Abstract by Zachary Brown
Democratizing Cancer Screening: Deep Methods for Reducing Costs in Low-Resource Communities
There is a tremendous disparity among those afflicted with cervical cancer world-wide: 90% of deaths resulting from this disease occur in low and middle income countries, comprising only 85% of world-wide cases. This is significant considering we are dealing with mortality figures in the hundreds of thousands (250,000 worldwide in 2017). Our research focuses on closing this gap, particularly in India.
We begin by describing the current best methods for digital pathology and then highlight the key challenges in India and how we are addressing them. The three key challenges are: 1) staining slides due to cost and poor environmental conditions, 2) annotating slides by trained professionals, and 3) an absence of certified pathologists.
To address these challenges, we present methods we are using to produce a dataset that more closely resembles the situation in India, give a high-level description of how we will annotate our data, and discuss our pipeline for automated cancer screening.