Short title: PENIMA – Penile Imaging

Aim: In WP II, Penima, we aim to develop an image recognition software as an explainable artificial intelligence (AI) based decision-supporting tool for health professionals evaluating penile lesions. The ambition is that this tool could catalyze a new and optimized patient trajectory in men with penile lesions. We plan to develop the AI using digital images of genital lesions and their clinical course. The AI will be available to health care professionals for assistance in recognizing rare changes, biopsy strategies and the need to refer the patient to the right treatment at the correct health care site.

Method: HESMEGLE WP II is carried out as a clinical prospective data acquisition study in private dermatology practice, dermatology and urology university hospital settings. After written consent from participants, digital images will be collected on separate and uniquely identified memory cards in a safe and structured data-secure environment. The logistics and data-handling have previously been thoroughly tested and approved for a similar project on bladder lesions and proved to function well in a clinical setting. Images are entered into the artificial intelligence learning platform and retrospectively validated according to histology or clinical course (response to treatment or progression). This will build and reinforce our explainable artificial intelligence-based decision-supporting tool prototype for genital lesions during a period of two years. The application for morphology recognition is already within the study group and developed by our collaborator Jacob Elmose Jensen from the company Cystotech and will be customized to fit study needs. It will be the role of the PhD student to train the artificial intelligence in collaboration with supervisors, company employees and experts in the field.

Active Inclusion Sites: 

Dept. of Urology at AUH

Urological outpatient clinic at Regionshospitalet Randers

Urological outpatient clinic at Regionshospitalet Horsens

Included patients: 

Included patients as of May 2024: 340

Collected photos as of May 2024: 2600