Aim

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers and is expected to become the second leading cause of cancer-related deaths worldwide by the year 2030. In about 7% of PDAC, a special growth pattern is recognized histologically, called “large-duct pattern”, characterized by ecstatic glands in more than 50% of the tumor with diameters exceeding 0.5 mm. Further research is needed for better understanding of clinico-pathological implications of this growth pattern. In our project, we aim to create an algorithm with HALO AI to identify cases with large-duct pattern. The algorithm will be trained on hematoxylin-eosin (HE) stained whole slide images (WSI) of patients with diagnosed PDAC and the trained algorithm will be applied to an existing clinico-pathological characterized cohort of 117 PDAC cases. The final HALO markup data will be exported to the image analysis software QuPath; there, quantification of gland diameters and determination of its proportion to the total tumor areas to identify large-duct pattern cases will be performed. Correlation with clinico-pathological, and comparison of large-duct pattern and conventional PDAC will be investigated.