Finding the “true diagnosis”
AI algorithm versus histopathology in diagnosing skin cancer, aims to investigate the diagnostic disagreement between an AI algorithm (clinical images) and histopathology (tissue sample) and to characterize these cases.
The cases that the AI algorithm is diagnosing will consist of one or more dermoscopic pictures captured before a suspicious nevus is removed. The diagnosis given by the AI algorithm will be compared to the diagnosis given by the pathologist. In cases of diagnostic discrepancies, we will investigate, why the diagnosis differs, and which method gives the correct diagnosis. In cases with disagreement between pathologist and the AI algorithm, there will be a second and third opinion from two experienced dermatopathologists. The agreed diagnosis from the second and third opinion will be the reference standard.
With this study we will challenge the golden standard of skin cancer diagnosis and try to prove that AI algorithms will increase the diagnostic accuracy of skin cancer.
The planned algorithm validation study at OUH. The study will examine whether disconcordances between the AI and the histopathology are caused by an incorrect histopathological assessment or a fault in the AI algorithm.
Christoffer is currently part time, but will be a full time researcher at AISC Research Consortium and focuses on the underlying reasons for diagnostic discrepancies between human an AI. This study is Christoffer’s masters thesis.
+45 3868 9236
The Research Unit at the Department of Plastic Surgery
Ground Floor, Entrance 7
Borgmester Ib Juuls Vej 7