Abstract
BACKGROUND:
Thermographic imaging is a non-invasive and radiation free imaging modality that measures the infrared radiation released by the body. Recently, there is a renewed interest regarding the scope of thermal imaging for breast cancer.
OBJECTIVE:
To evaluate the efficacy of thermographic breast imaging in detecting breast cancer.
METHODS:
A Prospective observational study was carried out from January 2014 to December 2014 at Kasturba Hospital, Manipal, India. Patients in whom breast cancer was confirmed on FNAC or biopsy, were included in the study and further evaluated with thermographic imaging of the breast.
RESULTS:
65 patients with FNAC or biopsy proven breast carcinoma were included in the study. Using thermographic imaging, malignancy was accurately detected in 60 patients (92.31%). Mammography was able to detect malignancy in 62 out of the 65 patients (95.38%). Thermography was able to detect malignancy in all 3 cases in which conventional mammography missed it.
CONCLUSION:
Thermography may have a role in detection of breast cancer. However, it is too early to recommend thermographic imaging as a standard imaging modality for breast cancer. Larger studies are required to evaluate the usefulness of thermography in diagnosis and/or screening of breast cancer.
Introduction
Over the years, mammography has become the gold standard imaging modality for breast lesions [1]. Ultrasound and MRI have emerged as alternative imaging modalities for detecting tumours in the breast. In the context of breast cancer, mammogram has both screening and diagnostic roles. Screening for breast cancer using mammography has resulted in a considerable decrease in breast cancer mortality [2–5].
However, mammography is not without its own disadvantages. These include- need of expensive equipment, exposure to radiation, missing of interval cancers (between 2 screening rounds) and overdiagnosis [6,7]. Further, mammography cannot be accurately used for radiologically firm breasts and in women with breast implants.
Thermographic imaging is a non-invasive and radiation free imaging modality that measures the infrared radiation released by the body [3]. Such infrared radiation is directly related to the vascularity and metabolic activity of the tissue being imaged [4]. Most cancers derive their own blood supply by facilitating angiogenesis. In addition, cancerous tissues also have higher metabolic activity compared to normal tissues. Therefore cancerous tissue is likely to emit more infrared radiation as compared to surrounding normal tissue. In principle, thermographic imaging aims to capture and demonstrate such excessive infrared rays emitted by cancerous tissue.
The early interest and promises shown regarding thermographic imaging of breast cancer waned over a period of time due to lack of sensitivity and specificity of older thermal cameras [8]. However, with the evolution of newer, higher resolution infrared cameras, which provide excellent thermal images, there is a renewed interest regarding the scope of thermal imaging for breast cancer. Thermographic images are easier to interpret than the conventional imaging modalities for breast cancer. Also, infrared cameras used for the same are cheaper and more portable when compared to the devices used in mammography, MRI and ultrasound [9].
In this preliminary study, we have attempted to evaluate the efficacy of thermographic breast imaging in identifying breast cancer.
Aim
To detect the accuracy of thermographic imaging in detecting breast cancer in biopsy/FNAC (Fine needle aspiration cytology) proven cases of breast cancer.
Materials and methods
This study was conducted in Kasturba Hospital, Manipal, which is a tertiary referral centre, attached to a medical college. Ethical committee clearance was obtained as per standard protocol. Written informed consent was obtained from all patients before enrollment in the study.
Study methodology:
Patients aged more than 18 years, presenting to Kasturba Hospital with breast related symptoms from January 2014 to December 2014, were evaluated with mammography. As an institutional protocol, most of these patients were also evaluated with an ultrasound scan of the breast (sonomammogram). The mammogram and the ultrasound were performed and interpreted by trained radiologists. Depending upon the findings, some of these patients were further evaluated by either FNAC (Fine needle aspiration cytology) or core needle biopsy. Patients in whom breast cancer was confirmed on FNAC or biopsy, were included in our study and further evaluated with thermal imaging of the breast. The thermal imaging was performed at least one week after the FNAC/core needle biopsy to decrease artefacts in the imaging caused by hematoma and/or inflammatory reaction.
Inclusion criteria:
All adult patients >18 years, confirmed to have breast cancer, by histo/cytopathological evaluation. Such patients should have undergone both mammographic (xeromammography, with or without sonomammography) evaluation and thermal imaging.
Exclusion criteria:
Patients with associated inflammatory conditions in breast (e.g. tumour necrosis, abscess etc.).
Technical aspects of thermographic imaging of breast
All the patients included in the study were subjected to thermographic video imaging in identical environment. Patient was seated on a revolving chair with hands raised over the shoulder and thermographic video imaging was recorded with the static camera fixed over tripod. The chair was rotated gently 180 degrees so that the lateral, anterior and medial aspects of the breast were comprehensively imaged. The camera settings were fixed throughout the study. Two thermal cameras (FLIR-E60 and T650sc) were used in our study. FLIR E60 is a low resolution thermal video camera (320 × 240 pixels) and FLIR T650sc provides higher resolution video images (640 × 480 pixels). Out of the 65 subjects 31 were imaged with low resolution camera. Radiographers provided the mammography and sonomammography reports. The clinical case sheet containing patient complaints and a comprehensive clinical examination by the breast clinician was provided to the team as additional data. Thermal videos were captured by a clinical nurse. Five medical and engineering experts (all five authors) met on a routine basis for evaluating the medical data. Since no reports were generated by the radiologists with thermal videos, experts followed a rigorous process to look for suspicious abnormalities in each thermal video manually. Isotherm visualization tool (without the automated detection feature) was used heavily to assist during this manual detection process. Isotherm map extracted temperature values of pixels within the region covered by the isotherm contours. Figs 1 and 2 show typical appearance of breast cancer on thermogram and Figs 3 and 4 show appearance of cancer with 1 ∘ C and 0.5 ∘ C isotherm contours respectively. Figure 1 is the thermogram of a 48 years old female, having BIRADS 4c lesion in upper outer quadrant of right breast as per mammogram. The thermogram matches with clinical findings, mammogram and sonomammogram findings. Linear hotlines are probably due to neovascularization. In Fig. 2, we have the thermogram of a 58 year old lady with BIRADS 5 lesion in the sub-areolar region of right breast. The tool also provided various isothermic views at different viewing angles and additional functionalities for hotspot selection such as cropping, zooming, etc., which helped in the visual assessment of thermal abnormalities. Figure 3 shows the isotherm contour map (1 degree C) for subject shown in Fig. 1. There is a clear lack of symmetry in both the raw image and the contour map, between similar regions of the right and left breast, with one hot spot evident in the upper outer quadrant of right breast.
Clinical finding, mammogram, sonomammogram, FNAC/Biopsy reports were reviewed, interpreted and compared to thermal results only after distilling summaries from each thermogram video. Results documented from thermal videos were not influenced by data from standard medical practice such as mammogram, sonomammogram, FNAC/biopsy or clinical examination reports.
Results
65 patients with FNAC or biopsy proven breast carcinoma were included in the study. All 65 patients were subjected to thermographic imaging and conventional mammography for both breasts. Correlated sonomammogram (ultrasonography) of both breasts was done along with conventional mammogram to help in diagnosis in 62 of the patients.
Mammogram was classified as having correctly detected malignancy if it was reported by the radiologist as BIRADS ≥ IV.
The average age of the patients included in the study was 50.8 years (Standard deviation -9.46 years).
Figure 5 compares the accuracy of thermography with mammography for detecting breast cancer. Using thermographic imaging, malignancy was accurately detected in 60 patients (92.31%). Mammography was able to detect malignancy in 62 out of the 65 patients (95.38%).
Thermography was able to detect malignancy in all 3 cases in which conventional mammography missed it. The average age of these 3 patients was 39.67 years.
Out of the 5 cases where thermography failed to detect malignancy, mammography and sonomammography were able to pick up the cancer in all 5 patients. The average age of these 5 patients was 51 years.
Table 1 shows the comparison between thermography and mammogram in detecting tumours with various T-stages. If both thermography and mammography are considered together, the accuracy of detecting breast cancer increases to 100%. In 5 patients, thermography falsely detected a malignant lesion in the opposite breast. Whereas mammography was false positive in the opposite breast in 2 patients.
Discussion
Attempts have been made since the 1960s to use thermographic imaging of the breast as a tool to detect breast cancer.
Thermal imaging of the breast involves detecting lesions in the breast which are hotter than the surrounding normal breast tissue. Those lesions which have increased blood supply due to angiogenesis are more likely to be hotter, and these will be detected by the thermal camera as a ‘hotspot’. This concept is appealing as there is no exposure to harmful radiation. Furthermore, a conventional mammogram diagnoses breast cancer by identifying abnormal calcifications or breast architectural distortions, whereas thermogram relies on vascularity to identify breast cancer.
However, in spite of promising conceptualization, the modality fell to disrepute as the earlier thermographic imaging lacked sensitivity and specificity [8,10]. Over the years the quality of thermal cameras have improved and several semi-automated algorithms have been developed for tumour detection. In the past few years there has been renewed interest in thermography in breast cancer. Attempts to evaluate the accuracy of thermography for screening and diagnosis of breast cancer have yielded mixed results [11–16].
In this study we have attempted to evaluate the relevance of thermography in comparison to mammography for detecting breast cancer.
Our results show that thermographic imaging of breast has an accuracy of 92.31% in FNAC/biopsy proven cases of breast cancer. This is comparable to that of conventional mammography. In this study, when we combine thermography with mammography, the accuracy increases to 100%.
The average age of patients in which mammogram failed to detect the lesion was lower than the overall average age (39.67 versus 50.8 years). Although the numbers are small, our results reiterate the fact that mammography is less accurate in younger patients with denser breast tissue. In these patients, thermography was found to be a more valuable tool for detecting breast cancer.
Further, thermographic imaging of the contralateral breasts in the subjects did not show any false positive results. However, calculating false positivity was not one of the aims of this study. This was a pilot study aimed at determining whether thermography can actually detect a carcinoma of the breast. We will be conducting a larger scale study in which subjects with normal breasts will also be included in order to ascertain the false positivity and to determine its role as a screening tool for breast cancer.
One limitation of thermography is that regions such as axilla and inframammary folds may produce misreading due to the higher surface temperatures in these regions.
Since all the patients had undergone a prior FNA or a core needle biopsy, the findings in the thermographic imaging may have been distorted by inflammatory reaction and/or hematoma formation. In order to minimize these artefacts, thermography was performed a minimum of one week after the FNA or core needle biopsy. However, this remains an important drawback of the study.
Thermographic imaging is not performed routinely at our centre. The imaging was done free of cost for all the participants in the study. The actual cost of the imaging in a diagnostic/screening setup has not been calculated. Owing to the significantly lower cost of equipment and ease of taking the images, the cost is expected to be much lower than that of a conventional mammogram.
Conclusion
Our results suggest that thermography may have a role in detection of breast cancer. In a country like India, where the infrastructure for mammography is available only in select tertiary referral centres, thermography may serve as a useful tool at primary health centres. Even when facilities for mammogram are available, thermography may be used as an adjunct imaging modality to provide information for improving diagnostic accuracy.
However, it is still a very early stage to recommend thermographic imaging as a standard imaging modality in breast cancer. Further studies are needed, involving large number of subjects of both benign and malignant breast diseases, and in screening setups, to validate the role of thermal imaging in breast diseases.