Abstract
Background
Benefits and harms of screening mammography have been disputed in recent years. This fact, along with the limitations of mammography as well as its unavailability in all our medical centers, tempted us to evaluate the accuracy of thermography in detecting breast abnormalities.
Patients and Methods
All patients who were candidates for breast biopsy were examined by both mammography and thermography before tissue sampling in a referral center between January 2013 and January 2014. We defined sensitivities and specificities, and positive predictive values (PPVs) and negative predictive values (NPVs), of the 2 modalities in comparison with histologic results as the gold standard.
Results
132 patients were included. The median age of all patients was 49.5 ± 10.3 years (range 24-75 years). The sensitivity, specificity, PPV, NPV, and accuracy for mammography were 80.5%, 73.3%, 85.4%, 66.0%, and 76.9%, respectively, whereas for thermography the figures were 81.6%, 57.8%, 78.9%, 61.9%, and 69.7%, respectively.
Conclusion
Our study confirms that, at the present time, thermography cannot substitute for mammography for the early diagnosis of breast cancer.
Key Words: Breast cancer, Infrared, Mammography, Neoplasm, Thermography
Introduction
Nowadays mammography is used as the gold standard for breast cancer screening and diagnosis; this modality can demonstrate masses or microcalcifications in the earliest stages of the disease. Nevertheless, mammography has a low sensitivity in young women and in dense breasts [1]; also, its facility requires high-quality, expensive equipment and entails high costs [2].
A non-invasive means of breast imaging, thermography was reported by Lawson in 1957 as a new tool in the investigation of breast lesions [3]. This entity was considered rather extensively in the 1970s and 1980s [4], and was approved by the US Food and Drug Administration (FDA) as an adjunctive tool for the diagnosis of breast cancer, but not as a useful screening or diagnostic tool for breast malignancy, due to its low sensitivity and specificity [5,6]. However, since advanced digital equipment and recent advancements in technology and image processing methods have led to superior detection of heat waves, the current study focuses again on the subject, in order to investigate the potentially improved detection of breast pathology by digital thermographs. Considering these facts, we conducted a study to determine the sensitivity and specificity of thermography, in comparison with mammography, in differentiating benign and malignant disease in breast lesions undergoing biopsy and histological review.
Methods
This prospective study was undertaken in the breast clinic of the Cancer Institute of Tehran, Iran, between January 2013 and January 2014 among women attending the clinic with any breast-related complaint or just for opportunistic screening. Inclusion criteria consisted of any clinical, mammographic, or ultrasonographic finding in need of tissue histological exam. Exclusion criteria consisted of age under 35 years for cases harboring benign-appearing lesions, because mammography was not necessary in these cases. The study was approved by the ethics committee of the Cancer Institute of Tehran on September 2012.
First, a breast exam was performed for all patients by 1 of 2 breast surgeons in the clinic, and then all included cases underwent both mammography and thermography after obtaining informed consent. The mammography apparatus in this study was a Hologic fully digital direct mammography system. Digital mammography images were interpreted by 2 expert radiologists blinded to the thermography results; they were assigned numbers from B1 to B5 according to the Breast Imaging Report and Data System (BI-RADS) of the American College of Radiology staging system. Breast thermography was done by a trained certified medical thermographer, using a medical thermography system (FMG-MED IR; Fanavaran Madoon Ghermez Company, Tehran, Iran), which was a non-cooled microbolometer with a focal plane array detector. The image matrix size was 640 × 480, the pixel size 25 × 25 µm, with a response wavelength of 8-12 µm and a temperature resolution of 0.08 °C. Images were viewed with dedicated software with manual brightness and contrast adjustment and were displayed with either a gray palette scale or a preset color palette. Digital infrared thermal imaging (DITI) was carried out at 21 °C room temperature with the patient disrobed from the waist up (fig. 1). The overall timing for each thermography scan was 20 min including an ice test. In order to perform the ice test, the hands of the patients were placed in ice and cold water for 45-60 s after the first round of thermography (after taking the first 5 thermal images). Then, the next 5 thermal images were taken. The ice test could enhance the sensitivity and specificity of the thermography results because of the small temperature changes before and after ice immersion, which may modify the thermography levels (TH, see below) and show details of the suspicious lesions as hyperthermic areas. This test is more important in the dense breast.
Fig. 1.
Digital infrared thermal imaging (DITI) scan in a 48-year-old woman with infiltrating ductal carcinoma in the upper outer quadrant of the left breast.
Results of breast thermography were categorized by the Marseille system, a risk assessment system demonstrated in table 1. This analytic method provides for a TH1-TH5 scale as a summary based on specific, objective, and quantitative thermal features, and differential levels of infrared energy. Further assessment and screening are recommended when images are classified from TH3 to TH5.
Table 1.
The Marseille risk assessment system for breast thermography results
| TH1 | no unusual features, normal breast tissue |
| TH2 | area(s) of increase in heat, responsive to cold challenge |
| TH3 | area(s) of atypical increase in heat, unresponsive to cold challenge |
| TH4 | area(s) of abnormal increase in heat, unresponsive to cold challenge |
| TH5 | area(s) of severely abnormal increase in heat, unresponsive to cold challenge |
Subsequent to breast imaging, tissue sampling was performed in all participants via core needle biopsy or surgery. Histologic results were considered as the gold standard test. We defined the sensitivities and specificities of mammography and thermography, and of thermography after applying the ice test, in comparison with the gold standard. We also defined a test package, consisting of hybridization of the 2 tests (mammography and thermography); this package was also compared with the gold standard test. Besides, we used Cohen's kappa coefficient to calculate the agreement between the mammography and thermography (with and without ice test) interpretations, and between each of the two tests as well as the hybrid test and the histologic results.
Results
Overall, 132 patients were included in the study. The median age of all patients was 49.5 ± 10.3 years (range 24-75 years). The personal and family history of cancer, menopausal status, side and size of tumors, and histologic result are demonstrated in table 2. The final pathologic result consisted of 45 benign lesions and 87 malignant ones, including 79 cases of invasive ductal carcinoma and 8 cases of invasive lobular carcinoma.
Table 2.
Patient and tumor characteristics
| Variable | n (%) |
|---|---|
| Menopausal status | |
| Premenopause | 52 (39.4) |
| Perimenopause | 9 (6.8) |
| Postmenopause | 57 (43.2) |
| Hysterectomized | 6 (4.5) |
| Family history of breast cancer | |
| Positive | 61 (46.2) |
| Negative | 71 (53.8) |
| History of breast cancer | |
| Positive | 4 (3) |
| Negative | 128 (97) |
| Tumor side | |
| Right breast | 57 (43.2) |
| Left breast | 75 (56.8) |
| Tumor size | |
| < 2cm | 27 (20.5) |
| 2–5 cm | 51 (38.6) |
| > 5 cm | 1 (0.8) |
| Unknown | 53 (40.2)a |
| Tumor pathology | |
| Benign | 45 (34.1) |
| Malignant | 87 (65.9) |
| Mammographically suspicious sign | |
| Spiculated mass | 52 (59.7) |
| Microcalcification | 23 (26.4) |
| Paranchymal distortion | 12 (13.7) |
Operated elsewhere, so exact pathologic size not available.
The numbers of histologically benign and malignant cases in each class of mammography (BI-RADS) and thermography (TH) as well as thermography after applying the ice test are demonstrated in table 3. We divided the mammography and thermography results into 2 broader categories as given below:
Table 3.
Number of benign and malignant histologies for each score in mammography and thermography
| Pathology | Mammography |
||||||
|---|---|---|---|---|---|---|---|
| B0 | B1 | B2 | B3 | B4 | B5 | Total | |
| Malignant | 10 (41.7%) | 1 (25%) | 1 (12.5%) | 6 (40%) | 43 (79.6%) | 26 (96.3%) | 85 |
| Benign | 14 (58.3%) | 3 (75%) | 7 (87.5%) | 9 (64.3%) | 11 (20.4%) | 1 (3.7%) | 45 |
| Pathology | Thermography |
||||||
| TH1 | TH2 | TH3 | TH4 | TH5 | Total | ||
| Malignant | 6 (40%) | 10 (37%) | 30 (76.9%) | 30 (88.2%) | 11 (64.7%) | 87 | |
| Benign | 9 (60%) | 17 (63%) | 9 (23.1%) | 4 (11.8%) | 6 (35.3%) | 45 | |
B0 = Inconclusive, B1 = normal, B2 = benign, B3 = probably benign, B4 = suspicious, B5 = highly suspicious, TH1–2 negative, TH3–5 = positive.
Mammography: B0 = inconclusive, B1-3 = negative, B4-6 = positive
Thermography: TH1-2 = negative, TH3-5 = positive
Category BI-RADS 0 (B0) needs additional viewing or additional imaging like ultrasonography for further evaluation. We consider only B0 cases if the last BI-RADS assessment was defined by an additional mammographic view.
The numbers of the results in each group and the hybrid imaging package in patients with benign and malignant tumors are demonstrated in table 4. The sensitivities and specificities of the 2 imaging modalities alone and in combination are shown in table 5. The accuracy of thermography was obviously lower than that of mammography (69.7% vs. 76.9%). The sensitivity of the combined imaging package was higher than that of each imaging (96.2% vs. 80.5% and 81.6%), but the specificity and accuracy became lower.
Table 4.
Numbers of positive and negative results in mammography, thermography, and package in histology groups
| Histology |
||
|---|---|---|
| Malignant | Benign | |
| Mammography | ||
| + (B4,5) | 70 | 12 |
| – (B1–3) | 17 | 33 |
| Thermography | ||
| + (TH3–5) | 71 | 19 |
| – (TH1,2) | 16 | 26 |
| Thermography + ice test | ||
| + (TH3–5) | 77 | 25 |
| – (TH1,2) | 6 | 18 |
| Imaging packagea | ||
| B4,5 and TH3–5 | 84 | 25 |
| B1–3 and TH1,2 | 3 | 20 |
Combination of mammography and thermography.
Table 5.
Diagnostic value indices of mammography, thermography, and the imaging package
| Sensitivity, % | Specificity, % | PPV, % | NPV, % | Accuracy, % | |
|---|---|---|---|---|---|
| Mammography | 80.5 | 73.3 | 85.4 | 66.0 | 76.9 |
| Thermography | 81.6 | 57.8 | 78.9 | 61.9 | 69.7 |
| Thermography + ice test | 92.8 | 41.9 | 75.5 | 75 | 67.35 |
| Imaging packagea | 96.2 | 44.4 | 77.1 | 87.0 | 70.5 |
PPV = Positive predictive value, NPV = negative predictive value.
Combination of mammography and thermography.
In order to evaluate the agreement coefficient of different imaging modalities in differentiating benign from malignant masses, we obtained Cohen's kappa values, as demonstrated in table 6. This value was 0.52 for mammography (p-value 0.000) and was 0.209 for thermography (p-value 0.005), which means that mammography was more powerful than thermography for the differentiation of benign and malignant lesions.
Table 6.
Agreement between diagnostic modalities according to Cohen's kappa values
| Mammography (p), agreement | Histology (p), agreement | |
|---|---|---|
| Mammography | – | 0.52 (< 0.001), moderate |
| Thermography | 0.22 (0.009), weak | 0.40 (< 0.001), moderate |
| Thermography + ice test | 0.388 (< 0.001), moderate | – |
| Imaging package | – | 0.47 (< 0.001), moderate |
Discussion
It is anticipated that the increased metabolic rate and new angiogenesis in tumors cause some increase in temperature of the area, which could be detected by thermography [7]. In addition, rising nitric oxide levels produced by proliferating malignant cells result in local vasodilation, which subsequently leads to heat emission from that spot [8].
Studies conducted in the past were not entirely in favor of this theory; in 1975, the American College of Radiology and the American Thermographic Society defined thermography as only ‘a complementary diagnostic tool that may be useful in evaluation of breast disease when it is combined with both physical examination under the supervision of a qualified physician, and mammography by a trained radiologist’ [9]. In 1976, Moskowitz et al. [6] detected a low rate of true positives for early-stage breast cancer in thermograms interpreted by expert specialists, and concluded that this modality is probably not suitable for cancer screening. In more recent years, animal models have shown an early detection of breast malignancy by thermography before any mammographic change [10,11,12].
The present prospective study, performed on 132 women undergoing thermography and mammography before biopsy, showed a low sensitivity (47.1%) for thermography compared to mammography (80.5%) when only TH4-5 were considered as positive. In one of the earliest studies carried out by Gautherie and Gros in 1980 [13], patients with a thermogram stage 4 or 5 had a 90% chance of being affected by cancer at the time of study, while 38% of the TH3 cases (suspicious but not conclusive) developed cancer within 1-4 years of follow-up. Similarly, in a recent review by Lahiri et al. [14], an abnormal pattern in infrared images could identify a high risk for breast cancer development in the future. In our study, by considering stage TH3 as positive, and by using the ice test in the majority of patients, the sensitivity of thermography became higher (81.6% and 92.8%).
Similar studies in the literature have shown more favorable results. In a multicentric study by Parisky et al. [15], 769 patients were imaged using a DITI prior to breast biopsy; a total of 875 biopsies were performed and the sensitivity of DITI was 97%, with a specificity of 14%.
In the study by Aora et al. [7], DITI identified 58 of 60 malignancies, with 97% sensitivity and 44% specificity, and 82% negative predictive value (NPV). In another work, Wishart et al. [16] used 4 different ways for interpreting the scans and found DITI to be effective in women under 50 years with a high sensitivity (78%) and specificity (75%), by exerting an expert manual review. They concluded that reduced vascularity in breasts of older women, especially over 70 years of age, may account for the poor result of thermography in this group of patients.
Yao et al. [17] found that the sensitivity and specificity of infrared thermography was superior to that of mammography and ultrasonography in lesion less than 2 cm in diameter, and that mammography had a better diagnostic accuracy only in lesion larger than 2 cm in diameter. In a study on chemically induced rat mammary tumors, the maximal tumor temperature and the thermal amplitude determined by thermography were significantly correlated with the tumor volume [18]. Other studies on breast cancer cases had shown a direct correlation between infrared thermography stage and tumor size, grade, and vascularity, as well as lymph node involvement [19,20]. In another study by Zore et al. [21], the tumor size had no influence on the increased temperature. The increased temperature in that study was more dependent on immunohistochemistry phenotypes of the tumor than on other clinically and histopathologically prognostic factors.
The results of a literature review regarding the effectiveness of DITI in an imaging program are controversial [22,23,24]. A systematic search of 7 biomedical databases by Vreugdenburg et al. [23] during March 2011 found a wide variation in sensitivity (0.25-0.97) and specificity (0.12-0.85) of DITI and concluded that there is insufficient evidence to recommend the use of thermography for breast cancer screening. As well, a review by Sajadi et al. [24] did not show any acceptable diagnostic value for DITI in comparison with other diagnostic modalities.
In the current study, the accuracy of DITI was lower than that of mammography (69.7% vs. 76.9%), but the study has some limitations: The sample size is small, and all mass or mass-like densities were examined. In addition, cases fairly inadequate for DITI, such as elderly or morbid obese women, advanced stages of cancer, and large breast sizes, were included in the study.
We believe that DITI is a non-invasive, inexpensive and accessible imaging modality in our country, but its real role in clinical practice could only be evaluated through a large multicenter trial that would estimate the accuracy of digital thermography in breast cancer screening. At present, our suggestion is to perform thermography as a complementary test to a breast clinical exam. In conclusion, it seems that, despite technical advances in thermography, it cannot substitute for mammography at the present time; this modality can only be proposed as a complementary tool in breast cancer diagnosis.
Disclosure Statement
All the authors formally declared no conflict of interest with ongoing research. This work has not been supported by any company or organization.
References
- 1.Garcia-Manso A, Garcia-Orellana CJ, Gonzalez-Velasco HM, Gallardo-Caballero R, Macias-Macias M. Study of the effect of breast tissue density on detection of masses in mammograms. Comput Math Methods Med. 2013;2013:213794. doi: 10.1155/2013/213794. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Artz DR, Brown ML, Barrett MJ. The supply of mammography resources in West Virginia. W V Med J. 1992;88:142–146. [PubMed] [Google Scholar]
- 3.Lawson RN. Thermography - a new tool in the investigation of breast lesions. Can Serv Med J. 1957;13:517–524. [PubMed] [Google Scholar]
- 4.Gohagan JK, Rodes ND, Blackwell CW, Darby WP, Farrell C, Herder T, Pearson DK, Spitznagel EL, Wallace MD. Individual and combined effectiveness of palpation, thermography, and mammography in breast cancer screening. Prev Med. 1980;9:713–721. doi: 10.1016/0091-7435(80)90016-x. [DOI] [PubMed] [Google Scholar]
- 5.Threatt B, Norbeck JM, Ullman NS, Kummer R, Roselle PF. Thermography and breast cancer an analysis of a blind reading. Ann N Y Acad Sci. 1980;335:501–527. doi: 10.1111/j.1749-6632.1980.tb50775.x. [DOI] [PubMed] [Google Scholar]
- 6.Moskowitz M, Milbrath J, Gartside P, Zermeno A, Mandel D. Lack of efficacy of thermography as a screening tool for minimal and stage I breast cancer. N Engl J Med. 1976;295:249–252. doi: 10.1056/NEJM197607292950504. [DOI] [PubMed] [Google Scholar]
- 7.Aora N, Martins D, Ruggerio D, Tousimis E, Swistel AJ, Osborne MP, Simmons RM. Effectiveness of a noninvasive digital infrared thermal imaging system in the detection of breast cancer. Am J Surg. 2008;196:523–526. doi: 10.1016/j.amjsurg.2008.06.015. [DOI] [PubMed] [Google Scholar]
- 8.Lee MY, Yang CS. Entropy-based feature extraction and decision tree induction for breast cancer diagnosis with standardized thermograph images. Comput Methods Programs Biomed. 2010;100:269–282. doi: 10.1016/j.cmpb.2010.04.014. [DOI] [PubMed] [Google Scholar]
- 9.Strax P. Health Control in Detection of Cancer. Stockholm: Almqvist & Wiksell International; 1976. Benefit of breast cancer screening on morbidity and mortality; in Skandia International Symposia; pp. 133–140. [Google Scholar]
- 10.Poljak-Blazi M, et al. Specific thermographic changes during Walker 256 carcinoma developments: differential infrared imaging of tumour, inflammation and haematoma. Cancer Detect Prev. 2009;32:431–436. doi: 10.1016/j.cdp.2009.01.002. [DOI] [PubMed] [Google Scholar]
- 11.Song C, Appleyard V, Murray K, Frank T, Sibbett W, Cuschieri A, Thompson A. Thermographic assessment of tumor growth in mouse xenografts. Int J Cancer. 2007;121:1055–1058. doi: 10.1002/ijc.22808. [DOI] [PubMed] [Google Scholar]
- 12.Xie W, McCahon P, Jakobsen K, Parish C. Evaluation of the ability of digital infrared imaging to detect vascular changes in experimental animal tumours. Int J Cancer. 2004;108:790–794. doi: 10.1002/ijc.11618. [DOI] [PubMed] [Google Scholar]
- 13.Gautherie M, Gros CM. Breast thermography and cancer risk prediction. Cancer. 1980;45:51–56. doi: 10.1002/1097-0142(19800101)45:1<51::aid-cncr2820450110>3.0.co;2-l. [DOI] [PubMed] [Google Scholar]
- 14.Lahiri B, Bagavathiappan S, Jayakumar T, et al. Medical applications of infrared thermography: a review. Infrared Phys Technol. 2012;55:221–235. doi: 10.1016/j.infrared.2012.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Parisky YR, Sadri A, Hamm R, Hughes K, Esserman L, Rust S, Callahan K. Efficacy of computerized infrared imaging analysis to evaluate mammographically suspicious lesions. AJR Am J Roentgenol. 2003;180:263–269. doi: 10.2214/ajr.180.1.1800263. [DOI] [PubMed] [Google Scholar]
- 16.Wishart GC, Campisi M, Boswell M, Chapman D, Shackleton V, Iddles S, Hallett A, Britton PD. The accuracy of digital infrared imaging for breast cancer detection in women undergoing breast biopsy. Eur J Surg Oncol. 2010;36:535–540. doi: 10.1016/j.ejso.2010.04.003. [DOI] [PubMed] [Google Scholar]
- 17.Yao X, Wei W, Li J, Wang L, Xu Z, Wan Y, Li K, Sun S. A comparison of mammography, ultrasonography, and far-infrared thermography with pathological results in screening and early diagnosis of breast cancer. Asian Biomed. 2014;8:11–19. [Google Scholar]
- 18.Faustino-Rocha AI, Silva A, Gabriel J, Teixeira-Guedes CI, Lopes C, Gil da Costa R, Gama A, Ferrreira R, Oliveira PA, Ginja M. Ultrasonographic, thermographic and histologic evaluation of MNU-induced mammary tumor in female Sprague-Dawley rats. Biomed Pharmacother. 2013;67:771–776. doi: 10.1016/j.biopha.2013.06.011. [DOI] [PubMed] [Google Scholar]
- 19.Sterns EE, Zee B, SenGupta S, Saunders FW. Thermography. Its relation to pathologic characteristic, vascularity, proliferation rate, and survival of patients with invasive ductal carcinoma of the breast. Cancer. 1996;77:1324–1328. doi: 10.1002/(SICI)1097-0142(19960401)77:7<1324::AID-CNCR15>3.0.CO;2-3. [DOI] [PubMed] [Google Scholar]
- 20.Head JF, Elliott RL. Thermography. Its relation to pathologic characteristic, vascularity, proliferation rate, and survival of patients with invasive ductal carcinoma of the breast. Cancer. 1997;79:186–188. doi: 10.1002/(sici)1097-0142(19970101)79:1<186::aid-cncr28>3.0.co;2-#. [DOI] [PubMed] [Google Scholar]
- 21.Zore Z, Filipovic-Zore I, Stanec M, Batinjan G, Matejcic A. Association of clinical, histopathological and immunohistochemical prognostic factor of invasive breast tumors and thermographic findings. Infrared Phys Technol. 2015;68:201–205. [Google Scholar]
- 22.Etehadtavakoli M, Ng EYK. Breast thermography as a potential non-contact method in the early detection of cancer: a review. J Mech Med Biol. 2013;13:330001. [Google Scholar]
- 23.Vreugdenburg TD, Willis CD, Mundy L, Hiller JE. A systemic review of elastography, electrical impedance scanning, and digital thermography for breast cancer screening and diagnosis (review) Breast Cancer Res Treat. 2013;137:665–676. doi: 10.1007/s10549-012-2393-x. [DOI] [PubMed] [Google Scholar]
- 24.Sajadi HS, Seyedin H, Aryankhesal A. A systematic review on the effectiveness of thermography in diagnosis of disease (review) Int J Imaging Syst Technol. 2013;23:188–193. [Google Scholar]