Skin cancer is the most common cancer worldwide: one in six people in the Netherlands develops basal cell carcinoma, the mildest type. Yet every excised lesion ends up with a pathologist, and this diagnosis forms the largest workload in any pathology laboratory. This thesis shows that artificial intelligence (AI) recognizes basal cell carcinoma at the level of dermatopathologists. It also holds up during surgery, while the patient is still on the operating table, and in several international hospitals whose tissue the system had never seen. Such a system saves precious time in overburdened care by filtering these cases out of the daily stream of diagnoses. Does the AI replace the pathologist? No. The aim is to take over routine work, leaving the pathologist time for more complex cases. Whether daily practice bears that out is the next step, and with it the hope of healthcare that stays available and affordable.
About the PhD candidate
Daan Geijs (1992) obtained Master's degrees in Biomedical Engineering and physics education at the University of Twente in 2019, and taught physics at secondary school. In 2020, he started with his PhD research as part of the Computational Pathology Group and Department of Pathology of Radboud university medical center. Currently, he is working as Product Discovery Lead at kaiko.ai.