Introduction : Pulmonary and cardiothoracic disorders are among the leading causes of morbidity and mortality worldwide. Although chest radiography is an economical and most widely used diagnostic tool for assessing chest abnormalities, the correct interpretation of the information can sometimes be challenging for a medical practitioner. DeepTek Chest Pathology AI Analyzer has been designed to help healthcare practitioners in identifying chest pathologies from chest radiographs. The features of smart reporting aid in quick assessment and improvement in chest pathology detection resulting in worklist prioritization.
Results :
● The standalone AI had an aggregate AUROC of 91.2% and a sensitivity of 88.4% in detecting suspicious ROIs in the Lungs, Pleura, and Cardiac regions across two different hospital datasets. It outperformed unaided human readers, who achieved an aggregate AUROC of 84.2% and sensitivity of 74.6% for the same task.
● AI assistance resulted in a significant improvement in aggregate AUROC of test readers from 84.2% to 87.9% across both hospital datasets.
Conclusion : DeepTek Chest Pathology AI Analyzer outperformed human readers in detecting suspicious ROIs in the lungs, pleura, and cardiac regions from chest radiographs. When using the AI aid, the average sensitivity and AUROC of the test readers improved by 14.2% and 4.4%, respectively. The features such as smart tags and smart reporting helped the readers in reporting chest pathologies in a short duration of time.