Programming a wound care computer vision algorithm with open-source electronics

MSc assignment

Client Information

The UT researcher Pep Canyelles Pericas is developing a new generation of smart patches for wound healing, particularly in complex wounds. Pep has published research works using Raspberry Pi to control complex hardware setups. The researcher has on-going collaborations with ZGT hospital in Almelo, where complex wounds are part of the strategic research. This project explores the development of an imaging system to be used at home for wound care. Recently, a UT Design Lab fellows fund project in a similar domain was awarded.

Project Background

The treatment of complex wounds requires combined clinical and home medical treatments. Currently large hospitals (such as ZGT) have wound expertise centres that are composed by clinicians from different disciplines. Patients with difficult wounds are assessed by a multidisciplinary team to arrange a suitable treatment. Commonly, the TIME (Tissue, Infection, Moist and Edge) decision making tree is followed, largely relying on the clinical’s staff expertise to diagnose and monitor wound healing. Severe patients start treatments at the hospital but as they recover, they continue treatments at home. Patient compliance to the given instructions and the lack of automated wound monitoring commonly hinder healing, setting recovery back. These are all conclusions obtained from our recent context analysis and stakeholder feedback collection from wound expertise centres, facilitated by the UT Design Lab Fellows Fund 2021 and researched by a team of interdisciplinary students from the UT and Saxion. Thus, there is a need for automated wound care monitoring technology for home use with direct connectivity to the clinic.


In this project, we want to develop an open-source-based monitoring system that displays the wound type and monitors the healing process using computer imaging means. This will be done employing colorimetry methods applied to the established medical colour code for wound progression: red (fresh wound), yellow (infection) and black (necrosis or dead tissue). We will use a combined colorimetry with wound morphology monitoring to assess healing over time. Thanks to the embedded electronics approach, home monitoring can be linked to the hospital to conveniently connect the patient with the wound expertise centres for personalised monitoring and feedback. The project has two strands: software and hardware. In this project, focus is given in programming the image recognition algorithm in Python using the Raspberry Pi high-quality imaging module. This is a challenging project aimed at proactive, responsible, and professional students willing to kick start a career in electronic engineering/computer science R&D.

Constraints & Practical Aspects

We will make use the information obtained in previous projects (via written reports), particularly the treatment decision-making process based on colorimetry. We will translate the information into a set of rules for the computer vision algorithm. This project benefits from the user interface concept developed by a previous graduation project. The algorithm will make use of the findings and recommendations reflected in the project report for the user interface appearance, functionality, and usability. Continuous interaction with end users and hospitals will be sought. The project deliverable is a Python algorithm implemented in Raspberry Pi hardware for colorimetry analysis and morphology tracking in wound healing.

Core skills needed: Python programming at medium to advanced level. Interest and previous experience in computer vison/imaging and/or in Raspberry Pi or open-source electronics projects would be highly beneficial. Affinity to the maker movement and ability to work in multidisciplinary teams will also be advantageous. The student needs to be able to work autonomously.

The project is linked to Design Lab fellows fund project. The student will manage a budget provision for prototyping.