AI: New horizons in urological practice

In his presentation “AI for pathology reporting”, Prof. Geert Litjens (NL) reviewed diagnostic, prognostic and predictive applications of AI in urological cancers, highlighting that AI can support cancer detection and Gleason grading at expert level.  However, he stresses the continued need for more transparent systems, as seen in the work by Sun et al. in Medical image computing and computer assisted interventions (2025).

Prof. Litjens also emphasised the need for better multimodal integration in AI. “The current AI models lack multimodal integration for accurate biochemical prediction occurrence (BCR)”. He highlights the recent CHIMERA (Combining Histology, medical (radiology) and molecular data for medical pRognosis and diagnosis) Challenge, which aims to advance precision medicine through its uniquely composed multimodal dataset. CHIMERA is a multimodal AI model combining transcriptomics, histopathology and radiology. Tasks include pairing MRI plus pathology of the prostate to predict biochemical recurrence (BCR), and pairing H & E (haematoxylin and eosin staining) plus RNA (ribonucleic acid) data of bladder cancer to predict overall survival. He expects this to have a big impact on AI-driven PCa research.

_____

“The future is not man versus machine but man with machine”, stated Ass. Prof. Giuseppe Fanelli (IT) in his presentation on ‘prostate cancer pathology in the AI era: Impact and horizons’, where he highlighted that AI tools are powerful assistants, but accountability remains with the pathologists. “Vendors and institutions share responsibility through proper validation, monitoring, and compliance with regulatory frameworks.”

He illustrates how AI is being used for computational pathology with its ability to extract clinically actionable knowledge using computational methods from complex, high-dimensional laboratory and clinical data, thereby yielding more precise diagnosis, disease stratification, and selection of patient-specific treatments.

Ass. Prof. Fanelli stresses that data alone is not enough, analysis tools are required. “There are many vendor solutions and general-purpose machine learning tools, but none satisfied all our requirements, so we built our own.” He shares details of the PathML, which is a fully open-source research toolkit able to support the entire digital pathology research workflow.

The next step to advancing pathology digitalisation according to Ass. Prof. Fanelli is the integration of digital slides data with clinical, radiological and genomic information.

You can watch the full presentations via webcast recordings at the EMUC25 Resource Centre.

The post Controversial issues in urological pathology: a multidisciplinary approach appeared first on EMUC25.

ESUI22 is being held in conjunction with EMUC22, the 14^th European Multidisciplinary Congress on Urological Cancers, and a variety of satellite meetings. Together, this four-day event gives an extensive update on urological cancers in a multidisciplinary perspective, featuring speakers representing ESMO, ESTRO and the EAU.

Potential of AI

In the first plenary session “Standard today, but what about tomorrow?” radiologist Prof. Jelle Barentsz (Nijmegen, NL) presented the need for new protocols of quality control in imaging, and re-assessed the role of artificial intelligence (AI) now that its use has widened.

AI had made improvements, even in the past two years, but it still cannot compete with an expert gaze. It is however approaching the level of skill of a typical radiologist and this opens up new possibilities. According to Prof. Barentsz, AI is the radiologist’s friend and can help diagnosis in several ways. “Use of AI can shorten reporting time, and helps the radiologist with easy, more obvious cases. It can make the report and annotate the region, speeding up evaluation time. It also improves detection by offering a ‘double read’: an extra check by a computer that’s never tired.”

Importantly, the radiologist “remains in control” but AI can ease the workload, especially in the post-COVID period where medical experts are prone to burn-outs. The radiologist can focus on consulting, and has more time for creative thought. Interestingly, use of AI also allows integration with non-imaging tools and other “-omics”.

Prof. Barentsz went on to give a quick overview of his experiences with the Transurethral Ultrasound Ablation, or TULSA, procedure at the Busch Center (in Alpharetta GA, USA). Initial impressions left him “flabbergasted”, and he concluded that it will certainly play a role in the coming years as a viable treatment option.

Other developments on the horizon

During the same session, Dr. Vincenzo Scattoni (Milan, IT) offered a urologist’s perspective on the development of TRUS over the past few decades. Dr. Scattoni concluded: “TRUS’s utility in clinical practice has been continuously confirmed over the years, but unfortunately there are no current developments that have proven to significantly improve cancer detection. Based on well-designed controlled studies, the combination of targeted biopsy schemes and systematic biopsies provides the highest detection rate.”

Prof. Frederik Giesel (Dusseldorf, DE) had a pre-recorded presentation on the different tracers that were beginning to see use in PSMA PET-CT. A particular highlight was the emergence of PSMA-ligand therapy, currently in Phase II trials and carefully starting to be included in medical guidelines: “Exciting times are waiting for us in the future of PSMA,” said Prof. Giesel.

The post ESUI22: Examining the near future of imaging appeared first on EMUC25.