Would you like to see your presentation here, made available to a global audience of researchers?
Add your own presentation or have us affordably record your next conference.
Current methods for diagnosing acute cellular rejection (ACR) in heart transplant patients rely on manual histopathological analysis, which is time-consuming, subjective, and prone to variability. This project aims to develop an AI- driven approach to assist pathologists and streamline detection in myocardial biopsies, which would enable faster, more accurate, and objective identification from ACR, and the potential to apply the model’s ability to detect other diseases based on lymphocytic infiltration in the future. Methods including using ImageJ’s Trainable Weka Segmentation to analyze biopsy samples following heart transplant stained with H&E and standardized with a similar lymphocyte size. The primary task for the model was to identify clusters of lymphocytes, defined as ‘a cluster of over 3 blue-staining naked nuclei’. Classes included red for lymphocyte clusters, and green for ‘not-lymphocyte-clusters’. Model will be tested across new images to analyze the accuracy of detection. Preliminary outcomes include training across 7 samples, and 9 iterations of development for the model. Next steps include testing the model across 7-10 new biopsy slides, and comparing the identification across areas of manually-identified lymphocytic infiltration (the current gold standard).