Master Thesis
Master Thesis: Multimodal Transcriptomic/Histology AI for PDAC Precursor Detection
10.07., Studentische Hilfskräfte, Praktikantenstellen, Studienarbeiten
The Schuefflerlab for Computational Pathology at the TUM Institute for Pathology is offering a CIT Master’s thesis in the field of medical machine learning and pathology AI (artificial intelligence). The study explores the potential unimodal and multimodal deep learning models to predict precursors of pancreatic ductal adenocarcinoma (PDAC) from histology and/or spatial transcriptomics data.
Pancreatic ductal adenocarcinoma (PDAC) has several precursors, such as acinar to ductal metaplasia (ADM) and mucinous tubular complexes (MTC), but their exact progression pattern is still unknown. Spatial transcriptomics is a technique that quantifies the local transcriptome of a tissue sample in high resolution, allowing for detailed investigation of PDAC tissue and its precursors on a molecular level. However, spatial transcriptomics is slow and expensive, limiting its application to small tissue samples. In this work, we explore the potential of deep learning (DL) in histology to detect precursors of PDAC and related expression profiles. We consider DL as a replacement technology (can DL in histology predict the precursors similar as spatial transcriptomics can?) and as an enhancement technology (can a multimodal histology/RNA model improve the precursor prediction?). A publication is aimed at the end of the thesis, depending on the results.
Please find the attached PDF for a project description.