Tethys AI

Radiology image interpretation training has remained largely static since the 1980s. The mainstay of interpretation is exposure via residency training to a variety of cases, both normal and abnormal, with interpretation guided and corrected by an attending physician. These occur mostly randomly via the day’s worklist, with supplementation of less common cases via lectures. Post-residency graduation, recurrent training is still required annually in the form of continuing medical education (CME). However, CME is provided almost exclusively in live or prerecorded lecture format. Targeted evaluation of skill set gaps and personalized remediation is not currently available unless the radiologist is willing to pay for an expensive, in-person course. The transition from a highly supervised training setting to self-guided CME is particularly difficult for junior attendings. Literature has demonstrated that early career physicians interpreting medical images are generally less qualified, less accurate and less efficient than more experienced radiologists. Our software provides an online, targeted experience that assesses a radiologist and assembles a case set that addresses specific image interpretation weaknesses identified by our novel AI algorithm. This allows the radiologist to access expert-level targeted cases and remediate skill gaps in that imaging modality, improving their confidence and efficiency.