The cytological and histological assessment of human tissues has emerged as a key challenge for detection and treatment of multiple clinical conditions, including cancer. Experts analyze slide images in order to characterize samples by performing tasks like whole-slide classification, patch-wise classification, tumor localization, region segmentation etc. The results of these tasks inform the diagnosis. But a mismatch between demand and supply of experts, especially in the Indian setting and the rural context, has led to suboptimal quality and turnaround times for these analyses.

An automated histopathological analysis platform could help the lives of millions in resource-constrained communities who require screening and diagnostic services every year, by enabling a reduction in reliance on hard-to-acquire experts, increase in throughput, and improvement in efficiencies. It could be effectively deployed into the diagnostic pipeline of existing organized pathology lab chains, recuperating initial costs effectively over time while simultaneously getting better in terms of accuracy as access to more continuous data is made available.

This research aims to build a platform for automated histopathological analysis, powered by machine learning, in order to significantly reduce reliance on experts, price and time, and provide a cloud-based image analysis solution targeted at hospitals, usable by operators requiring much lower level of expertise, especially in resource and data constrained contexts.