e13586 Background: Screening mammography is often found to have low sensitivity in women with high density breast tissues. Alternate modalities of breast USG and MRI require high-quality expensive equipment making the regular screening with these modalities less affordable and accessible, particularly in resource-constrained settings This study evaluates the clinical performance of an AI-based test (Thermalytix) that uses machine learning on breast thermal images which could potentially be a low-cost solution for breast screening in low- and middle-income countries (LMICs). Methods: The prospective comparative study conducted from December 2018 to January 2020 evaluated the performance of Thermalytix in women with dense and non-dense breast tissue who presented for a health check-up at a hospital. All women underwent Thermalytix and mammography. Further investigations were recommended for participants who were reported as positive on either test. Sensitivity and specificity of Thermalytix were evaluated across age-groups, menopausal status, and breast densities. Results: Among the 687 women recruited for the study, 459 women who satisfied the inclusion criteria were included in the analysis. 168 women had ACR categories ‘c’ or ‘d’ dense breasts, of which 37 women had an inconclusive mammography report (BI-RADS 0). Overall, 21 women were detected with breast cancer in the study. Thermalytix demonstrated an overall sensitivity of 95.2% (95% CI, 76.1-99·9) and a specificity of 88.6% (95% CI, 85.2-91.4). Among women with dense breast tissue (n=168), Thermalytix showed a sensitivity of 100% (95% CI, 69.2-100) and a specificity of 81.7% (95% CI, 74.7-87.4). In women with ACR categories ‘c’ and ‘d’ dense breasts, mammography reported 22% of them as inconclusive (BI-RAD 0), while in the same sub-set of the population Thermalytix demonstrated a sensitivity of 100%. Conclusions: The AI-based Thermalytix demonstrated high sensitivity and specificity in the study cohort. It also fared well in women younger than 50 years and pre-menopausal women where routine mammography screening yields low sensitivity. Overall, this study introduces Thermalytix, a promising radiation-free, automated, and privacy-aware test that can supplement mammography for routine screening of women, especially in women with dense breast tissue, and has the potential to influence the clinical practice in LMICs by making breast cancer screening portable and affordable. Performance of Thermalytix and mammography in women with high breast densities (ACR categories ‘c’ and ‘d’ breasts). Clinical trial information: NCT04688086. [Table: see text]
Correction: Computer-Aided Detection for Breast Cancer Screening in Clinical Settings: Scoping Review
