Master's Thesis: Improvement of a Species Identification Assay and AI-based Time-Series Colour Analysis for the Detection of Bacterial Species and Concentration in Urinary Tract Infections.

08.04.2026, Abschlussarbeiten, Bachelor- und Masterarbeiten

The student is required to integrate testing for additional bacterial species present in UTIs, while improving upon the current ones already tested. This work is based on an existing method that has been tested in a previous performance study.

The existing model analyzes time-series colour changes of the test fields to determine bacterial species and approximate concentration. The goal of the project is to optimize this pipeline using machine learning and regression approaches, improving classification accuracy, robustness, and early prediction capability.

Requirements:
- Proven previous laboratory experience, microniology experience preferred.
- An applied, industry-aware mindset, balancing scientific rigor with manufacturability and usability
- Practical expertise in image analysis, quantitative readout extraction, and time-resolved data evaluation.
- Independent working style and willingness to learn.

Project Goals:
- Improve a paper-based species identification field, enabling improved sensitivity/specificity and expanding the possible species to test for.
- Quantitatively evaluate identification field performance using image-absed readouts and time-series analysis.
- Improve the current AI-based colour time-series model used for UTI species identification and bacterial concentration estimation in a paper-based diagnostic device.
- Perform clinical validation with real human urine samples, producing more data for improvement of current identification model.

BugSense
Heinz-Nixdorf-Chair for Biomedical Electronics
Einsteinstraße 25, 81675 Munich

Kontakt: seoras.russell@tum.de