May 2021
·
206 Reads
·
1 Citation
Thermology International
Breast thermography is a non-contact, non-invasive, and irradiation free imaging modality that is found to be applicable for detection of breast cancers. One of the main challenges with breast thermography is the subjectivity and the accuracy of visual interpretation of thermal images. In this paper, we present a systematic retrospective study that measures the performance of manual interpretation of breast thermal images and also evaluates the benefit of using a software visualization tool to aid this interpretation. The thermal analysis and visualization tool considered for this study was SMILE-100, which automatically transforms the temperature values into contrast-enhanced colour images and annotates elevated temperature regions to help thermologists in detection of signs of malignancy. The study was conducted in 258 symptomatic women and the reference used was the standard recommendation for diagnosing breast cancer involving a combination of mammography, ultrasound, and biopsy. The sensitivity and specificity of the thermologist's assessment without tool support was found to be 60.3% (95%CI: 48.2% to 72.4%) and 81.5% (95%CI: 76.1% to 87.0%), respectively. When the same thermologist was aided with SMILE-100 tool after a washout period of two months, a sensitivity and specificity of 81.0% (95% CI: 71.2% to 90.6%) and 75.9% (95% CI: 69.9% to 81.9%), respectively, was observed. Furthermore, the sensitivity and specificity of tool generated Hotspot annotations was found to be 85.7% (95%CI: 77.1% to 94.4%) and 70.3% (95%CI: 63.8% to 76.7%), respectively. These results show that SMILE-100 can be a good tool to improve the interpretation accuracy of thermal images and reduce the subjectivity in interpretation.