Smith ET (HUMI Director) created the AI Metrics platform, an imaging trial management system and CT and MR image viewer that leverages artificial intelligence (AI) and guided workflows that adhere to the rules of tumor response criteria (e.g. RECIST 1.1, iRECIST, Lugano, RANO). The AI assists with tumor measurement, labelling and tracking (Figure 1) and provides advanced reports that include a graph, table, key images, and structured text (Figure 2).

Smith ET led a multi-institutional comparative effectiveness study with 24 radiologists and 20 oncologic providers. AI Metrics increased accuracy by 25%, decreased major errors by 99%, improve efficiency by 2X, and increased agreement by 45% compared to current practice. Furthermore, 96% of radiologist and 100% of oncologic providers preferred AI Metrics.

The AI Metrics platform was upgraded to as an imaging trial management system and is now FDA 510(k) cleared and 21 CFR Part 11 compliant. The AI Metrics platform was also upgraded to perform radiomics and extract >100 imaging features from each tumor, facilitating CT and MR imaging biomarker development. Finally, AI Metrics exports all tumor segmentations, which allows for additional analysis and training of new AI algorithms.