eduzhai > Applied Sciences > Engineering >

CrossPath Top-down Cross Data Type Multi-Criterion Histological Analysis by Shepherding Mixed AI Models

  • Save

... pages left unread,continue reading

Document pages: 14 pages

Abstract: Data-driven AI promises support for pathologists to discover sparse tumorpatterns in high-resolution histological images. However, three limitationsprevent AI from being adopted into clinical practice: (i) a lack ofcomprehensiveness where most AI algorithms only rely on singlecriteria examination; (ii) a lack of explainability where AI models work as black-boxes with little transparency; (iii) a lack of integrability where itis unclear how AI can become part of pathologists existing workflow. Toaddress these limitations, we propose CrossPath: a brain tumor grading toolthat supports top-down, cross data type, multi-criterion histological analysis,where pathologists can shepherd mixed AI models. CrossPath first uses AI todiscover multiple histological criteria with H and E and Ki-67 slides based onWHO guidelines. Second, CrossPath demonstrates AI findings with multi-levelexplainable supportive evidence. Finally, CrossPath provides a top-downshepherding workflow to help pathologists derive an evidence-based, precisegrading result. To validate CrossPath, we conducted a user study withpathologists in a local medical center. The result shows that CrossPathachieves a high level of comprehensiveness, explainability, and integrabilitywhile reducing about one-third time consumption compared to using a traditionaloptical microscope.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×