Sage Journals: Discover world-class research

Abstract

Background

Our purpose is to develop a model combining radiomic features of radiotherapy localisation computed tomography and clinical characteristics that can be used to estimate overall survival in patients with nasopharyngeal carcinoma treated with intensity-modulated radiotherapy following induction chemotherapy.

Methods

We recruited 145 patients with pathologically confirmed nasopharyngeal carcinoma between February 2012 and April 2015. In total, 851 radiomic features were extracted from radiotherapy localisation computed tomography images for the gross tumour volume of the nasopharynx and the gross tumour volume of neck metastatic lymph nodes. The least absolute shrinkage and selection operator algorithm was applied to select radiomics features, build the model and calculate the Rad-score. The patients were divided into high- and low-risk groups based on their Rad-scores. A nomogram for estimating overall survival based on both radiomic and clinical features was generated using multivariate Cox regression hazard models. Prediction reliability was evaluated using Harrell’s concordance index.

Results

In total, seven radiomic features and one clinical characteristic were extracted for survival analysis, and the combination of radiomic and clinical features was a better predictor of overall survival (concordance index = .849 [confidence interval: .782-.916]) than radiomic features (concordance index = .793 [confidence interval: .697-.890]) or clinical characteristics (concordance index = .661 [confidence interval: .673-.849]) alone.

Conclusion

Our results show that a nomogram combining radiomic features of radiotherapy localisation computed tomography and clinical characteristics can predict overall survival in patients with nasopharyngeal carcinoma treated with intensity-modulated radiotherapy following induction chemotherapy more effectively than radiomic features or clinical characteristics alone.

Keywords

radiomic nasopharyngeal carcinoma radiotherapy localisation computed tomography nomogram survival prediction

Introduction

Nasopharyngeal carcinoma (NPC) is a unique form of head and neck cancer characterised by its distinct geographical distribution, with an average incidence of approximately 1.1% per 100,000 people worldwide.^1-3 NPC is also distinct from other head and neck cancers owing to its association with the Epstein–Barr virus, naturally aggressive behaviour and high risk of distant metastases.^4-6 Although improvements in living conditions and quality of life have helped to decrease the incidence of NPC, it is relatively more prevalent in East and Southeast Asia than in other regions.

Treatment options for NPC include radiotherapy alone, concurrent chemoradiotherapy and induction chemotherapy plus concurrent chemoradiotherapy.⁷ Radiotherapy is the primary treatment for patients with NPC. The current standard treatment for locoregionally advanced NPC is concurrent chemoradiotherapy with or without induction chemotherapy. Advancements in imaging technology, such as the development of intensity-modulated radiotherapy (IMRT), have substantially improved survival rates among patients with advanced NPC.⁸ However, there is no effective method for predicting prognosis in patients with NPC.

Several recent studies have demonstrated that radiomic-based methods can be used to predict the prognosis of patients and guide clinical decision-making.^9-11 Radiomic analysis, in which several high-throughput quantitative features are extracted from radiological images, provides valuable information associated with tumoural heterogeneity.¹² To date, radiomic analyses have been used to predict survival in patients with various types of tumours, such as hepatocellular carcinoma, gastric carcinoma and breast cancer.^13-15 Some scholars have reported that a nomogram combining radiomic and clinical features can be used as a convenient and reliable tool for predicting prognosis in patients with NPC.^16-18 Radiomic features are extracted from radiotherapy localisation computed tomography (CT) images obtained prior to radiotherapy, which differ from conventional CT images, in that they allow clinicians to directly sketch regions of interest (ROIs), saving time for both clinicians and patients. Although radiotherapy localisation CT is widely used in hospitals, radiomic analysis based on radiotherapy localisation CT is rare. Therefore, in this study, we aimed to develop and validate a novel radiotherapy localisation CT-based radiomic model for predicting survival among patients with NPC treated with IMRT following induction chemotherapy.

Material and Methods

Patients

Patients were considered eligible if they met the following inclusion criteria: age ≥18 years; pathologically confirmed NPC; Karnofsky functional status score of at least 70 points; at least one measurable lesion that has not received local treatment, such as radiotherapy, according to the Response Evaluation Criteria in Solid Tumours 1.1; and expected survival ≥3 months. Treatments and data analyses were conducted in accordance with the Declaration of Helsinki. All patients were treated at the Affiliated Hospital of Southwest Medical University in Luzhou City, Sichuan Province, China, from February 2012 to April 2015. Ethical approval was obtained from the Medicine Ethics Committee of the Affiliated Hospital of Southwest Medical University. Diagnoses and treatments for NPC were based on the guidelines of the Chinese Society of Clinical Oncology. All patients underwent magnetic resonance imaging (MRI) initially and were subsequently diagnosed with NPC based on the results of a pathological biopsy.

First, all patients received induction chemotherapy with either paclitaxel plus cisplatin or 5-fluorouracil plus cisplatin every 3 weeks for a total of two cycles. Concurrent chemoradiotherapy with cisplatin plus IMRT was performed at week 7 to 14 of the total treatment cycle

IMRT

In recent years, IMRT has been widely used in the treatment of various head and neck tumours, including NPC and other rare cancers.^19,20 In addition, a clinical study reported that IMRT not only significantly improved treatment outcomes but also reduced the incidence of some side effects.²¹ Intensity-modulated radiation allows the clinician to adjust the intensity of each radiation according to the actual situation, so each beam field can have one or more high-intensity radiation areas and any number of low-intensity areas within the same field, thus allowing greater control over the dose distribution with the target.²² IMRT uses linear accelerators to safely deliver a precise dose of radiation to the tumour while minimising the dose delivered to the surrounding normal tissue.

Clinical Endpoints and Follow-Up

After the completion of treatment, all patients were monitored every 3 months during the first 2 years, every 6 months for the next 3 years and annually thereafter. The follow-up rate was 93.3%. All patients were followed-up for more than 2 years, and the median follow-up duration was 27 months. Overall survival (OS) was calculated from the start of IMRT to the date of death or the last follow-up visit.

ROI Segmentation

ROIs included the gross tumour volume of the nasopharynx (GTVnx) and gross tumour volume of neck metastatic lymph nodes (GTVnd). They were sketched by a specialist with more than 15 years of experience in head and neck cancer research using a Philips Pinnacle 9.10 planning system (Philips, Cleveland, Ohio, USA).

Radiomic Analysis

Image Acquisition

The images for radiomic analysis were obtained from radiotherapy localisation CT performed prior to IMRT. Radiotherapy localisation CT was performed using a LightSpeed RT 4 scanner (GE Healthcare, Chicago, IL, USA). The scanning parameters used in this study were as follows: tube voltage, 120 kVp; field-of-view, 250-400 mm; pixel size, 512 × 512; slice thickness, .25 cm; and average number of slices, 116. Before feature extraction, all images were resampled according to a voxel size of 1×1×1 mm³. Wavelet-based methods were used to process the CT images from which the radiomics features of the GTVnx and GTVnd were extracted.

Feature extraction

The radiomic features of GTVnx and GTVnd were extracted via the three-dimensional (3 D) slicer platform using the ‘pyradiomics’ package: (available at: http://PyRadiomics.readthedocs.io/en/latest/) (accessed on June 10, 2020).²³

Statistical Analyses

All statistical analyses were performed using R software (version 3.6.3; R Foundation for Statistical Computing, Vienna, Austria) and X-tile software (version 3.6.1; Yale University School of Medicine, New Haven, Conn). Data dimension reduction, feature selection and radiomics feature building were achieved through least absolute shrinkage and selection operator (LASSO) Cox regression modelling to select the most valuable predictive radiomic features from the gross tumour volume. Multivariate Cox regression hazard models were used to assess survival outcomes, radiomic features and clinical characteristics. Radiomic features and clinical characteristics were presented using a nomogram, which graphically displayed the prediction model using numerical scale lines based on traditional statistical methods. Radiomic features were selected via LASSO regression to fit the Cox proportional model using the ‘glmnet’ package in the R software. The ‘survival’ and ‘rms’ packages in R software were used for the multivariate Cox regression hazard models, nomogram and calibration curve, respectively. Harrell’s concordance index (C-index) was used to evaluate the nomogram model. The linear combination of selected radiomic features was used to calculate the radiomic scores (Rad-scores) for each patient, and the Rad-scores were weighted by their respective coefficients. X-tile software was used to calculate the cut-off value of the Rad-score, which was used to divide patients into high-risk and low-risk groups. Associations between radiomic features and clinical parameters were investigated using heat maps, and heatmap analysis was performed using the ‘pheatmap’ package in R (version 3.6.3).

Results

Table 1 shows the characteristics of the 145 patients enrolled in the study. In total, 851 radiomic features were extracted, including grey-level dependence matrix features (GLDM, which quantify grey-level dependencies in an image), grey-level co-occurrence matrix features (GLCM, which describe the second-order joint probability function of an image region constrained by the mask), grey-level size zone matrix features (GLSZM, which quantify grey-level zones in an image) and neighbouring grey-tone difference matrix features (NGTDM, which quantify the difference between a grey value and the average grey value of its neighbours within some distance). LASSO regression reduced the high-throughput radiation signature (Figure 1), and seven radiomic features and one clinical feature were extracted for OS analysis. The radiomic features included large-dependence low grey-level emphasis (which measures the joint distribution of large dependence with lower grey-level values) of the GLDM of the original (feature 1), autocorrelation (which is a measure of the magnitude of the fineness and coarseness of texture) of the GLCM of the wavelet-LHL (H = high-frequency band, L = low-frequency band) (feature 2), large-area low grey-level emphasis (which measures the proportion in the image of the joint distribution of larger size zones with lower grey-level values) of the GLSZM of the wavelet-LLH (feature 3), busyness (which measures the change from a pixel to its neighbour) of the NGTDM of the wavelet-LLH (feature 4), busyness of the NGTDM of the wavelet-HLH (feature 5), busyness of the NGTDM of the wavelet-HHH (feature 6) and cluster prominence (which is a measure of the skewness and asymmetry) of the GLCM of the wavelet-HHL (feature 7). The overall stage was the selected clinical characteristic. Figure 2 shows the coefficients of the selected radiomic features. Features 1-7 consisted of radiomic features, and the Rad-score was calculated using the following formula: Rad-score = Feature 1 × .0,951,401 + Feature 2 × .0,427,186 + Feature 3 × .1,886,388 + Feature 4 × .0,810,722+ Feature 5 × .0,115,232 + Feature 6 × .1,299,493 + Feature 7 × .0,232,920.

Table 1.

Patient Characteristics.

Characteristics	N (%)
Age (years)
Median (Range)	50 (18-84)
Sex
Male	108 (74.4%)
Female	37 (25.5%)
Clinical T stage
T1	13 (8.9%)
T2	57 (39.3%)
T3	47 (32.4%)
T4	28 (19.3%)
Clinical N stage
N0	15 (10.3%)
N1	44 (30.3%)
N2	68 (46.8%)
N3	18 (12.4%)
Overall stage
II	21 (14.4%)
III	79 (54.4%)
IV	45 (31.0%)
<B>KPS</B>
70	26 (17.9%)
80	55 (37.9%)
90	64 (44.1%)

Abbreviations: KPS, Karnofsky Performance Status.

Figure 1.

LASSO regression reduced the high-throughput radiation signature. LASSO coefficient profiles of the 851 texture features (A). Tuning parameter (λ) selection in the LASSO model using 10-fold cross-validation with minimum criteria (B). LASSO: least absolute shrinkage and selection operator.

Figure 2.

Histograms showing the roles of individual features contributing to the radiomic features. Contributing features are plotted on the y-axis, and their LASSO coefficients are plotted on the x-axis.

The cut-off Rad-score was −.06, which was used to classify patients into a high-risk group (Rad-score ≥ −.06) and a low-risk group (Rad-score < −.06). The survival curves for both groups are shown in Figure 3. The high-risk group exhibited shorter OS than the low-risk group. The nomograms for radiomic and clinical features, as well as the nomogram for the combination of the two, and the corresponding calibration curves are shown in Figure 4. Heat maps revealed associations between the radiomic signature selected by LASSO regression and clinical parameters (Figure 5). The results of the heat map analysis indicated that features 1-7 were correlated with T stage (all P <.05). Our results also indicated that the combination of radiomic and clinical features was a better predictor of OS (C-index = .849 [confidence interval {CI}: .782-.916]) than radiomic features (C-index = .793 [CI: .697-.890]) or clinical features (C-index = .661 [CI: .673-.849]) alone.

Figure 3.

Kaplan–Meier curves for overall survival in the low-risk and high-risk groups.

Figure 4.

The nomogram and the associated calibration curve. The nomogram (left) and the associated calibration curve (right) developed when clinical characteristics alone were used (A). The nomogram (left) and the associated calibration curve (right) when radiomic features alone were used (B). The nomogram (left) and the associated calibration curve (right) for the combination of radiomic features and clinical characteristics (C).

Figure 5.

Heat map for radiomic features and clinical characteristics.

Discussion

In this study, we used radiotherapy localisation CT images to develop a novel radiomic model for predicting survival among patients with NPC treated with IMRT following induction chemotherapy. A total of 145 patients with NPC underwent radiotherapy localisation CT, and the images obtained were directly used by clinicians to outline the ROIs. Seven radiomic features and one clinical feature were selected for model construction. Our findings indicated that a nomogram combining these features can predict OS in patients with NPC treated with IMRT following induction chemotherapy more effectively than models based on radiomic features or clinical characteristics alone.

Radiomic analysis has been applied to many types of cancer, such as colorectal cancer, breast cancer, pancreatic ductal adenocarcinoma and hepatocellular carcinoma,^24-27 and previous studies have suggested its advantages in predicting treatment efficacy and prognosis among patients. Some studies have demonstrated that LASSO regression can be used for the accurate analysis of radiological characteristics in relatively small sample sizes and can also prevent overfitting of the model.^28,29 Radiomics combines radiomic features with clinical features to better predict OS, progression-free survival and disease-free survival, which effectively compensates for the lack of single clinical or radiomic predictions. In this study, we developed and validated a nomogram based on clinical and radiomic features obtained from radiotherapy localisation CT scans acquired before IMRT, which helped predict OS in patients with NPC. The method used in this study is simple and repeatable because these features are generated by a validated package, which is freely available from the 3 D slicer.²³ Furthermore, the information provided by this method may aid doctors in making clinical decisions for patients.

Several previous studies have reported that MRI-based radiomic models can be used to predict survival in patients with NPC.^30-32 At the same time, some scholars have reported the value of positron emission tomography-CT-based radiomic analysis in predicting NPC prognosis.³³ However, unlike these studies, we used radiotherapy localisation CT-based radiomic models to predict patient survival. Therefore, the biggest innovation of this study is that we used radiotherapy localisation CT as the basis for model construction to predict the prognosis of patients with NPC. Prior to radiotherapy, each patient underwent radiotherapy location CT. The images obtained from radiotherapy localisation CT can be directly used by clinicians to outline ROIs, which saves clinicians more time and improves clinical work efficiency compared with the use of images obtained from ordinary CT and MRI. All images used for radiomics analysis were acquired via radiotherapy localisation CT before IMRT. Ultimately, our results indicated that the combination of radiomic and clinical features was a better predictor of OS than radiomic features or clinical features alone.

This study had some limitations. Owing to the small sample size, a validation cohort was lacking. To overcome these limitations, our research group intends to recruit more patient groups to conduct a multicentre study. In addition, previous research has shown that OS should be assessed by following patients up for more than 5 years³⁴; however, the median follow-up duration in our study was only 27 months.

In summary, our findings indicated that radiotherapy localisation CT-based radiomic models can effectively predict the survival of patients with NPC. Recent studies have reported that MRI-based radiomic models can be used to diagnose NPC stage and as adjunct tools to evaluate local recurrence, thus providing personalised treatment for different patients.^31,35 Therefore, future studies should aim to develop and validate radiotherapy localisation CT-based radiomic models for staging and the prediction of local recurrence in patients with NPC. Furthermore, additional studies are required to determine whether radiotherapy localisation CT-based radiomic analyses can be applied to other cancers, such as lung cancer, liver cancer and other head and neck tumours. Such studies will not only provide diversified and more effective means for clinical diagnosis and treatment but will also provide patients with more personalised and effective treatments.

Conclusion

The combination of radiotherapy localisation CT-based radiomic features and clinical characteristics may serve as a reliable and powerful tool for predicting prognosis in patients with NPC treated with IMRT following induction chemotherapy.

Footnotes

Declarations of Conflicting Interest

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Approval

The study was approved by the Approved by Luzhou Medical College Affiliated Hospital drug clinical trial Committee.The approval number is K2012001.

ORCID iDs

Xiaoyue Li

Han Chen

References

Chen

Chan

ATC

Blanchard

Sun

. Nasopharyngeal carcinoma. Lancet. 2019; 394(10192): 64-80. doi: 10.1016/S0140-6736(19)30956-0.

Shield

Ferlay

Jemal

, et al. The global incidence of lip, oral cavity, and pharyngeal cancers by subsite in 2012. CA Cancer J Clin. 2017; 67(1): 51-64. doi: 10.3322/caac.21384.

Chan

Grégoire

Lefebvre

, et al. Nasopharyngeal cancer: EHNS-ESMO-ESTRO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2012; 23(Suppl 7): vii83-vii85. doi: 10.1093/annonc/mds266

Chan

KCA

Woo

JKS

King

, et al. Analysis of plasma epstein-barr virus DNA to screen for nasopharyngeal cancer. N Engl J Med. 2017; 377(6): 513-522. doi: 10.1056/NEJMoa1701717.

Vokes

Liebowitz

Weichselbaum

. Nasopharyngeal carcinoma. Lancet. 1997; 350(9084): 1087-1091. doi: 10.1016/S0140-6736(97)07269-3.

Lee

Lin

. Current management of nasopharyngeal cancer. Semin Radiat Oncol. 2012; 22(3): 233-244. doi: 10.1016/j.semradonc.2012.03.008.

Pfister

Spencer

Adelstein

, et al. Head and neck cancers, version 2.2020, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2020; 18(7): 873-898. doi: 10.6004/jnccn.2020.0031.

Xiao

Qiu

. The effectiveness of intensity-modulated radiation therapy versus 2D-RT for the treatment of nasopharyngeal carcinoma: A systematic review and meta-analysis. PLoS One. 2019; 14(7): e0219611. doi: 10.1371/journal.pone.0219611.

Han

Zhu

Liu

, et al. Radiomic nomogram for prediction of axillary lymph node metastasis in breast cancer. Eur Radiol. 2019; 29(7): 3820-3829. doi:10.1007/s00330-018-5981-2.

10.

Sah

Owczarczyk

Siddique

Cook

GJR

Goh

. Radiomics in esophageal and gastric cancer. Abdom Radiol (NY). 2019; 44(6): 2048-2058. doi:10.1007/s00261-018-1724-8.

11.

Zhang

Fang

Dong

, et al. Development and validation of a CT-based radiomic nomogram for preoperative prediction of early recurrence in advanced gastric cancer. Radiother Oncol. 2020; 145: 13-20. doi:10.1016/j.radonc.2019.11.023.

12.

Lambin

Rios-Velazquez

Leijenaar

, et al. Radiomics: Extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012; 48(4): 441-446. doi: 10.1016/j.ejca.2011.11.036.

13.

Park

Lim

, et al. Radiomics signature on magnetic resonance imaging: Association with disease-free survival in patients with invasive breast cancer. Clin Cancer Res. 2018; 24(19): 4705-4714. doi: 10.1158/1078-0432.CCR-17-3783.

14.

Zhang

Liu

, et al. Radiomic analysis of contrast-enhanced CT predicts microvascular invasion and outcome in hepatocellular carcinoma. J Hepatol. 2019; 70(6): 1133-1144. doi: 10.1016/j.jhep.2019.02.023.

15.

Jiang

Chen

Xie

, et al. Radiomics signature of computed tomography imaging for prediction of survival and chemotherapeutic benefits in gastric cancer. EBioMedicine. 2018; 36: 171-182. doi: 10.1016/j.ebiom.2018.09.007.

16.

Zhou

Pan

, et al. Development and validation of a nomogram for predicting long-term overall survival in nasopharyngeal carcinoma: A population-based study. Medicine (Baltim). 2020; 99(4): e18974. doi: 10.1097/MD.0000000000018974.

17.

Wang

Peng

, et al. Development and external validation of a nomogram for predicting the overall survival of patients with stage II nasopharyngeal carcinoma after curative treatment. Cancer Manag Res. 2019; 11: 4403-4412. doi: 10.2147/CMAR.S202151.

18.

Wang

, et al. Establishment and validation of a novel nomogram to predict overall survival in nasopharyngeal carcinoma with lymph node metastasis. Head Neck. 2021; 43(8): 2353-2363. doi: 10.1002/hed.26687.

19.

Luo

Jin

. The efficacy of induction chemotherapy in the treatment of stage II nasopharyngeal carcinoma in intensity modulated radiotherapy era. Oral Oncol. 2018; 85: 95-100. doi: 10.1016/j.oraloncology.2018.08.016.

20.

Doi

Uemoto

Masai

Tatsumi

Shiomi

. Definitive re-irradiation using intensity-modulated radiation therapy in cancers of the head and neck, focusing on rare tumors. Acta Otolaryngol. 2018; 138(8): 750-758. doi:10.1080/00016489.2018.1438662.

21.

Sun

Chen

, et al. Long-term outcomes of intensity-modulated radiotherapy for 868 patients with nasopharyngeal carcinoma: an analysis of survival and treatment toxicities. Radiother Oncol. 2014; 110(3): 398-403. doi:10.1016/j.radonc.2013.10.020.

22.

Bucci

Bevan

Roach

3rd . Advances in radiation therapy: Conventional to 3D, to IMRT, to 4D, and beyond. CA Cancer J Clin. 2005; 55(2): 117-134. doi: 10.3322/canjclin.55.2.117.

23.

Fedorov

Beichel

Kalpathy-Cramer

, et al. 3D slicer as an image computing platform for the quantitative imaging network. Magn Reson Imaging. 2012; 30(9): 1323-1341. doi: 10.1016/j.mri.2012.05.001.

24.

Zhang

Huang

Zhao

, et al. Development and validation of magnetic resonance imaging-based radiomics models for preoperative prediction of microsatellite instability in rectal cancer. Ann Transl Med. 2021; 9(2): 134. doi:10.21037/atm-20-7673.

25.

Tan

Xie

, et al. Development and validation of a preoperative magnetic resonance imaging radiomics-based signature to predict axillary lymph node metastasis and disease-free survival in patients with early-stage breast cancer. JAMA Netw Open. 2020; 3(12): e2028086. doi: 10.1001/jamanetworkopen.2020.28086.

26.

Xie

Wang

Tong

Zhou

. Pancreatic ductal adenocarcinoma: A radiomics nomogram outperforms clinical model and TNM staging for survival estimation after curative resection. Eur Radiol. 2020; 30(5): 2513-2524. doi:10.1007/s00330-019-06600-2.

27.

Peng

Zhang

Zhou

Liu

. A radiomics nomogram for preoperative prediction of microvascular invasion risk in hepatitis B virus-related hepatocellular carcinoma. Diagn Interv Radiol. 2018; 24(3): 121-127. doi:10.5152/dir.2018.17467.

28.

Tibshirani

. The lasso method for variable selection in the cox model. Stat Med. 1997; 16(4): 385-395.doi: 10.1002/(sici)1097-0258(19970228)16:4<385::aid-sim380>3.0.co.

29.

Gui

. Penalized cox regression analysis in the high-dimensional and low-sample size settings, with applications to microarray gene expression data. Bioinformatics. 2005; 21: 3001-3008. doi: 10.1093/bioinformatics/bti422.

30.

Zhong

Fang

Dong

, et al. A deep learning MR-based radiomic nomogram may predict survival for nasopharyngeal carcinoma patients with stage T3N1M0. Radiother Oncol. 2020; 151: 1-9. doi: 10.1016/j.radonc.2020.06.050.

31.

Zhang

Zhou

, et al. Radiomic nomogram: Pretreatment evaluation of local recurrence in nasopharyngeal carcinoma based on MR imaging. J Cancer. 2019; 10(18): 4217-4225. doi: 10.7150/jca.33345.

32.

Spadarella

Calareso

Garanzini

Ugga

Cuocolo

. MRI based radiomics in nasopharyngeal cancer: Systematic review and perspectives using radiomic quality score (RQS) assessment. Eur J Radiol. 2021; 140: 109744. doi: 10.1016/j.ejrad.2021.109744.

33.

Peng

Dong

Fang

, et al. Prognostic value of deep learning PET/CT-based radiomics: Potential role for future individual induction chemotherapy in advanced nasopharyngeal carcinoma. Clin Cancer Res. 2019; 25(14): 4271-4279. doi: 10.1158/1078-0432.CCR-18-3065.

34.

Yang

Elhalawani

Shi

, et al. A large-scale retrospective study of the overall survival outcome in nasopharyngeal carcinoma with hypertension in Chinese population. Oncotarget. 2017; 8(43): 75577-75586. doi: 10.18632/oncotarget.17483.

35.

Feng

Liang

Wang

, et al. Radiomics analysis and correlation with metabolic parameters in nasopharyngeal carcinoma based on PET/MR imaging. Front Oncol. 2020; 10: 1619. doi: 10.3389/fonc.2020.01619.

Development of a Radiotherapy Localisation Computed Tomography-Based Radiomic Model for Predicting Survival in Patients With Nasopharyngeal Carcinoma Treated With Intensity-Modulated Radiotherapy Following Induction Chemotherapy

Abstract

Background

Methods

Results

Conclusion

Keywords

Introduction

Material and Methods

Patients

IMRT

Clinical Endpoints and Follow-Up

ROI Segmentation

Radiomic Analysis

Image Acquisition

Feature extraction

Statistical Analyses

Results

Discussion

Conclusion

Footnotes

Declarations of Conflicting Interest

Funding

Ethical Approval

ORCID iDs

References