Sage Journals: Discover world-class research

Abstract

Background

Image segmentation is an essential step in the analysis and subsequent characterisation of brain tumours through magnetic resonance imaging. In the literature, segmentation methods are empowered by open-access magnetic resonance imaging datasets, such as the brain tumour segmentation dataset. Moreover, with the increased use of artificial intelligence methods in medical imaging, access to larger data repositories has become vital in method development.

Purpose

To determine what automated brain tumour segmentation techniques can medical imaging specialists and clinicians use to identify tumour components, compared to manual segmentation.

Methods

We conducted a systematic review of 572 brain tumour segmentation studies during 2015–2020. We reviewed segmentation techniques using T1-weighted, T2-weighted, gadolinium-enhanced T1-weighted, fluid-attenuated inversion recovery, diffusion-weighted and perfusion-weighted magnetic resonance imaging sequences. Moreover, we assessed physics or mathematics-based methods, deep learning methods, and software-based or semi-automatic methods, as applied to magnetic resonance imaging techniques. Particularly, we synthesised each method as per the utilised magnetic resonance imaging sequences, study population, technical approach (such as deep learning) and performance score measures (such as Dice score).

Statistical tests

We compared median Dice score in segmenting the whole tumour, tumour core and enhanced tumour.

Results

We found that T1-weighted, gadolinium-enhanced T1-weighted, T2-weighted and fluid-attenuated inversion recovery magnetic resonance imaging are used the most in various segmentation algorithms. However, there is limited use of perfusion-weighted and diffusion-weighted magnetic resonance imaging. Moreover, we found that the U-Net deep learning technology is cited the most, and has high accuracy (Dice score 0.9) for magnetic resonance imaging-based brain tumour segmentation.

Conclusion

U-Net is a promising deep learning technology for magnetic resonance imaging-based brain tumour segmentation. The community should be encouraged to contribute open-access datasets so training, testing and validation of deep learning algorithms can be improved, particularly for diffusion- and perfusion-weighted magnetic resonance imaging, where there are limited datasets available.

Keywords

Brain tumour magnetic resonance imaging segmentation systematic review brain artificial intelligence

Introduction

Brain tumours are malignancies of brain tissues. Characterising such tissues and identifying related genes can help to estimate tumour spread, and further help to identify tumour grades for the treatment planning.^1,2 Such characterisation comprises of the different tumour components assessment. Components such as tumour core, boundary of tumour core, peritumoral oedema, cellular proliferation (an increase of the number of cells), cellular infiltration (migration of cells or excessive growth) and vascular proliferation (leaky blood vessels) are of great clinical interest.^1,3 Current practice involves various magnetic resonance imaging (MRI) approaches to visualise these tumour components.⁴ Particularly, the tumour core can be visible in T2-weighted MRI, and T1-weighted MRI, the enhanced boundary of tumour core can be seen in gadolinium-based T1-weighted MRI (T1-Gd), peritumoral oedema can be visible in fluid-attenuated inversion recovery (FLAIR) MRI, while cellular proliferation, cellular infiltration and vascular proliferation can be visualised in diffusion-weighted MRI and perfusion-weighted MRI.

In the literature, MRI-based image processing methods addressed the outlining of tumour components.⁵ These methods can distinguish between healthy and tumour tissues. Moreover, they can distinguish among different tumour components within the tumour. Such methods are commonly known as segmentation methods,⁵ which can be manual, semi-automatic or automatic. With the increase of segmentation methods, the medical image analysis community has reviewed them using publicly available benchmark datasets (e.g. the brain tumour segmentation – BraTS dataset) to assess their performance.^5,6 Additionally, individual reviews addressing only deep learning methods,⁷ only automated methods,⁸ classical reviews,⁹ and practical implications¹⁰ are reported. However, a systematic review to identify promising, and widely adapted brain tumor segmentation methods is not reported in the literature. Moreover, in the literature, individual methods, and their validation show technical advancements in most MR image-based approaches, but they are not fully explored in all MRI sequences in the imaging of brain tumours. For example, diffusion-weighted, and perfusion-weighted MRI are often overlooked or limited in brain tumor segmentation method development, and validation. In addition, a recent trend of deep learning methods has extensively contributed to the development of automatic segmentation methods, in order to avoid the subjective, and time-consuming nature of manual techniques. However, the performance of these deep learning methods relies on the data types used, single or multi-centre data collection, the number of available data samples for training, validation and testing of the approach. Therefore, it is crucial to explore the various studies in deep learning methods. Such a study can help understanding and justifying the need for further large open-access data repositories and alternative artificial intelligence (AI)-based techniques, such as transfer learning.

In this systematic review, we addressed the following PICO (P: population, I: intervention, C: comparison, O: outcome) format research question:

What automated brain tumour segmentation techniques can medical imaging specialists, and clinicians use to identify tumour components, compared to manual segmentation?

To answer the above question, we reviewed brain tumour segmentation methods which are based on physics or mathematics models, deep learning models, and software or semi-automatic methods. The methods, which use at least four types of MRI sequences (T1-weighted, T1-Gd, T2-weighted and FLAIR MRI), are included in the synthesis as they are common in clinical practice. Articles based on images from other MRI sequences are discussed, individually. The accuracy measures and study population of the various segmentation methods are also reviewed. Moreover, deep learning architectures are reviewed for their underlying network architecture (for example, U-Net, VGG etc.) with a list of articles that adapted such architectures in their studies. Technical specifications of such architectures are listed in the discussion.

Materials and methods

Study protocol

A protocol for this study is prepared internally but not registered elsewhere. However, the PROSPERO database is checked to ensure the originality of the study.

Article search

For source articles’ systematic identification, we have searched the following online databases: PubMed, Embase Ovid and Engineering Village. The different combinations of the keywords ‘glioma’, ‘medulloblastoma’, ‘brain tumour’ and ‘segmentation’ were used. After discussion among the authors, the combination, ‘glioma’, and ‘segmentation’, and ‘brain tumour’, was used for the article search. The search duration was defined to include articles published in the period from 2015 until 2020. In PubMed, the specific search filters were full-text, humans, English and segmentation keyword must be in the article title or abstract. In Engineering Village, the specific search filters were that the segmentation keyword must be in the abstract, and glioma, and brain tumour keywords must be in the subject or title or abstract. All the articles were stored using Zotero¹¹ software. After removing the duplicates, study inclusion/exclusion criteria were applied.

Study inclusion/exclusion criteria

Articles are screened at two levels for inclusion/exclusion. First, the articles are screened at the abstract, and title level. At this level, the studies, which are not segmentation studies, are excluded. The excluded studies are grouped into the following categories: clinical analysis studies, case studies, image pre-processing studies, general overall surveys, tumour classification studies, tumour detection and identification studies, information learning for model training studies, surgical planning studies, datasets and not accessible studies.

Second, the segmentation studies are screened by reading the full text. At this level, the articles, which are eligible for synthesis, are identified by reviewing the imaging modalities, and associated data types used in their segmentation method. The studies, which are not MRI-based, are excluded. As a result, among the MRI-based studies, articles that utilised T1-weighted, T2-weighted, T1-Gd and FLAIR MRI are included for the final synthesis. All the inclusion/exclusion criteria are mentioned in the PRISMA¹² diagram of Figure 1.

Figure 1.

PRISMA diagram. PRISMA diagram of the systematic review of brain tumour segmentation methods.

Data extraction

After removing duplicates, all the articles are organised, as a table, in Microsoft Excel (Please refer to S1 Appendix). For each article, data is extracted according to the following parameters: publication id, publication type, publication year, author, article title, publisher, DOI, issue number, volume number, type of study, type of technical methodology, type of imaging modality, type of MRI, type of tumour, single or multi-centre data collection, the number of data samples and performance score.

Among the aforementioned parameters, the ‘data samples’ parameter is sub-categorised into study population, training data, validation data and test data. The types of technical methods in image-based tumour segmentation are organised into three categories: physics or mathematics-based methods, deep learning-based methods and software or semi-automatic methods. The performance score of each segmentation method is retrieved from the respective article and included in our table. In articles, authors have used various performance score measures, which include match, accuracy, Jaccard similarity coefficient, Tanimoto similarity, Hausdorff distance, dice score, positive predictive value, specificity, sensitivity, negative predictive value, precision, recall, misclassification error, intersection over union, Lin's concordance correlation coefficient, quality measure, balanced error rate, kappa, correlation, mean square error, false-positive rate per patient, the extent of resection, mutual information, residual tumour volume, root mean square error, the ratio of overlap, coefficient of variation, agreement index, interoperator variance, F1 score, volume difference, peak signal-to-noise ratio, qualitative evaluation, difference ratio of pixels, similarity, overlap index, absolute error, percent error, the difference with the gold standard, paired t-test mean difference, linear regression and area under the curve. The details of each performance score matrix can be found in their respective articles as mentioned in the data extraction matrix (S1 Appendix). We synthesised the performance of each method in segmenting whole tumour (WT), non-enhancing tumour core (TC) and enhanced tumour (ET). The values are stored as a tuple of ‘whole tumour, tumour core, enhanced tumour’. Each missing value in a performance score tuple is recorded as null. As shown in the PRISMA diagram of Figure 1, the imaging modalities, and MRI sequences are identified at the eligibility level. Broadly, four imaging modalities have been identified: magnetic resonance imaging (MRI), computed tomography, positron emission tomography, and ultrasound. Specifically in MRI, the identified imaging sequences include T1-weighted MRI, gadolinium-enhanced T1-weighted MRI (T1-Gd), T2-weighted MRI, FLAIR MRI, diffusion-weighted MRI, fMRI, perfusion-weighted MRI, magnetic resonance spectroscopy, apparent diffusion constant, fractional anisotropy, diffusion tensor, dynamic susceptibility contrast, dynamic contrast enhancement, diffusion kurtosis, magnetisation prepared rapid gradient echo, T1-MPRAGE and T1-weighted with turbo field echo.

Risk of bias assessment

All reviewers assessed studies or validated data extraction matrix independently. JMB assessed each study and populated data extraction matrix records. SLCK validated the data extraction matrix. TNA verified, and confirmed the data extraction matrix. The whole process was performed manually, and without using any automation tools.

Synthesis methods

Studies are synthesised according to the publication year, technical methods used in method development (type of study), MRI sequences used in the segmentation method development, deep learning methods (technical architectures used in deep learning studies), performance score to evaluate accuracy in segmenting three tumour components – whole tumour, and tumour core, and enhanced tumour, study population, and specific studies with additional MRI sequences. The studies having at least T1-weighted, T2-weighted, T1-Gd and FLAIR MRI sequences used are included for the synthesis. The synthesis of a number of articles is visualised as a bar plot, and pie charts showing the total number of articles in each year, and category. The synthesis of the study population in terms of data samples used, and performance score measure in terms of median Dice score are visualised as results. The outcome is depicted as Matlab box plots as commonly found in the synthesised articles. Deep learning studies are tabulated to identify the most widely adapted deep learning technology.

Benchmark review articles and particular imaging sequences studies are mentioned in the specific imaging studies’ section. Studies of segmentation methods using diffusion-weighted, and perfusion-weighted MRI, are reported individually.

During the synthesis, the unavailable values of performance score measures and data samples are considered as null values.

Results

Article identification

In this systematic review, 572 articles are identified from online publication repositories. Out of these, 14 duplicates are removed and the remaining 558 articles are screened for the eligibility criteria. After screening titles and abstracts, 246 articles are excluded, and the remaining 312 articles are screened by reading the full text. After the full-text screening, 89 articles are excluded with reasons, and 223 articles are included for synthesis. The summary of exclusion reasons and the number of excluded articles are shown in the PRISMA diagram (Figure 1).

Publications over the time

As shown in the PRISMA diagram, after screening at the abstract and title level, we selected 312 articles. Following that we applied the eligibility criteria, which provided us with 223 articles for the synthesis. These articles are categorised over the publication year, as shown in Figure 2 bar plot. Further, the articles are categorised as per the technical method category, as shown in Figure 2 pie charts.

Figure 2.

A number of articles (2015–2020). The bar plot represents the number of articles published over the review period (2015–2020), and pie charts depict published articles in each category of technical method in each corresponding year. Total articles = 223 refers to the articles included for the synthesis.

Type of study

After the screening at the title and abstract level, 312 studies are included. Among them, 102 are physics or mathematics-based, 189 are deep learning-based and 21 are software-based or semi-automatic methods articles. After the full-text screening, 223 studies are included for synthesis. Among them, 61 are physics or mathematics-based,^13–74 156 are deep learning-based and six are software-based or semi-automatic^75–80 methods articles.

Imaging sequences in synthesised studies

A total of 223 studies have used at least four MRI sequences (T1-weighted, T2-weighted, T1-Gd and FLAIR MRI) which are included in the final synthesis.

Technical architectures in studies

We synthesised 156 deep learning articles to identify the commonly reported deep learning architecture in automatic brain tumour segmentation. We reported deep learning architectures such as convolutional neural network (CNN),^81–83 visual geometry group (VGG) network,^84–86 DeepMedic,^84,87–92 U-Net,^{84,91,93–114} autoencoder,^115–117 generative adversarial network (GAN),^118–120 W-net,¹¹³ a cascade of W-Net E-Net, and T-Net,¹¹³ squeeze, and excitation network (SENet),¹²¹ multiresolution neural network,¹²² holistically-nested edge detection (HED) network,¹²³ multi-level upsampling network,¹²⁴ V-net,¹²⁵ residual network (ResNet),^125,126 hourglass network,¹²⁷ multi-view network (MvNet),¹²⁸ DeepSCAN,^129,130 densely connected,⁸⁵ inception,^85,99 ensemble net,¹³¹ PixelNet,^132,133 ContextNet,¹³⁴ dense neural network,¹³⁵ MC-Net,¹³⁶ OM-Net,¹³⁶ ConvNet,¹³⁷ wide residual network, and pyramid pool network (WRN-PPNet),¹³⁸ deep convolutional network,^139,140 neuromorphic neural network,¹⁴¹ DeepLabv3 +,¹⁴² recurrent neural network¹⁴³ and German cancer research centre (DFKZ) network.¹⁰⁷ Moreover, we mentioned the extensions of these architectures in the discussion section. Deep architectures, and their extensions, with associated publications, are summarised in Table 1.

Table 1.

Deep architectures, and their extensions used in tumour segmentation.

Deep architecture	Associated publications
CNN	^{81–83,87,107,144–165}
VGG	^{84–86,166,167}
DeepMedic	^{84,87–92,168,169}
U-Net	^{84,91,93,94,95,96,97,98–100,101–110,111–114,155,157,162,163,170,171,172–181,182–191,192–201,202–210}
Autoencoder	^{115–117,211}
GAN	^{118–120,212–214}
W-net and cascade of W-net E-net and T-net	^{113,215–218}
SENet	¹²¹
Multiresolution neural network	¹²²
HED	¹²³
Multi-level upsampling network	¹²⁴
V-net	^125,219
ResNet	^{125,126,197,220,221}
Hourglass network	¹²⁷
MvNet	¹²⁸
DeepSCAN	^129,130
Densely connected	⁸⁵
Inception	^85,99
Ensemble net	¹³¹
PixelNet	^132,133
ContextNet	¹³⁴
Dense neural network	^135,222
MC-Net	¹³⁶
OM-Net	¹³⁶
ConvNet	^137,223
WRN-PPNet	^138,224
Deep convolutional network	^{139,140,159,225}
Neuromorphic neural network	¹⁴¹
DeepLabv3 +	¹⁴²
Recurrent neural network	¹⁴³
DFKZ	¹⁰⁷
EMMA	⁹¹
SegNet	²²⁶
ResNeXt	²²⁷
DenseAFPNet	²²⁸
DMFNet	^229–231
P-Net	¹⁹⁷
MFNet	²³²
HNF-Net	²³³
Deep neural network	^234,235
D2C2N	²³⁶
DeepSeg	²³⁷

CNN: convolutional neural network; VGG: visual geometry group; GAN: generative adversarial network; SENet: squeeze, and excitation network; HED: holistically-nested edge detection; ResNet: residual network; MvNet: multi-view network, WRN-PPNet: wide residual network, and pyramid pool network, DFKZ: German cancer research centre; EMMA: ensembles of multiple models, and architecture; DenseAFPNet: Dense atrous feature pyramid network; DMFNet: dilated multi-fibre network; MFNet: multi-direction fusion network; HNF-Net: high-resolution, and non-local feature network; D2C2N: dilated densely connected convolutional network.

Performance score evaluation matrices of studies

From our data extraction matrix, it was apparent that the most common accuracy measure used is the Dice score. Therefore, in Figure 3, we have shown the median Dice score values in segmenting the WT, TC and ET areas of brain tumours, considering all the 223 articles. We have compared Dice score among physics or mathematics-based methods, deep learning methods and software-based or semi-automatic methods.

Figure 3.

Comparison of segmentation results. Performance score evaluation, in segmenting WT, TC and ET, by considering all 223 articles.

Study population

The study population is summarised as the total number of data samples used in the study. The median data sample used in segmentation studies is 351 (median ± stdev = 351 ± 232.67). Moreover, deep learning segmentation methods are noted to have data samples divided into three categories: training data, validation data and test data. As shown in Figure 4, the median ± stdev of each training, validation and test data sample in deep learning methods are also reported, which is 285 ± 154.41, 54 ± 41.60 and 110 ± 85.29, respectively.

Figure 4.

Data samples in deep learning studies. Summary of training, validation and test data samples reported in deep learning methods. Median of training, validation and test data samples are 285, 54 and 110, respectively.

Specific imaging studies

A rigorous review, using an open-access dataset, is published as the gliomas segmentation benchmark article,⁵ which encouraged a tumour segmentation method development using T1-weighted, T2-weighted, T1-Gd and FLAIR MRI sequences. Among the 223 synthesised articles, five studies have used additional imaging sequences. Three studies^71,117,236 have used diffusion imaging sequences, and one study²¹² has used perfusion imaging and multiple CT imaging sequences. These three studies are noted to have deep learning, or physics or mathematics-based methods. Moreover, we noted that a few studies have used perfusion imaging sequences, and diffusion imaging sequences but did not use the aforementioned all four MRI sequences. These studies are perfusion imaging studies^{212,238–241} and diffusion imaging studies.^{117,169,236,239–247} Also, two studies are noted to have used ultrasound imaging.^248,249

Discussion

In this systematic review, we addressed a PICO style question to put the brain tumour segmentation methods in the context of clinical utility. Our population (P) is brain tumour patients, intervention (I) is segmentation methods, comparison (C) with manual segmentation evaluated as Dice score and outcome (O) is automated segmentation methods. Accordingly, we systematically found a promising technology, U-Net, which performs automated brain tumour segmentation using multiple MRI sequences. Its validation with manually segmented ground truth has achieved a Dice score of 0.9. Moreover, U-Net is majorly cited compared to other automated methods. Such merits show that U-Net has the potential to be considered for the clinical efficacy studies of automated brain tumour segmentation.

Research in context
Evidence before this study: In the literature, individual studies of brain tumour segmentation methods, a benchmark framework (BraTS) to assess these methods using an open-access dataset of T1-weighted, T2-weighted, T1-Gd and FLAIR MRI are reported. A few studies used diffusion-weighted and perfusion-weighted MRI to validate their methods.
Added value of this study: U-Net is systematically identified as a promising technology (Dice score 0.9 and cited the most) for brain tumour segmentation. Moreover, it is noted that the contribution of open-access datasets, including diffusion-weighted, and perfusion-weighted MRI, should be encouraged for training, validation and testing of brain tumour segmentation algorithms.
Implications of all the available evidence: The identified technologies can help medical imaging specialists, and clinicians to semi-automatically or automatically segment brain tumours, compared to manually.

As shown in Figure 2 orange sections, deep learning articles in segmenting brain tumours are increasingly dominating the literature compared to software-based or semi-automatic methods, and considerably increased compared to physics or mathematics-based methods during the 2015–2020 period. The potential reasons could be the subjective nature of a software-based approach, limiting its efficiency, and limited transfer learning in physics-based models. Deep learning methods are attractive for efficient automatic segmentation, and the learned weights using one dataset can be further utilised with another dataset.

From a clinical point of view, the articles, which have reported segmentation of all three regions (whole tumour-WT, non-enhancing tumour core-TC, and enhanced tumour-ET), are synthesised. The whole tumour, including peritumoral oedema, is visible in FLAIR MRI, non-enhancing solid tumour core is visible in T2-weighted MRI, and necrotic/cystic core and enhanced core are visible in T1-weighted, and T1-Gd MRI.^5,250 Specific details, such as cellular proliferation and cellular infiltration, can be visualised with diffusion-weighted imaging.^251,252 Moreover, vascular proliferation can be visualised with perfusion-weighted imaging.²⁵¹ The inclusion of tumour components, such as cellular proliferation, infiltration and vascular proliferation, are addressed in very few articles. The potential reason could be the limited availability of open-access diffusion and perfusion-weighted imaging datasets.

Segmentation methods are commonly validated in terms of Dice scores ranging from 0 to 1.0. 1.0 shows the best matching between segmentation ground truth and method output. As shown in Figure 3, physics or mathematics-based methods, deep learning methods and software-based or semi-automatic methods have median Dice score, as (WT, TC, ET) tuple, (0.84, 0.76, 0.7), (0.87, 0.78, 0.73) and (0.88, 0.69, 0.71), respectively. Particularly, physics or mathematics-based methods, and deep learning methods studies are enabled to explore accuracy in segmenting WT, TC and ET due to the availability of ground truth in datasets such as BraTS.⁵ As the whole tumour covers a larger area, the value of WT is higher compared to TC or ET in each category.

As shown in the PRISMA diagram Figure 1, we have synthesised articles with physics or mathematics-based methods, deep learning-based methods and software or semi-automatic methods. The technical architectures are specifically reviewed for deep learning-based methods, as they have the potential to transfer learning from one dataset to another dataset.¹¹⁷ Modifications of CNN are fully CNN,^81–83 hyper-dense CNN,¹⁶¹ cascaded CNN,^162,163 cascaded fully CNN,¹⁶⁴ multi-path CNN,^87,165 ensembled CNN,¹⁴⁴ holistic CNN,¹⁴⁵ full-resolution residual CNN,¹⁴⁶ two-phase patch-based CNN¹⁴⁷ and cascaded anisotropic CNN.¹⁰⁷ U-Net is modified as cascaded U-Net,^162,163 residual U-Net,^170,171 domain adapted U-Net¹⁷² and efficient spatial pyramid (ESP) network.¹⁷³ A modification of W-net is reported as the cascade of W-net, E-net and T-net,¹¹³ and a modification of V-net is reported as the cascaded V-net.²¹⁹ Modifications of ResNet are dilated ResNet²²⁰ and fully convolutional residual neural network.²²¹ An ensemble of multiple deep architectures is also reported as ensembles of multiple models, and architecture (EMMA).⁹¹ A modification of GAN is reported as conditional GAN.²¹² Modifications of ConvNet are reported as classification ConvNet²²³ and detection ConvNet.²²³

As summarised in Table 1, among all these deep architectures, U-Net based architectures are reported the most among the U-Net based articles, the maximum accuracy of segmenting a tuple of (WT, TC, ET) is reported as (0.92, 0.95, 0.94) in terms of Dice score. Several other architectures such as VGG, DeepMedic, autoencoder and GAN are also frequently reported, and extended. A summary of the technical specifications of U-Net, VGG, DeepMedic, autoencoder and GAN-based architectures is reported in Table 2. In physics or mathematics-based studies, and software-based or semi-automatic studies, the maximum reported Dice score in segmenting a tuple of (WT, TC, ET) is (0.97, 0.86, 0.95) and (0.88, 0.69, 0.71), respectively.

Table 2.

Articles of widely used deep architectures and their technical details.

Deep architecture	Article	Technical details
U-Net	⁸⁴	3D U-Net, which synthesises information at each scale by combining local and contextual information
	¹⁰³	Modified 3D U-Net with better gradient flow
	¹⁰⁸	Modified U-Net with an upsampled component, which is based on the nearest neighbour algorithm and elastic transformation
	¹⁰⁹	2D U-Net network using a biophysics-based domain adaptation method with a generative adversarial model, which synthesises known ground truth data
	¹¹⁰	Modified U-Net with up skip connection, inception module and efficient cascade training
	¹¹¹	3D U-Net with DenseNet, which was pre-trained on ImageNet
	¹¹²	U-Net network
	¹¹³	3D U-Net with test time augmentation
	¹¹⁴	U-Net with dice loss function to tackle class imbalance problem, and extensive data augmentation to prevent over-fitting
	⁹³	Ensemble of 3D U-Nets with different hyperparameters
	⁹⁴	U-Net training, which is followed by Bit-plane method output
	⁹⁵	U-Net with double convolutional layers, inception modules and dense modules
	⁹⁶	Modified U-Net, which is addressing class imbalance problem, with weighted cross-entropy and generalised dice loss function
	⁹⁷	Deep learning radiomics algorithm model with 3D patch-based U-Net
	⁹⁸	U-net with encoder adaption block and densely connected fusion blocks in the decoder
	⁹¹	An ensemble of two 3D U-Nets in which skip connections are used as a summation of signals in the up-sampling part of one network, and the other network uses concatenated skip connections and stridden convolutions
	⁹⁹	Inception modules with U-Net
	¹⁰⁰	Multi-scale images as input to the 3D U-Net and including 3D atrous spatial pyramid pooling layer to boost the performance of the network
	¹⁰¹	U-Net training improvement using large patch sizes, region-based training, additional data and a combination of loss functions
	¹⁰²	U-Net with separable 3D convolution by dividing each 3D convolution block into three parallel branches
	¹⁰⁴	Two 3D U-Nets in which the first detect the tumour, and the second one segments multiple regions of the tumour
	¹⁰⁵	U-Net
	¹⁰⁶	The tree structure of 3D U-Nets such that the first node of the tree predicts oedema, and then feed the output to the subsequent nodes to detect tumorous subregions of oedema
	¹⁰⁷	U-Net in an ensemble of networks
VGG	⁸⁴	3D fully connected network which is based on VGG with skip connections that combine coarse high scale information with fine low scale information
	⁸⁵	3D convolutions, except max pool layers, VGG-based, an ensemble of multiple architectures
	⁸⁶	CNN, which is based on VGG-16 and initially trained on ImageNet weights, and then fine-tuned with MICCAI data, relies on a pseudo-3D method which enables 3D segmentation from 2D colour-like images, and ultimately gives faster segmentation
DeepMedic	⁸⁴	Two path network based on DeepMedic network which allows gathering low, and high resolution features together
	⁸⁷	Multi-path CNN, which is inspired by DeepMedic which includes large, and small patches
	⁸⁸	DeepMedic network
	⁸⁹	A computer-aided diagnosis that combined DeepMedic, and radiomics features such as first-order features, shape features and texture features
	⁹⁰	DeepMedic with additional residual connections
	⁹¹	An ensemble of two DeepMedic architectures
	⁹²	A DeepMedic-based network is followed by a fully connected network to remove false positives
Autoencoder	¹¹⁵	Encoder, and decoder based 3D architecture that includes a variational auto-encoder branch to reconstruct the input image, which could be used as a regulariser for the shared decoder
	¹¹⁶	Stacked denoising auto-encoder
	¹¹⁷	Stacked denoising auto-encoder
GAN	¹¹⁸	Discriminator and generator based conditional generative adversarial network
	¹¹⁹	Adversarial network, discriminator is trained along with a generator to produce synthetic results, synthetic labels and ground truth are discriminated by discriminator, discriminator output is fed back to generator for improved segmentation accuracy
	¹²⁰	Generative adversarial network with a coarse-to-fine generator to generate generic augmented data

VGG: visual geometry group; GAN: generative adversarial network; CNN: convolutional neural network; MICCAI: Medical Image Computing, and Computer-Assisted Interventions.

In this study, we have synthesised the study population and performance measures from the articles. Among the 223 synthesised articles, 217 studies have used multi-centre datasets, five studies have used single-centre datasets, and one study has used only a synthetic dataset. Note that, data variety could be limiting in the generalisation of Dice score performance comparison reported in our study. A potential solution is to develop an open-access data repository and review the studies with the same data samples. Moreover, some of the studies have missing values, which could limit our synthesis results. A benchmark framework to report the evaluation measures could be useful to mitigate such limitations in the future. In this study, we identified brain tumour segmentation techniques and synthesised results as found from their respective articles, which could be a limitation. In order to apply these methods at clinics, a separate efficacy study should be performed by clinical staff members utilising data at respective clinics.

Conclusion

In conclusion, we systematically addressed a review question that can help medical imaging specialists, and clinicians to identify automatic brain tumour segmentation techniques, compared to manual segmentation. Our specific inclusion criteria emphasised having multiple MRI sequences in the method development. We noted that four MR-based sequences, i.e. T1-weighted, T2-weighted, T1-Gd and FLAIR MRI, are used the most. Diffusion weighted, and perfusion-weighted MRIs are rarely used. Among the segmentation methods, deep learning methods have contributed the most compared to other methods during the 2015–2020 period. Within the deep learning methods, U-Net-based methods are adapted the most and have an accuracy of approximately 0.9 Dice score in segmenting a brain tumour. We also noticed that the benchmark BraTS dataset does not have perfusion-weighted and diffusion-weighted MRI data, motivating the development of an open-access data repository with such MRI sequences.

In the future, a novel dataset can be developed with additional imaging data such as diffusion-weighted, and perfusion-weighted MRI, similar to the frequently reported open-access dataset^5,6 which contains the data samples of T1-weighted, T2-weighted, T1-Gd and FLAIR MRI of gliomas patients. Creating such an open-access dataset can help to include cellular proliferation, infiltration and vascular proliferation in brain tumour segmentation techniques (cellular proliferation: increase of the number of cells; cellular infiltration: migration of cells or excessive growth; vascular proliferation: leaky blood vessels). Moreover, medulloblastoma cases are rare (European annual rate: 6.8/million, age: 0–14 years, duration: 2000–2007).^253,254 Therefore, a comprehensive adult dataset may be useful to address childhood tumours with transfer learning methods.

Supplemental Material

sj-xlsx-1-dhj-10.1177_20552076221074122 - Supplemental material for Magnetic resonance image-based brain tumour segmentation methods: A systematic review

Supplemental material, sj-xlsx-1-dhj-10.1177_20552076221074122 for Magnetic resonance image-based brain tumour segmentation methods: A systematic review by Jayendra M Bhalodiya, Sarah N Lim Choi Keung and Theodoros N Arvanitis in Digital Health

Footnotes

Acknowledgements

The authors thank HDR UK for receiving funding for this work. HDR UK is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation and Wellcome Trust Funder: HDR UK – Medical Research Council – Award Reference HDR-3001.

Conflict of interest

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Professor Theodoros N. Arvanitis (TNA) is joint Editor-in-Chief of Digital Health.

Contributorship

All the authors reviewed this manuscript. JMB prepared data extraction matrix, analyse data, wrote, and edited manuscript. SLCK validated the data extraction matrix and reviewed the manuscript. TNA verified the data extraction matrix, reviewed the manuscript, and acquired funding for the project. All authors have full access to all the data in this study.

Data availability

The data extraction matrix of this study is available in the Supplementary Information files.

Ethical approval

Not applicable. No human subjects involvement.

Guarantor

Professor Theodoros N. Arvanitis (TNA) is the guarantor for this study.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Health Data Research UK (grant number HDR-3001).

ORCID iDs

Jayendra M Bhalodiya

Theodoros N Arvanitis

Supplemental material

Supplemental material for this article is available online.

References

Lam

WWM

Poon

Metreweli

. Diffusion MR imaging in glioma: does it have any role in the pre-operation determination of grading of glioma? Clin Radiol 2002; 57: 219–225.

Louis

Perry

Wesseling

, et al. The 2021 WHO classification of tumors of the central nervous system: a summary. Neuro Oncol 2021; 23: 1231–1251.

Price

. The role of advanced MR imaging in understanding brain tumour pathology. Br J Neurosurg 2007; 21: 562–575.

Nilsson

Englund

Szczepankiewicz

, et al. Imaging brain tumour microstructure. Neuroimage 2018; 182: 232–250.

Menze

Jakab

Bauer

, et al. The multimodal brain tumor image segmentation benchmark (BRATS). IEEE Trans Med Imaging 2015; 34: 1993–2024.

Bakas

Akbari

Sotiras

, et al. Advancing the cancer genome atlas glioma MRI collections with expert segmentation labels and radiomic features. Sci Data 2017; 4: 170117.

Işin

Direkoǧlu

Şah

. Review of MRI-based brain tumor image segmentation using deep learning methods. Procedia Comput Sci 2016; 102: 317–324.

Angulakshmi

Lakshmi Priya

. Automated brain tumour segmentation techniques—a review. Int J Imaging Syst Technol 2017; 27: 66–77.

Wadhwa

Bhardwaj

Singh Verma

. A review on brain tumor segmentation of MRI images. Magn Reson Imaging 2019; 61: 247–259.

10.

Abd-Ellah

Awad

Khalaf

AAM

, et al. A review on brain tumor diagnosis from MRI images: practical implications, key achievements, and lessons learned. Magn Reson Imaging 2019; 61: 300–318.

11.

AGPL. Zotero, https://www.zotero.org/.

12.

Page

McKenzie

Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Br Med J 2021; 372. Epub ahead of print 2021. DOI:

10.1136/bmj.n71

13.

Njeh

Sallemi

Ben

, et al. 3D Multimodal MRI brain glioma tumor and edema segmentation: a graph cut distribution matching approach. Comput Med Imaging Graph 2015; 40: 108–119.

14.

Menze

Van Leemput

Lashkari

, et al. A generative probabilistic model and discriminative extensions for brain lesion segmentation—with application to tumor and stroke. IEEE Trans Med Imaging 2016; 35: 933–946.

15.

Osman

AFI

. Automated brain tumor segmentation on magnetic resonance images and patient’s overall survival prediction using support vector machines. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2018; 10670 LNCS: 435–449.

16.

Chen

Nguyen

Chui

, et al. Automated brain tumor segmentation using kernel dictionary learning and superpixel-level features. 2016 IEEE Int Conf Syst Man, Cybern SMC 2016 – Conf Proc 2017: 2547–2552.

17.

Juan-Albarracín

Fuster-Garcia

Manjón J

, et al. Automated glioblastoma segmentation based on a multiparametric structured unsupervised classification. PLoS One 2015; 10: 1–20.

18.

Zhao

Yang

Lin

, et al. Automated glioma detection and segmentation using graphical models. PLoS One 2018; 13: 1–22.

19.

Kapás

Lefkovits

Iclănzan

, et al. Automatic brain tumor segmentation in multispectral MRI volumes using a random forest approach. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2018; 10749 LNCS: 137–149.

20.

Latif

Butt

Khan

, et al. Automatic multimodal brain image classification using MLP and 3D glioma tumor reconstruction. In: 2017 9th IEEE-GCC Conference and Exhibition (GCCCE). IEEE, 2017, pp. 1–9.

21.

Chen

. Automatic multimodal brain-tumor segmentation. In: Proceedings - 5th International Conference on Instrumentation and Measurement, Computer, Communication, and Control, IMCCC 2015. IEEE, 2016, pp. 939–942.

22.

Kadkhodaei

Samavi

Karimi

, et al. Automatic segmentation of multimodal brain tumor images based on classification of super-voxels. Proc Annu Int Conf IEEE Eng Med Biol Soc EMBS 2016; 2016–Octob: 5945–5948.

23.

Lee

Kim

Chen

, et al. Brain tumor image segmentation using kernel dictionary learning. Proc Annu Int Conf IEEE Eng Med Biol Soc EMBS 2015; 2015–Novem: 658–661.

24.

Ellwaa

Hussein

AlNaggar

, et al. Brain tumor segmentation using random forest trained on iteratively selected patients. In: Crimi, et al. A (ed) BrainLes. Springer, Cham: LNCS, 2016, pp.129–137.

25.

Zhan

Shen

Hong

, et al. A glioma segmentation method using cotraining and superpixel-based spatial and clinical constraints. IEEE Access 2018; 6: 57113–57122.

26.

Bakas

Reyes

Jakab

, et al. Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the BRATS challenge. arXiv, http://arxiv.org/abs/1811.02629 (2018).

27.

Lefkovits

Sziĺagyi

Lefkovits

. Brain tumor segmentation and survival prediction using a cascade of random forests. Cham: Springer International Publishing, Epub ahead of print 2019, DOI: 10.1007/978-3-030-11726-9.

28.

Kao

P-Y

Ngo

Zhang

, et al. Brain tumor segmentation and tractographic feature extraction from structural MR images for overall survival prediction. In: Crimi, et al. A (ed) BrainLes 2019, LNCS, Springer, Cham: Springer International Publishing, pp.128–141.

29.

Baid

Talbar

. Brain tumor segmentation based on non-negative matrix factorization and fuzzy clustering. BIOIMAGING 2017 – 4th Int Conf Bioimaging, Proceedings; Part 10th Int Jt Conf Biomed Eng Syst Technol BIOSTEC 2017 2017; 2017-Janua: 134–139.

30.

Rios Piedra

Ellingson

Taira

, et al. Brain tumor segmentation by variability characterization of tumor boundaries. In: A. Crimi, et al. (Eds.): BrainLes 2016, LNCS, pp. 206–216.

31.

Jia

Qin

. Brain tumor segmentation from multimodal magnetic resonance images via sparse representation. Artif Intell Med 2016; 73: 1–13.

32.

Anwar

Yousaf

Majid

. Brain tumor segmentation on multimodal MRI scans using EMAP algorithm. Proc Annu Int Conf IEEE Eng Med Biol Soc EMBS 2018; 2018–July: 550–553.

33.

Agn

Puonti

Rosenschöld

PMA

, et al. Brain tumor segmentation using a generative model with an RBM prior on tumor shape. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2016; 9556: 168–180.

34.

Lefkovits

Szilágyi

. Brain tumor segmentation with optimized random forest. In: A. Crimi, et al. (Ed.): BrainLes 2016, LNCS, pp. 88–99.

35.

Pinto

Pereira

Correia

, et al. Brain tumour segmentation based on extremely randomized forest with high-level features. Proc Annu Int Conf IEEE Eng Med Biol Soc EMBS 2015; 2015–Novem: 3037–3040.

36.

Kaur

Saini

Gupta

. A joint intensity and edge magnitude-based multilevel thresholding algorithm for the automatic segmentation of pathological MR brain images. Neural Comput Appl 2018; 30: 1317–1340.

37.

Serrano-Rubio

Everson

. Brain tumour segmentation method based on supervoxels and sparse dictionaries. In: A. Crimi, et al . (Eds.): BrainLes 2019, LNCS. Springer, Cham: Springer International Publishing, pp. 210–221.

38.

Luo

Wang

. Concatenated and connected random forests with multiscale patch driven active contour model for automated brain tumor segmentation of MR images. IEEE Trans Med Imaging 2018; 37: 1943–1954.

39.

Meier

Knecht

Wiest

, et al. CRF-Based brain tumor segmentation: Alleviating the shrinking bias. In: A. Crimi, et al. (Ed.): BrainLes 2016, LNCS, pp. 100–107.

40.

Abbasi

Tajeripour

. Detection of brain tumor in 3D MRI images using local binary patterns and histogram orientation gradient. Neurocomputing 2017; 219: 526–535.

41.

Selvapandian

Manivannan

. Fusion based glioma brain tumor detection and segmentation using ANFIS classification. Comput Methods Programs Biomed 2018; 166: 33–38.

42.

Bakas

Zeng

Sotiras

, et al. GLISTRboost: combining multimodal MRI segmentation, registration, and biophysical tumor growth modeling with gradient boosting machines for glioma segmentation. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2016; 9556: 144–155.

43.

Pinto

Pereira

Rasteiro

, et al. Hierarchical brain tumour segmentation using extremely randomized trees. Pattern Recognit 2018; 82: 105–117.

44.

Pei

Reza

SMS

Iftekharuddin

. Improved brain tumor growth prediction and segmentation in longitudinal brain MRI. In: Proceedings – 2015 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2015. IEEE, 2015, pp. 421–424.

45.

Steed

Treiber

Patel

, et al. Iterative probabilistic voxel labeling: automated segmentation for analysis of the cancer imaging archive glioblastoma images. Am J Neuroradiol 2015; 36: 678–685.

46.

Le Folgoc

Nori A

Ancha

, et al. Lifted auto-context forests for brain tumour segmentation. In: A. Crimi, et al. (Eds.): BrainLes 2016, LNCS, pp. 171–183.

47.

Agn

Munck af Rosenschöld

Puonti

, et al. A modality-adaptive method for segmenting brain tumors and organs-at-risk in radiation therapy planning. Med Image Anal 2019; 54: 220–237.

48.

Szilágyi

Iclănzan

Kapás

, et al. Low and high grade glioma segmentation in multispectral brain MRI data. Acta Univ Sapientiae, Inform 2018; 10: 110–132.

49.

Bonte

Goethals

Van Holen

. Machine learning based brain tumour segmentation on limited data using local texture and abnormality. Comput Biol Med 2018; 98: 39–47.

50.

Ali

HAM

Ahmed

MAA

Hussein

. MRI brain tumour segmentation based on multimodal clustering and level-set method. 2018 Int Conf Comput Control Electr Electron Eng ICCCEEE 2018 2018; 1–5.

51.

Phophalia

Maji

. Multimodal brain tumor segmentation using ensemble of forest method. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2018; 10670 LNCS: 159–168.

52.

Mote

Baid

Talbar

. Non-negative matrix factorization and self-organizing map for brain tumor segmentation. Proc 2017 Int Conf Wirel Commun Signal Process Networking, WiSPNET 2017 2018; 2018-Janua: 1133–1137.

53.

Tustison

Shrinidhi

Wintermark

, et al. Optimal symmetric multimodal templates and concatenated random forests for supervised brain tumor segmentation (simplified) with ANTsR. Neuroinformatics 2015; 13: 209–225.

54.

Meier

Karamitsou

Habegger

, et al. Parameter learning for CRF-based tissue segmentation of brain tumors. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2016; 9556: 156–167.

55.

Szilágyi

Brady

Berényi

. Phase congruency map driven brain tumour segmentation. In: SPIE Medical Imaging 2015: Image Processing. 2015, p. 941330.

56.

Pinto

Pereira

Dinis

, et al. Random decision forests for automatic brain tumor segmentation on multi-modal MRI images. Proc – 2015 IEEE 4th Port Meet Bioeng ENBENG 2015. Epub ahead of print 2015. DOI: 10.1109/ENBENG.2015.7088842.

57.

Zeng

Bakas

Sotiras

, et al. Segmentation of gliomas in pre-operative and post-operative multimodal magnetic resonance imaging volumes based on a hybrid generative-discriminative framework. In: A. Crimi, et al. (Eds.): BrainLes 2016, LNCS, pp. 184–194.

58.

Alberts

Charpiat

Tarabalka

, et al. A nonparametric growth model for brain tumor segmentation in longitudinal MR sequences. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2016; 9556: 69–79.

59.

Seow

Win

Wong

JHD

, et al. Segmentation of solid subregion of high grade gliomas in MRI images based on active contour model (ACM). J Phys Conf Ser 2016: 1–5. Epub ahead of print 2016; 694. DOI: 10.1088/1742-6596/694/1/012043

60.

Shanker

Singh

Bhattacharya

. Segmentation of tumor and edema based on K-mean clustering and hierarchical centroid shape descriptor. Proc - 2017 IEEE Int Conf Bioinforma Biomed BIBM 2017 2017; 1105–1109.

61.

Zhao

Meng

Wei

, et al. Supervised brain tumor segmentation based on gradient and context-sensitive features. Front Neurosci 2019; 13: 1–11.

62.

Bharath

Colleman

Sima

, et al. Tumor segmentation from multimodal mri using random forest with superpixel and tensor based feature extraction. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2018; 10670 LNCS: 463–473.

63.

Shanker

Bhattacharya

. Brain tumor segmentation of normal and lesion tissues using hybrid clustering and hierarchical centroid shape descriptor. Comput Methods Biomech Biomed Eng Imaging Vis 2019; 7: 676–689.

64.

Barbosa

Moreira

Ribeiro

, et al. Automatic classification and segmentation of low-grade gliomas in magnetic resonance imaging. In: A. M. Madureira, et al. (Eds.): SoCPaR 2018, AISC. 2020, pp. 43–50.

65.

Sharif

Amin

Raza

, et al. Brain tumor detection based on extreme learning. Neural Comput Appl 2020; 32: 15975–15987.

66.

Gyorfi

Csaholczi

Fulop

, et al. Brain tumor segmentation from multi-spectral magnetic resonance image data using an ensemble learning approach. IEEE Trans Syst Man, Cybern Syst 2020–Octob; 2020: 1699–1704.

67.

Csaholczi

Iclănzan

Kovács

, et al. Brain tumor segmentation from multi-spectral MR image data using random forest classifier. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2020; 12532 LNCS: 174–184.

68.

Pei

Bakas

Vossough

, et al. Longitudinal brain tumor segmentation prediction in MRI using feature and label fusion. Biomed Signal Process Control 2020; 55: 101648.

69.

Kaur

Saini

Gupta

. A novel fully automatic multilevel thresholding technique based on optimized intuitionistic fuzzy sets and tsallis entropy for MR brain tumor image segmentation. Australas Phys Eng Sci Med 2018; 41: 41–58.

70.

Mostafaie

Teimouri

Shahre Babak

, et al. Region of interest identification for brain tumors in magnetic resonance images. 2020 28th Iran Conf Electr Eng ICEE 2020 2020; 1–5.

71.

Soltaninejad

Yang

Lambrou

, et al. Supervised learning based multimodal MRI brain tumour segmentation using texture features from supervoxels. Comput Methods Programs Biomed 2018; 157: 69–84.

72.

Cordier

Delingette

Ayache

. A patch-based approach for the segmentation of pathologies: application to glioma labelling. IEEE Trans Med Imaging 2016; 35: 1066–1076.

73.

Bhattacharya

Sinha

. An improved approach of high graded glioma segmentation using sparse autoencoder and fuzzy c-means clustering from multi-modal MR images. In: SPIE Medical Imaging 2018: Biomedical Applications in Molecular, Structural, and Functional Imaging, 11-13 Feb. 2018. 2018, p. 38.

74.

Song

Chou

C-R

Chen

, et al. Anatomy-guided brain tumor segmentation and classification. In: A. Crimi, et al. (Eds.): BrainLes 2016, LNCS, pp. 162–170.

75.

Meier

Porz

Knecht

, et al. Automatic estimation of extent of resection and residual tumor volume of patients with glioblastoma. J Neurosurg 2017; 127: 798–806.

76.

Reyes AM

Buemi M

Aleman

, et al. Development of a graphic interface for the three-dimensional semiautomatic glioblastoma segmentation based on magnetic resonance images. In: Congreso Argentino de Ciencias de la Informatica y Desarrollos de Investigacion, CACIDI 2018. 2018. Epub ahead of print 2018. DOI: 10.1109/CACIDI.2018.8584357.

77.

Porz

Habegger

Meier

, et al. Fully automated enhanced tumor compartmentalization: man vs. machine reloaded. PLoS One 2016; 11: 1–16.

78.

Velazquez E

Meier

Dunn

, et al. Fully automatic GBM segmentation in the TCGA-GBM dataset: prognosis and correlation with VASARI features. Sci Rep 2015; 5: 1–10.

79.

Yushkevich

Pashchinskiy

Oguz

, et al. User-guided segmentation of multi-modality medical imaging datasets with ITK-SNAP. Neuroinformatics 2019; 17: 83–102.

80.

Ren

Sun

. Overall survival prediction using conventional MRI features. Springer, Cham: Springer International Publishing. 2020. Epub ahead of print 2020. DOI: 10.1007/978-3-030-46643-5_24.

81.

Zhao

Song

, et al. 3D Brain tumor segmentation through integrating multiple 2D FCNNs. In: BrainLes, LNCS, pp. 138–149.

82.

Jesson

Arbel

. Brain tumor segmentation using a 3D FCN with multi-scale loss. In: BrainLes, LNCS, pp. 138–149.

83.

Shaikh

Anand

Acharya

, et al. Brain tumor segmentation using dense fully convolutional neural network. In: BrainLes, LNCS, pp. 138–149.

84.

Casamitjana

Puch

Aduriz

, et al. 3D convolutional neural networks for brain tumor segmentation: a comparison of multi-resolution architectures. In: BrainLes. 2016, pp. 150–161.

85.

Albiol

. Extending 2D deep learning architectures to 3D image segmentation problems. In: BrainLes. Springer International Publishing, pp. 73–82.

86.

Puybareau

Tochon

Chazalon

, et al. Segmentation of gliomas and prediction of patient overall survival: a simple and fast procedure. In: BrainLes. Springer International Publishing, pp. 199–209.

87.

Sedlar

. Brain tumor segmentation using a multi-path. In: BrainLes. 2018, pp. 403–422.

88.

Perkuhn

Stavrinou

Thiele

, et al. Clinical evaluation of a multiparametric deep learning model for glioblastoma segmentation using heterogeneous magnetic resonance imaging data from clinical routine. Invest Radiol 2018; 53: 647–654.

89.

Chen

Liu

Peng

, et al. Computer-aided grading of gliomas combining automatic segmentation and radiomics. Int J Biomed Imaging 2018; 2018: 1–11.

90.

Kamnitsas

Ferrante

Parisot

, et al. Deepmedic for brain tumor segmentation. In: BrainLes. 2016, pp. 138–149.

91.

Kamnitsas

Bai

Ferrante

, et al. Ensembles of multiple models and architectures for robust brain tumour segmentation. In: BrainLes. 2018, pp. 450–462.

92.

Laukamp

Thiele

Shakirin

, et al. Fully automated detection and segmentation of meningiomas using deep learning on routine multiparametric MRI. Eur Radiol 2019; 29: 124–132.

93.

Feng

Tustison

Meyer

. Brain tumor segmentation using an ensemble of 3d u-nets and overall survival prediction using radiomic features. In: BrainLes. Springer International Publishing, pp. 279–288.

94.

Tuan

Bao

. Brain tumor segmentation using bit-plane and UNET. In: BrainLes. Springer International Publishing, pp. 466–475.

95.

Kim

. Brain tumor segmentation using deep fully convolutional neural networks. BrainLes 2018; 10670: 344–357.

96.

Kermi

Mahmoudi

Khadir

. Deep convolutional neural networks using U-net for automatic brain tumor segmentation in multimodal MRI volumes. In: BrainLes. Springer International Publishing, pp. 37–48.

97.

Baid

Talbar

Rane

, et al. Deep learning radiomics algorithm for gliomas (DRAG) model: A novel approach using 3D UNET based deep convolutional neural network for predicting survival in Gliomas. In: BrainLes. Springer International Publishing, pp. 369–379.

98.

Sun

Chen

Peng

, et al. DRRNet: dense residual refine networks for automatic brain tumor segmentation. J Med Syst 2019; 43: 1–9.

99.

Cahall

Rasool

Bouaynaya

, et al. Inception modules enhance brain tumor segmentation. Front Comput Neurosci 2019; 13: 1–8.

100.

Gong

, et al. Multi-scale masked 3-D U-net for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

101.

Isensee

Kickingereder

Wick

, et al. No new-net. BrainLes 2019; 2: 380–392.

102.

Chen

Liu

Peng

, et al. S3D-UNet: Separable 3D U-net for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

103.

Mehta

Arbel

. 3D U-Net for brain tumour. In: BrainLes. Springer International Publishing, pp. 254–266.

104.

Weninger

Rippel

Koppers

, et al. Segmentation of brain tumors and patient survival prediction: methods for the BraTS 2018 challenge. In: BrainLes. Springer International Publishing, pp. 380–392.

105.

Gering

Sun

Avery

, et al. Semi-automatic brain tumor segmentation by drawing long axes on multi-plane reformat. In: BrainLes. Springer International Publishing, pp. 380–392.

106.

Beers

Chang

Brown

, et al. Sequential neural networks for biologically-informed glioma segmentation. In: SPIE Medical Imaging. 2018, p. 108.

107.

Sun

Zhang

Luo

. Tumor segmentation and survival prediction in glioma with deep learning. In: BrainLes. Springer International Publishing, pp. 380–392.

108.

Hasan

SMK

Linte

. A modified u-net convolutional network featuring a nearest-neighbor re-sampling-based elastic-transformation for brain tissue characterization and segmentation. In: IEEE Western New York Image and Signal Processing Workshop (WNYISPW). IEEE, 2018, pp. 1–5.

109.

Gholami

Subramanian

Shenoy

, et al. A novel domain adaptation framework for medical image segmentation. In: BrainLes. Springer International Publishing, pp. 289–298.

110.

Wang

. A novel end-to-end brain tumor segmentation method using improved fully convolutional networks. Comput Biol Med 2019; 108: 150–160.

111.

Stawiaski

. A pretrained densenet encoder for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 105–115.

112.

Carver

Liu

Zong

, et al. Automatic brain tumor segmentation and overall survival prediction using machine learning algorithms. In: BrainLes. Springer International Publishing, pp. 406–418.

113.

Wang

Ourselin

, et al. Automatic brain tumor segmentation using convolutional neural networks with test-time augmentation. In: BrainLes. 2019, pp. 61–72.

114.

Isensee

Kickingereder

Wick

, et al. Brain tumor segmentation and radiomics survival prediction: contribution to the BRATS 2017 challenge. In: BrainLes. 2018, pp. 287–297.

115.

Myronenko

. 3D MRI brain tumor segmentation using autoencoder regularization. In: BrainLes. Springer International Publishing, pp. 380–392.

116.

Vaidhya

Thirunavukkarasu

Alex

, et al. Deep convolutional encoder networks for multiple sclerosis lesion segmentation. In: BrainLes, pp. 144–155.

117.

Alex

Vaidhya

Thirunavukkarasu

, et al. Semisupervised learning using denoising autoencoders for brain lesion detection and segmentation. J Med Imaging 2017; 4: 1.

118.

Rezaei

Harmuth

Gierke

, et al. A conditional adversarial network for semantic segmentation of brain tumor. In: BrainLes, pp. 138–149.

119.

Wang

. Brain tumor segmentation using an adversarial network. In: BrainLes, pp. 138–149.

120.

Mok

TCW

Chung

ACS

. Learning data augmentation for brain tumor segmentation with coarse-to-fine generative adversarial networks. In: BrainLes. Springer International Publishing, pp. 70–80.

121.

Yao

Zhou

Zhang

. Automatic segmentation of brain tumor using 3D SE-inception networks with residual connections. In: BrainLes. Springer International Publishing, pp. 380–392.

122.

Castillo

Daza

Rivera

, et al. Brain tumor segmentation and parsing on MRIs using multiresolution neural networks. BrainLes 2018; 10670: 138–149.

123.

Pourreza

Zhuge

Ning

, et al. Brain tumor segmentation in MRI scans using deeply-supervised neural networks. In: BrainLes, pp. 138–149.

124.

Tchikindas

Sparks

Baccon

, et al. Segmentation of nodular medulloblastoma using random walker and hierarchical normalized cuts. IEEE 37th Annu Northeast Bioeng Conf NEBEC 2011 2011; 1–2.

125.

Chandra

Vakalopoulou

Fidon

, et al. Context aware 3D CNNs for brain tumor segmentation. In: BrainLes, pp. 380–392.

126.

Pawar

Chen

Shah

, et al. Residual encoder and convolutional decoder neural network for glioma segmentation. BrainLes 2018; 10670: 138–149.

127.

Benson

Pound

French

, et al. Deep hourglass for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

128.

Shen

. Deep learning based multimodal brain tumor diagnosis. In: BrainLes, pp. 138–149.

129.

McKinley

Meier

Wiest

. Ensembles of densely-connected CNNs with label-uncertainty for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

130.

McKinley

Jungo

Wiest

, et al. Pooling-free fully convolutional networks with dense skip connections for semantic segmentation, with application to brain tumor segmentation. In: BrainLes, pp. 138–149.

131.

Naceur

Saouli

Akil

, et al. Fully automatic brain tumor segmentation using end-to-end incremental deep neural networks in MRI images. Comput Methods Programs Biomed 2018; 166: 39–49.

132.

Islam

Jose

VJM

Ren

. Glioma prognosis: segmentation of the tumor and survival prediction using shape, geometric and clinical information. In: BrainLes. Springer International Publishing, pp. 380–392.

133.

Islam

Ren

. Multi-modal PixelNet for brain tumor segmentation. In: BrainLes, pp. 138–149.

134.

Puch

Sanchez

Hernandez

, et al. Global planar convolutions for improved context aggregation in brain tumor segmentation. In: BrainLes, pp. 380–392.

135.

Randhawa

Modi

Jain

, et al. Improving boundary classification for brain tumor segmentation and longitudinal disease progression. In: BrainLes, pp. 65–74.

136.

Zhou

Chen

Ding

, et al. Learning contextual and attentive information for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

137.

Banerjee

Mitra

Shankar

. Multi-planar spatial-convnet for segmentation and survival prediction in brain cancer. In: BrainLes. Springer International Publishing, pp. 380–392.

138.

Wang

Zhu

, et al. Multimodal brain tumor image segmentation using WRN-PPNet. Comput Med Imaging Graph 2019; 75: 56–65.

139.

Colmeiro

RGR

Verrastro

Grosges

. Multimodal brain tumor segmentation using 3D convolutional networks. In: BrainLes, pp. 138–149.

140.

Hua

Huo

Gao

, et al. Multimodal brain tumor segmentation using cascaded V-Nets. In: BrainLes. Springer International Publishing, pp. 380–392.

141.

Han

W-S

Han

. Neuromorphic neural network for multimodal brain image segmentation and overall survival analysis. In: BrainLes. Springer International Publishing, pp. 380–392.

142.

Choudhury

Vanguri

Jambawalikar

, et al. Segmentation of brain tumors using DeepLabv3+. In: BrainLes. Springer International Publishing, pp. 380–392.

143.

Grivalsky

Tamajka

Benesova

. Segmentation of gliomas in magnetic resonance images using recurrent neural networks. In: 2019 42nd International Conference on Telecommunications and Signal Processing, TSP 2019. IEEE, 2019, pp. 539–542.

144.

Kori

Soni

Pranjal

, et al. Ensemble of fully convolutional neural network for brain tumor segmentation from magnetic resonance images. In: BrainLes. Springer International Publishing, pp. 380–392.

145.

Fidon

Garcia-Peraza-Herrera

, et al. Generalised Wasserstein dice score for imbalanced multi-class segmentation using holistic convolutional networks. In: BrainLes, pp. 138–149.

146.

Jungo

McKinley

Meier

, et al. Towards uncertainty-assisted brain tumor segmentation and survival prediction. In: BrainLes, pp. 138–149.

147.

Zhou

, et al. TPCNN: two-phase patch-based convolutional neural network for automatic brain tumor segmentation and survival prediction. In: BrainLes, pp. 138–149.

148.

Yang

Song

. A deep learning model integrating SK-TPCNN and random forests for brain tumor segmentation in MRI. Biocybern Biomed Eng 2019; 39: 613–623.

149.

Chahal

Haritosh

Gupta

, et al. Deep learning model for brain tumor segmentation analysis. 2019 3rd Int Conf Recent Dev Control Autom Power Eng RDCAPE 2019 2019; 378–383.

150.

Zhang

Song

Liu

, et al. Efficient 3D depthwise and separable convolutions with dilation for brain tumor segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2019, DOI: 10.1007/978-3-030-35288-2_45.

151.

Kaldera

HNTK

Gunasekara

Dissanayake

. MRI based glioma segmentation using deep learning algorithms. In: International Research Conference on Smart Computing and Systems Engineering (SCSE). 2019. Epub ahead of print 2019. DOI: 10.23919/SCSE.2019.8842668.

152.

Derikvand

Khotanlou

. Patch and pixel based brain tumor segmentation in MRI images using convolutional neural networks. 5th Iran Conf Signal Process Intell Syst ICSPIS 2019 2019; 18–19.

153.

Wacker

Ladeira

Nascimento

JEV

. Transfer learning for brain tumor segmentation. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2021; 12658 LNCS: 241–251.

154.

Derikvand

Khotanlou

. Brain tumor segmentation in MRI images using a hybrid deep network based on patch and pixel. Iran Conf Mach Vis Image Process MVIP 2020; 2020–Febru: 5–9.

155.

Ali

Gilani

Waris

, et al. Brain tumour image segmentation using deep networks. IEEE Access 2020; 8: 153589–153598.

156.

Hamghalam

Lei

Wang

. Convolutional 3d to 2d patch conversion for pixel-wise glioma segmentation in MRI scans. In: Crimi

Bakas

(eds) Brainles 2019, LNCS 11992. Springer, Cham: Springer International Publishing, 2020, pp.3–12.

157.

Mzoughi

Njeh

Ben

, et al. Glioblastomas brain tumor segmentation using optimized U-Net based on deep fully convolutional networks (D-FCNs). 2020 Int Conf Adv Technol Signal Image Process ATSIP 2020 2020; 1–6.

158.

Amian

Soltaninejad

. Multi-resolution 3d CNN for MRI brain tumor segmentation and survival prediction. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI:10.1007/978-3-030-46640-4_21.

159.

Michael Mahesh

Arokia Renjit

. Multiclassifier for severity-level categorization of glioma tumors using multimodal magnetic resonance imaging brain images. Int J Imaging Syst Technol 2020; 30: 234–251.

160.

van der Voort

Incekara

Wijnenga

MMJ

, et al. WHO 2016 subtyping and automated segmentation of glioma using multi-task deep learning. arXiv 2020; 1–49.

161.

Qamar

Jin

Zheng

, et al. 3D hyper-dense connected convolutional neural network for brain tumor segmentation. Proc - 2018 14th Int Conf Semant Knowl Grids, SKG 2018 2018; 123–130.

162.

Xia

. 3D deep neural network-based brain tumor segmentation using multimodality magnetic resonance sequences. In: BrainLes, pp. 138–149.

163.

Lachinov

Vasiliev

Turlapov

. Glioma segmentation with cascaded UNet. In: BrainLes. Springer International Publishing, pp. 380–392.

164.

Cui

Mao

Jiang

, et al. Automatic semantic segmentation of brain gliomas from MRI images using a deep cascaded neural network. J Healthc Eng 2018; 2018: 1–14. Epub ahead of print 2018. DOI: 10.1155/2018/4940593

165.

Liu

Qian

, et al. Multipath densely connected convolutional neural network for brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 81–91.

166.

Ahuja

Panigrahi

Gandhi

. Transfer learning based brain tumor detection and segmentation using superpixel technique. 2020 Int Conf Contemp Comput Appl IC3A 2020 2020; 244–249.

167.

Buatois

Puybareau

Tochon

, et al. Two stages CNN-based segmentation of gliomas, uncertainty quantification and prediction of overall patient survival. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2020; 11993 LNCS: 167–178.

168.

Wan

Rahmat

Price

. Deep learning for glioblastoma segmentation using preoperative magnetic resonance imaging identifies volumetric features associated with survival. Acta Neurochir (Wien) 2020; 162: 3067–3080.

169.

Rahmat

Saednia

Haji Hosseini Khani

, et al. Multi-scale segmentation in GBM treatment using diffusion tensor imaging. Comput Biol Med 2020; 123: 103815.

170.

Yang

H-Y

Yang

. Automatic brain tumor segmentation with contour aware residual network and adversarial training. In: BrainLes. Springer International Publishing, pp. 380–392.

171.

Zhao

, et al. Hierarchical multi-class segmentation of glioma images using networks with multi-level activation function. In: BrainLes. Springer International Publishing, pp. 380–392.

172.

Dai

Shu

, et al. Automatic brain tumor segmentation with domain adaptation. In: BrainLes. Springer International Publishing, pp. 380–392.

173.

Nuechterlein

Mehta

. 3D-ESPNet with pyramidal refinement for volumetric brain tumor image segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

174.

Noori

Bahri

Mohammadi

. Attention-guided version of 2D UNet for automatic brain tumor segmentation. 2019 9th Int Conf Comput Knowl Eng ICCKE 2019 2019; 269–275.

175.

Agravat

Raval

. Brain tumor segmentation and survival prediction. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2020; 11992 LNCS: 338–348.

176.

Xie

Liu

, et al. Deep cascaded attention network for multi-task brain tumor segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2019, DOI: 10.1007/978-3-030-32248-9_47.

177.

Shanis

Gerber

Gao

, et al. Intramodality domain adaptation using self ensembling and adversarial training. Springer, Cham: Springer International Publishing, 2019, pp. 90–98.

178.

Cheng

Liu

, et al. Multi-level glioma segmentation using 3D U-net combined attention mechanism with atrous convolution. Proc - 2019 IEEE Int Conf Bioinforma Biomed BIBM 2019 2019; 1031–1036.

179.

Chang

Beers

Bai

, et al. Automatic assessment of glioma burden: a deep learning algorithm for fully automated volumetric and bidimensional measurement. Neuro Oncol 2019; 21: 1412–1422.

180.

Rosas González

Birgui Sekou

Hidane

, et al. 3D automatic brain tumor segmentation using a multiscale input U-Net network. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2020; 11993 LNCS: 113–123.

181.

Agravat

Raval

. 3D semantic segmentation of brain tumor for overall survival prediction. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2021; 12659 LNCS: 215–227.

182.

Xue

Xie

Farhat

, et al. A multi-path decoder network for brain tumor segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_25.

183.

Aledhari

Razzak

. An adaptive segmentation technique to detect brain tumors using 2D UNet. Proc - 2020 IEEE Int Conf Bioinforma Biomed BIBM 2020 2020; 2328–2334.

184.

Zhenyu Z, Gao S and Huang Z. An automatic glioma segmentation system based on a separable attention U-Net (SAUNet). In 2020 9th International Conference on Bioinformatics and Biomedical Science (ICBBS '20). New York, NY, USA: Association for Computing Machinery, 95–101. 10.1145/3431943.3431959

185.

Dong

Jiang

, et al. Annotation-free gliomas segmentation based on a few labeled general brain tumor images. In: IEEE 17th International Symposium on Biomedical Imaging (ISBI). Iowa City, Iowa, USA, 2020, pp. 354–358.

186.

Sameer

Bayat

Mohammed

. Brain tumor segmentation and classification approach for MR images based on convolutional neural networks. Proc 2020 1st Inf Technol to Enhanc E-Learning other Appl Conf IT-ELA 2020 2020; 138–143.

187.

Naser

Deen

. Brain tumor segmentation and grading of lower-grade glioma using deep learning in MRI images. Comput Biol Med 2020; 121: 103758.

188.

Guo

Yang

, et al. Brain tumor segmentation based on attention mechanism and multi-model fusion. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_5.

189.

Chang

. Brain tumor segmentation on multimodal 3D-MRI using deep learning method. Proc - 2020 13th Int Congr Image Signal Process Biomed Eng Informatics, CISP-BMEI 2020 2020; 635–639.

190.

Feng

Tustison

Patel

, et al. Brain tumor segmentation using an ensemble of 3D U-nets and overall survival prediction using radiomic features. Front Comput Neurosci 2020; 14: 1–12.

191.

Zhao

. Brain tumor segmentation using attention-based network in 3D MRI images. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_1.

192.

Shi

Pang

, et al. Brain tumor segmentation using dense channels 2d u-net and multiple feature extraction networks. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46640-4_26.

193.

Baid

Shah

Talbar

. Brain tumor segmentation with cascaded deep convolutional neural network. In: Crimi

Bakas

(eds) Brainlesion: glioma, multiple sclerosis, stroke and traumatic brain injuries. BrainLes 2019. Lecture notes in computer science, vol 11993. Springer, Cham: Springer International Publishing, 2020, pp. 90–98.

194.

Feng

Dou

Tustison

, et al. Brain tumor segmentation with uncertainty estimation and overall survival prediction. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46640-4_29.

195.

Long

Liu

, et al. Cascaded hybrid residual U-Net for glioma segmentation. Multimed Tools Appl 2020; 79: 24929–24947.

196.

Kotowski

Nalepa

Dudzik

. Detection and segmentation of brain tumors from MRI using U-nets. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_17.

197.

Banerjee

Arora

Mitra

. Ensemble of CNNs for segmentation of glioma sub-regions with survival prediction. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_4.

198.

Bangalore Yogananda

Wagner

Nalawade

, et al. Fully automated brain tumor segmentation and survival prediction of gliomas using deep learning and MRI. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_10.

199.

Ribalta Lorenzo

Marcinkiewicz

Nalepa

. Multi-modal U-Nets with boundary loss and pre-training for brain tumor segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_13.

200.

Peng

Chen

Sun

, et al. Multi-Scale 3D U-nets: an approach to automatic segmentation of brain tumor. Int J Imaging Syst Technol 2020; 30: 5–17.

201.

Dutta

Chakraborty

Mondal

. Multimodal segmentation of brain tumours in volumetric MRI scans of the brain using time-distributed U-Net. Springer Singapore, Epub ahead of print 2020, DOI: 10.1007/978-981-13-9042-5_62.

202.

Wang

. Neural architecture search for gliomas segmentation on multimodal magnetic resonance Imaging, http://arxiv.org/abs/2005.06338 (2020).

203.

Juan-Albarracín

Fuster-Garcia

del Mar Álvarez-Torres

, et al. Oncohabitats glioma segmentation model. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2020; 11992 LNCS: 295–303.

204.

Yin

. Partial view segmentation: a novel approach to the brain tumor segmentation. 2020 IEEE 3rd Int Conf Comput Commun Eng Technol CCET 2020 2020; 76–79.

205.

Zhang

, et al. SU-Net: an efficient encoder-decoder model of federated learning for brain tumor segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-61609-0_60.

206.

Yin

Liu

, et al. The tumor mix-up in 3D unet for glioma segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_26.

207.

Kim

Luna

Chikontwe

, et al. Two-step u-nets for brain tumor segmentation and random forest with radiomics for survival time prediction. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46640-4_19.

208.

Kot

Krawczyk

Siwek

, et al. U-Net and Active Contour Methods for Brain Tumour Segmentation and Visualization. Proc Int Jt Conf Neural Networks. Epub ahead of print 2020. DOI: 10.1109/IJCNN48605.2020.9207572.

209.

Bangalore Yogananda

Shah

Vejdani-Jahromi

, et al. A novel fully automated MRI-based deep-learning method for classification of IDH mutation status in brain gliomas. Neuro Oncol 2020; 22: 402–411.

210.

Dong

Yang

Liu

, et al. Automatic brain tumor detection and segmentation using U-net based fully convolutional networks. Commun Comput Inf Sci 2017; 723: 506–517.

211.

Russo

Liu

Ieva

. Spherical coordinates transformation pre-processing in Deep Convolution Neural Networks for brain tumor segmentation in MRI. 1–26.

212.

Rezaei

Yang

Meinel

. Voxel-GAN: adversarial framework for learning imbalanced brain tumor segmentation. In: BrainLes. Springer International Publishing, pp. 380–392.

213.

Carver

Dai

Liang

, et al. Improvement of multiparametric MR image segmentation by augmenting the data with generative adversarial networks for glioma patients. Front Comput Neurosci 2021: 1–10. Epub ahead of print 2021; 14. DOI: 10.3389/fncom.2020.495075

214.

Hamghalam

Wang

Lei

. High tissue contrast image synthesis via multistage attention-GAN: application to segmenting brain MR scans. Neural Netw 2020; 132: 43–52.

215.

Stefani

Rahmat

Harris-Birtill

. Autofocus net: auto-focused 3D CNN for brain tumour segmentation. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-52791-4_4.

216.

Din

NKAM

Rahni

AAA

. Evaluation of a deep learning based brain tumour segmentation method. In: Journal of Physics: Conference Series 2020. Epub ahead of print 2020. DOI: 10.1088/1742-6596/1497/1/012009

217.

Nyholm

Löfstedt

. Multi-decoder networks with multi-denoising inputs for tumor segmentation. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2021; 12658 LNCS: 412–423.

218.

Nyholm

Löfstedt

. Tunet: end-to-end hierarchical brain tumor segmentation using cascaded networks. Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 2020; 11992 LNCS: 174–186.

219.

Casamitjana

Cata

Sanchez

, et al. Cascaded V-Net using ROI masks for brain tumor segmentation. In: BrainLes, pp. 138–149.

220.

Lopez

Ventura

. Dilated convolutions for brain tumor segmentation in MRI scans. In: BrainLes, pp. 138–149.

221.

Chang

. Fully convolutional deep residual neural networks for brain tumor segmentation. In: BrainLes, pp. 108–118.

222.

Ermiş

Jungo

Poel

, et al. Fully automated brain resection cavity delineation for radiation target volume definition in glioblastoma patients using deep learning. Radiat Oncol 2020; 15: 1–10.

223.

Banerjee

Mitra

Sharma

, et al. A CADe system for gliomas in brain MRI using convolutional neural networks, http://arxiv.org/abs/1806.07589 (2018).

224.

Wang

Zhu

, et al. A deep learning algorithm for fully automatic brain tumor segmentation. Proc Int Jt Conf Neural Networks 2019; 2019–July: 1–5.

225.

, et al. An intelligent diagnosis method of brain MRI tumor segmentation using deep convolutional neural network and SVM algorithm. Comput Math Methods Med 2020; 2020: 1–10. Epub ahead of print 2020. DOI: 10.1155/2020/6789306

226.

Alqazzaz

Sun

Yang

, et al. Automated brain tumor segmentation on multi-modal MR image using SegNet. Comput Vis Media 2019; 5: 209–219.

227.

Zarrar

Hussain

Khan

, et al. Latest trends in automatic glioma tumor segmentation and an improved convolutional neural network based solution. MACS 2019 - 13th Int Conf Math Actuar Sci Comput Sci Stat Proc. Epub ahead of print 2019. DOI: 10.1109/MACS48846.2019.9024815.

228.

Zhou

Shi

, et al. 3D dense connectivity network with atrous convolutional feature pyramid for brain tumor segmentation in magnetic resonance imaging of human heads. Comput Biol Med 2020; 121: 1–11.

229.

Guo

Yang

Lam

, et al. Domain knowledge based brain tumor segmentation and overall survival prediction. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46643-5_28.

230.

Ngo

Tran

Kim

, et al. Multi-task learning for small brain tumor segmentation from MRI. Appl Sci 2020; 10: 1–16.

231.

Nie

Ding

Shang

, et al. Spatial attention-based efficiently features fusion network for 3D-MR brain tumor segmentation. Proc 2020 IEEE Int Conf Prog Informatics Comput PIC 2020 2020; 67–74.

232.

Liu

Guo

. Improving brain tumor segmentation with multi-direction fusion and fine class prediction. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-46640-4_33.

233.

Jia

Xia

Cai

, et al. Learning high-resolution and efficient non-local features for brain glioma segmentation in MR images. Springer, Cham: Springer International Publishing, Epub ahead of print 2020, DOI: 10.1007/978-3-030-59719-1_47.

234.

Dorent

Booth

, et al. Learning joint segmentation of tissues and brain lesions from task-specific hetero-modal domain-shifted datasets. Med Image Anal Epub ahead of print 2021; 67: 1–16. DOI: 10.1016/j.media.2020.101862

235.

Acharya

Alsadoon

Al-Janabi

, et al. MRI-based diagnosis of brain tumours using a deep neural network framework. CITISIA 2020 - IEEE Conf Innov Technol Intell Syst Ind Appl Proc 2020; 1–5.

236.

Zhong

Shu

, et al. (TS)2WM: tumor segmentation and tract statistics for assessing white matter integrity with applications to glioblastoma patients. Neuroimage 2020; 223: 117368.

237.

Zeineldin

Karar

Coburger

, et al. Deepseg: deep neural network framework for automatic brain tumor segmentation using magnetic resonance FLAIR images. Int J Comput Assist Radiol Surg 2020; 15: 909–920.

238.

Kawa

Rudzki

Pietka

, et al. Computer aided diagnosis workstation for brain tumor assessment. Proc 22nd Int Conf Mix Des Integr Circuits Syst Mix 2015 2015; 98–103.

239.

Sauwen

Sima

Van Cauter

, et al. Hierarchical non-negative matrix factorization to characterize brain tumor heterogeneity using multi-parametric MRI. NMR Biomed 2015; 28: 1599–1624.

240.

Fathi Kazerooni

Mohseni

Rezaei

, et al. Multi-parametric (ADC/PWI/T2-w) image fusion approach for accurate semi-automatic segmentation of tumorous regions in glioblastoma multiforme. Magn Reson Mater Physics, Biol Med 2015; 28: 13–22.

241.

Sauwen

Acou

Sima

, et al. Semi-automated brain tumor segmentation on multi-parametric MRI using regularized non-negative matrix factorization. BMC Med Imaging 2017; 17: 1–14.

242.

Zhang

, et al. Non-model segmentation of brain glioma tissues with the combination of DWI and fMRI signals. Biomed Mater Eng 2015; 26: S1315–S1324.

243.

Rahmat

Brochu

, et al. Semi-automated construction of patient individualised clinical target volumes for radiotherapy treatment of glioblastoma utilising diffusion tensor decomposition maps. Br J Radiol Epub ahead of print 2020; 93: 1–7. DOI: 10.1259/bjr.20190441

244.

Zhang

Dou

Zhang

, et al. A framework of automatic brain tumor segmentation method based on information fusion of structural and functional MRI signals. Proc 2016 8th IEEE Int Conf Commun Softw Networks, ICCSN 2016 2016; 625–629.

245.

Sun

Wang

Guo

, et al. A potential field segmentation based method for tumor segmentation on multi-parametric MRI of glioma cancer patients. BMC Med Imaging 2019; 19: 1–9.

246.

Artzi

Blumenthal

Bokstein

, et al. Classification of tumor area using combined DCE and DSC MRI in patients with glioblastoma. J Neurooncol 2015; 121: 349–357.

247.

Beigi

Safari

Ameri

, et al. Findings of DTI-p maps in comparison with T2/T2-FLAIR to assess postoperative hyper-signal abnormal regions in patients with glioblastoma 08 information and computing sciences 0801 artificial intelligence and image processing. Cancer Imaging 2018; 18: 1–7.

248.

Huber

Alber

Bette

, et al. Reliability of semi-automated segmentations in glioblastoma. Clin Neuroradiol 2017; 27: 153–161.

249.

Nitsch

Klein

Dammann

, et al. Automatic and efficient MRI-US segmentations for improving intraoperative image fusion in image-guided neurosurgery. NeuroImage Clin 2019; 22: 101766.

250.

Niyazi

Brada

Chalmers

, et al. ESTRO-ACROP guideline ‘target delineation of glioblastomas’. Radiother Oncol 2016; 118: 35–42.

251.

Guo

Wang

Feng

, et al. Diffusion and perfusion weighted magnetic resonance imaging for tumor volume definition in radiotherapy of brain tumors. Radiat Oncol 2016; 11: 1–13.

252.

Kono

Inoue

Nakayama

, et al. The role of diffusion-weighted imaging in patients with brain tumors. Am J Neuroradiol 2001; 22: 1081–1088.

253.

Massimino

Biassoni

Gandola

, et al. Childhood medulloblastoma. Crit Rev Oncol Hematol 2016; 105: 35–51.

254.

Lassaletta

Ramaswamy

. Medulloblastoma in adults: they’re not just big kids. Neuro Oncol 2016; 18: 895–897.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB