Project Description:
Hirschsprung’s Disease, a congenital condition affecting 1 in 5000 newborns, is diagnosed through the detection of ganglion cells in rectal biopsy slides – a time consuming process traditionally reliant on expert pathologists. To address this, our team developed an AI-powered diagnostic tool that uses a convolutional neural network to automatically identify ganglion cells from biopsy images. Integrated into a user-friendly web application, the tool allows clinicians to upload slides and receive quick diagnostic results. Through iterative prototyping, we found that 256x256 grayscale image tiles paired with a large neural network and 50 training epochs produced the highest recall for detecting Hirschsprung’s Disease. Ethically, our solution promotes global healthcare access and diagnostic efficiency, though it necessitates careful mitigation of potential biases and over-reliance on AI to prevent misdiagnosis.
