1x Master / 2x Bachelor Arbeit in Computational Pathology bei Technische Universität München

Technische Universität München

The Computational Pathology lab at the TUM Institute of Pathology is currently offering a Master thesis or two Bachelor theses in digital pathology.

Problem:

In digital pathology, high resolution scanners digitize human tissue at high resolution (0.25-0.5 microns per pixel), resulting in gigapixel-sized whole-slide images. During the scanning process, a microscope automatically screens the tissue slide capturing image tiles that are then stitched to the complete image. High quality of digital tissue images is required for accurate diagnostic review by pathologists and machine learning models. However, two groups of artifacts commonly arise in digital pathology: artifacts during slide preparation (e.g. tissue folds, pen marks) and artifacts of the scanning process (e.g. blurriness, stripes, and missing tissue) making quality control (QC) crucial. The parameterized image processing software HistoQC(1) provides tools to analyze WSI, but needs to be adjusted and evaluated on the analyzed data set.

Goal:

To evaluate HistoQC as a QC tool for the artifacts (i) blurriness, (ii) pen marks (green, red, black), (iii) tissue folds (master thesis: additionally (iv) stripes, (v) tissue at coverslip, and (vi) dirt) on HE-stained and IHC-stained WSI (Figure 1). Based on previous work (2, 3), the student will implement and test HistoQC pipelines for accurate detection of the above artifacts for the two staining groups. The artifact detectors will then be evaluated on real TUM data to quantify the artifacts in our clinical scanning operation.
Once finalized, the artifact detectors will be installed in our scanning operation process to monitor the digital images’ quality. This work will have direct impact on the clinical workflow of TUM’s institute of pathology.

Requirements:

Knowledge in or interest in Python programming (for image analysis pipeline), possibly R programming (for statistics, can also be done in python), digital pathology, efficient image processing, pathology, debugging, visualizations, analytical thinking.

Notes:

This project can be done as one Master’s thesis (for all six artifact types) or be split into two Bachelor’s theses (for three artifact types, each).

TUM is an equal opportunity employer. TUM aims to increase the proportion of women, therefore, we particularly encourage applications from women. Applicants with severe disabilities will be given priority consideration given comparable qualifications.
Data Protection Information: As part of your application for a position at the Technical University of Munich (TUM), you submit personal data. Please note our privacy policy in accordance with Art. 13 General Data Protection Regulation (DSGVO).