According to the Cancer Registry Foundation, the number of skin cancer cases in Belgium increased by 342% between 2004 and 2016. This makes skin cancer the most common form of cancer in our country and the upward trend is expected to continue, both for non-melanoma skin tumors and for melanomas. Current skin cancer detection method involves evaluation of the blemish by an experienced dermatologist. In addition, biopsies are often performed to examine the skin tumor in the lab. This is a time consuming and expensive process, and causes unnecessary pain and discomfort to the patient.
Infrared thermography (IRT) is a technique that could potentially be used successfully as an aid to the dermatologist in the detection of melanomas and other types of skin cancer. The principle of infrared thermography for the analysis of skin spots is based on chemical and biophysical reactions, such as blood transport, perfusion and metabolic processes, which influence the temperature of the skin. In dynamic infrared thermography for the analysis of skin spots, the skin is cooled, and then the temperature changes of the skin are imaged by an infrared camera.
This research investigates whether a measurement system based on dynamic infrared thermography can help in the detection of skin cancer. It has been investigated to what extent it is possible, on the basis of static and dynamic thermal images obtained, to distinguish different types of skin spots from each other using machine learning.
Master thesis presentation by Jurgen Wittenberg June 2021 [PDF]