PAIP2021: Perineural Invasion in Multiple Organ Cancer (Colon, Prostate, and Pancreatobiliary tract)

MICCAI endorsed event

A distinct feature of malignant tumors compared to normal tissue or benign tumors is the invasiveness of transformed cells. Malignant tumor cells can invade vessels, nerves, stroma (through the basement membrane), lymph nodes, and other distant organs. Metastasis to lymph nodes and distant organs is a unique feature of malignant tumors, and it affects patient survival. Nerves are normal structures which connect the brain to peripheral organs, and transmit electrical signals which are responsible for voluntary and non-voluntary motor control and sensory perception. Nerves and blood vessels are the main pathways through which malignant cells can move from a primary site to other organs. Perineural invasion by malignant tumor cells has been reported as an independent indicator of poor prognosis in cancers. The College of American Pathologists recommended the inclusion of perineural invasion in pathology reports for resected specimens of malignant tumors.

Detection of perineural invasion in small nerves on glass slides is a labor-intensive task. Histologically, perineural invasion is composed of nerve and tumor cells attached to the nerve, and the nerve may be completely enclosed within the tumor mass. The morphology of nerves is the same in all organs, but tumor cells have various forms according to the histologic type and organs they are present in, thus modelling cancer in every organ is difficult. In the case of similar histology among the different cancers, it is challenging to determine whether an algorithm can be made for common histologic elements even if the background organs are different.

This challenge aims to promote the development of a common algorithm for automatic detection of perineural invasion in resected specimens of multi-organ cancers. This challenge will have a technical impact in the following fields: detection of composite target (nerve and tumor) and common modeling for target images in multiple backgrounds. This challenge will provide a good opportunity to overcome the limitations of current diseaseorgan- specific modeling and develop a technological approach to the universality of histology in multiple organs.