Image Recognition of Yeast Cells: Automated Detection and Classification of Cell

Institut

Lehrstuhl für Bioseparation Engineering (TUM-ED)

Typ

Masterarbeit /

Inhalt

experimentell / theoretisch /  

Beschreibung

Every division of a budding yeast cell leaves a scar on the mother cell’s surface. The number of scars can therefore indicate the cell’s replicative age. Counting these scars manually is time-consuming and can vary between observers.

In this thesis, you will develop an automated image-analysis workflow to detect individual yeast cells, classify them by scar count, and report the number of cells in each category. The project combines microscopy and image annotation with software development and machine learning. The goal is a documented, reproducible prototype that makes the analysis of larger image sets faster and more consistent.

Tasks

• Prepare a consistent microscopy image dataset and define scar-count categories.
• Annotate representative cells and assess annotation consistency.
• Create independent training, validation, and test sets.
• Develop a pipeline for image preprocessing, cell detection and segmentation, scar recognition, classification, counting, and visualization.
• Compare suitable classical image-analysis and deep-learning approaches.
• Validate the results against expert annotations, analyze errors and uncertainty, and document the prototype.

Voraussetzungen

• Student in biotechnology, bioinformatics, computer science, data science, engineering, or a related field.
• Interest in microscopy, biological image analysis, and machine learning.
• Programming experience, preferably in Python.
• Previous deep-learning experience is helpful, but not required.
• Motivation to work across experimental biology and software development.

Möglicher Beginn

sofort

Kontakt

Torben Bardel
Raum: MW 3434
torben.bardeltum.de

Ausschreibung