Table 1 Studies retrieved from a systematic review of literature on breast thermography published in the twenty-first century
From: Breast thermography: a systematic review and meta-analysis
Author, year, Ref | Sample size (n cancer) | Patient description | Protocol | Sensitivity specificity | Thermal resolution | Type of interpretation | Highlights and observations |
|---|---|---|---|---|---|---|---|
Parisky et al., 2003 [53] | 875 lesions in 769 patients (187) | Patients undergoing breast biopsy because of mammographically suspicious lesions | Dynamic | 97.37% 14.41% | Unknown | Proprietary software | Funded by Computerized Thermal Imaging (manufacturer) Thermogram interpretation required previous manual localization of the suspicious region Images were acquired with the patient lying prone with the breasts suspended through openings in the imaging bed |
Button et al., 2004 [54] | 29 (8) | Patients scheduled for breast biopsy | Unknown | 81.25% 53.66% | Unknown | Visual | Funded by Omnicorder Technologies Inc. (manufacturer) Images were acquired with the patient lying prone with the breasts suspended through openings in the imaging bed |
Yuan et al., 2005 [55] | 106 (49) | Patients with breast tumors scheduled for surgery | Static | 89.5% 91.9% | 0.1K | Visual | Â |
Arora et al., 2008 [56] | 94 lesions in 92 patients (60) | Patients undergoing breast biopsy because of abnormal mammographic and/or ultrasound findings | Dynamic | 90.0% 44.1% | 0.08K | Proprietary software | Funded by Infrared Sciences Corp. (manufacturer) |
Wishart et al., 2010 [57] | 106 lesions in 100 women (65) | Patients undergoing breast biopsy because of a lump on clinical breast examination or an abnormality on screening mammography | Dynamic | 78% 48% | Unknown | Visual | Equipment loaned by Infrared Sciences Corp. (manufacturer) Higher sensitivity (78%) and specificity (75%) with the proposed algorithm in women under 50. When combined with mammography, sensitivity increased to 89% |
Wang et al., 2010 [58] | 298 lesions in 276 patients (174) | Women undergoing excisional biopsy because of suspicious findings on mammograms or ultrasound | Static | 92.0% 44.3% | 0.1K | Linear combination of five IR signs | Funded by AG Digital Technology Corp. (manufacturer) Thermogram interpretation required previous manual localization of the suspicious region |
Kontos et al., 2011 [59] | 63 (20) | Symptomatic patients | Static | 25% 85% | Unknown | Color differences | Equipment loaned by Meditherm (manufacturer) |
Kolaric et al., 2013 [60] | 26 (20) | Patients scheduled for surgery with pathophysiological examination | Unknown | 100% 55.6% | 0.07K | Visual | 5 of the 20 patients had bilateral cancer Thermography detected 5 cancers that were missed by mammography |
Collet et al., 2014 [61] | 105 lesions in 99 patients (33) | Women undergoing biopsy for suspicious imaging findings | Dynamic | 78.8% 48.6% | Unknown | Proprietary software | Â |
Yao et al., 2014 [62] | 2036 (480) | Women undergoing biopsy for abnormal mammography and/or ultrasound results | Static | 84.4% 94.0% | Unknown | Thermal releasing curve | Thermography obtained higher sensitivity than mammography (78.3%) and ultrasound (83.1%), but specificity of mammography was higher (98%) |
Francis et al., 2014 [63] | 36 (12) | Patients referred to the breast clinic for suspected breast abnormalities | Static | 83.3% 83.3% | Unknown | Machine Learning | Rotational thermography. Images were acquired with the patient lying prone with the breasts suspended through openings in the imaging bed |
Omranipour et al., 2016 [64] | 132 (87) | Women undergoing biopsy for any clinical, mammographic, or ultrasonographic finding. They attended for any breast-related complaint or just for opportunistic screening | Dynamic | 92.8% 41.9% | 0.08K | Visual | Mammography had a lower sensitivity (80.5%) but a higher specificity (73.3%). The combination of mammography and thermography yielded a sensitivity of 96.6% with a specificity of 44.4% |
Araujo et al., 2018 [65] | 50 (14) | Patients with a suspected mass | Unknown | 92.86% 86.11% | Unknown | Parameterized Mahalanobis distance | Â |
Morales-Cervantes et al., 2018 [66] | 206 (8) | Patients with clinical evidence of a tumor suggestive of cancer had risk factors for breast cancer and were heading for consultation | Static | 100% 68.68% | 0.07K | Asymmetric thermal and vascular pattern | Â |
Sarigoz et al., 2018 [67] | 54 (21) | Patients with palpable breast masses that were scheduled for biopsy due to suspicious findings in clinical and radiological evaluation | Static | 95.24% 69.70% | 0.1K | Thermal asymmetry | Thermogram interpretation required previous manual localization of the suspicious region |
Hellgren et al., 2019 [68] | 1727 (12) | Asymptomatic women with dense breasts | Dynamic | 58.33% 87.11% | Unknown | Proprietary software | Funded by Real Imaging Ltd. (manufacturer) All patients with suspicious thermograms but normal mammograms and ultrasounds were referred to MRI Thermography detected 6 cancers in 5 women, in addition to the 7 that were identified by mammography |
Sun et al., 2019 [69] | 407 (196) | Patients with non-inflammatory, unilateral, and single-breast lesions undergoing biopsy | Static | 86.7% 86.7% | Unknown | Thermal releasing curve, thermal asymmetry, and vascular features | Â |
Kakileti et al., 2020 [70] | 470 (78) | 470 women who presented for a breast health checkup 50.6% symptomatic: 49.4% asymptomatic: | Static | 91.02% 82.39% 89.85% 69.04% 100% 92.41% | 0.02K—0.5K | Proprietary software that uses ML | Funded by Thermalytix (manufacturer) Images are acquired at two sites with three different thermal cameras Asymptomatic: 100% sensitivity and 92.41% specificity Symptomatic: 89.85% sensitivity and 69.04% specificity 27% young women (under 40): 100% sensitivity. 11 of the cancers were discovered in this group, all of them symptomatic Mammography had lower sensitivity (85.96%) but higher specificity (94.05%). Among the asymptomatic group, the sensitivity of mammography was only 50% Thermography correctly identified 71% of the T1 tumors (smaller than 2 cm), whereas mammography detected 68% of them |
Singh et al., 2021 [71] | 258 (63) | Symptomatic women | Static | 82.5% 80.5% | 0.02K | Proprietary software that uses ML | Funded by Thermalytix (manufacturer) |
Da Luz et al., 2022 [72] | 25 (10) | Patients referred for evaluation after breast lump detection 72% had dense breasts | Static | 83% 74% | Unknown | Thermal asymmetry | Thermogram interpretation required previous manual localization of suspicious regions Mammography obtained lower sensitivity (73%) and specificity (70%) and missed 4 cancers, all of them in patients with dense breasts. 6 of the biopsied benign nodules had BIRADS 4–6 on mammography (unnecessary biopsies) |
Bansal et al., 2023 [73] | 459 (21) | Women attending screening mammography; 36.6% with dense and 63.4% with non-dense breast tissue, 15% symptomatic and 85% asymptomatic | Unknown | 95.24% 88.58% | Unknown | Proprietary software that uses ML | Funded by Thermalytix (manufacturer) In women with dense breasts, the sensitivity was 100% and the specificity was 81.65%. Among these women, 22% were reported as BI-RADS 0 on mammography |
MartÃn-Del-Campo-Mena et al., 2023 [74] | 3337 (129) | Patients receiving screening mammography or complementary tests due to suspicion of breast cancer | Dynamic | 94.87% 72.26% | 0.02K | Machine Learning | Mammography had higher sensitivity (100%) and specificity (97.1%) |