Nondestructive direct digital capture of histopathology with minimal processing (minutes) combined with a dedicated AI image analysis system can reduce current rates of lumpectomy positive margins…

Nondestructive direct digital capture of histopathology with minimal processing (minutes) combined with a dedicated AI image analysis system can reduce current rates of lumpectomy positive margins requiring a second surgery to reexcise involved tissues from 25% or more to nearly zero. In pancreatic resections rates of discrepancy and or deferral to permanent sections of 10% or more can be reduced to nearly zero but importantly also with a concomitant decrease in the diagnostic difficulty and need for the most highly expert pathologists. By the end of the proposed project we will have a functional prototype system margincall which will be able to image specimen contours mark surgical resection margins serially cross section tissue fibi image serial tissue slices pathology image AI map cancer (and or in situ neoplasias) to each slice of tissue assess proximities to all margins reconstruct a 3D orientation of the serial slices and heat map the tumor proximity over the surface of the specimen contour image. This will provide the surgeon with precise areas of involvement as well as areas that are close and will distinguish areas of invasive extension from areas of in situ involvement of the glands. Importantly excision of invasive versus in situ can require a different approach surgically with in situ disease requiring an excision of the gland tree or lobe. Margincall will be able to address these challenges and function as an intraoperative end to end solution for enhanced visualization of surgical margins without the need for a pathologist. University of california davis indiana university and eindhoven university of technology are subrecipients participating in this project.