Computer Vision News - February 2022

35 Automated Echocardiographic Detection of ... About the second study, the findings show that the readers exhibit significant increases in the mean ± SE AUROC from 0.877 ± 0.019 without to 0.931 ± 0.028 with the assistance of the classifier, resulting in a 10% increase in number of confident reads (sensitivity). When they look at individual reader performance, the authors detect that, with the assistance of the AI-based classifier, the AUROC of two readers exceeded the performance of the AI classifier on its own (AUROC ¼ 0.927), with one of the readers achieving 100% sensitivity. Conclusions This paper strongly suggests that AI alone can achieve good diagnostic accuracy, and, more importantly, in a time when we are all wondering how the integration of AI models with humans would work, it shows that such a system can only improve the performance and confidence of a clinician. The study perfectly fits with the mission of Ultromics , optimize the performance of echo exams with the power of AI, providing Machine Learning decision support for Coronary Artery Disease . As such, it contributes to one of their two software packages, Echo Go Pro, designed to provides reports to support physicians in the treatment of patients with suspected CAD. Their other developed software, EchoGo Core, is instead built for the automatic analysis of the Left Ventricle, which interprets echocardiogram scans and provides accurate measurements of metrics such as global longitudinal strain (GLS), ejection fraction (EF), left ventricle end-diastolic volume (LV EDV) and so on from multi-views echo images. Due to their accuracy and cloud-based service, the solutions of Ultromics are really at the cutting-edge of medical imaging and technology. They are something we always look forward to, hoping that they will represent the future of medicine.