ID 2509: Computer Vision Models for Chronic Wound Diagnosis
Master Thesis: Computer Vision Models for Chronic Wound Diagnosis
Chronic wounds and wound healing disorders not only pose a significant burden on the healthcare system but also impact patients’ quality of life, often requiring prolonged treatment and frequent monitoring.
Using deep learning offers the possibility of supporting clinicians in wound documentation and diagnosis. In this master thesis you will research deep learning methods to support wound documentation and diagnosis on the basis of collected data from the university hospital Erlangen.
Details
You will develop a deep learning model for wound diagnosis based on the chronic wound data from the Erlangen university hospital.
Tasks
- Literature review on:
- (Chronic) Wound Diagnosis with Deep Learning
- Computer Vision Models for Medical Diagnosis.
- Open Source models for image anonymization (face detection/tattoo detection)
- Anonymize the images and remove images that contain identifying factors
- Develop, train and validate deep learning model to diagnose wound images and compare with state-of-the-art models
- Test the given model on unknown wound types.
- Analyze the errors of the model (When is it difficult for the model to make predictions?)
Requirements
- Python, Pytorch
- Preferred starting date: April/May 2025 (can be discussed)
- You should be prepared to review and analyze wound images. (For example, look up ‚Diabetic Foot Ulcer‘ on Google)
Supervisors
Please use the application form to apply for the topic. We will then get in contact with you.