AI-enhanced MRI analysis shows potential to accelerate metastatic prostate-cancer assessment and improve treatment decisions

Image credit: Michal Jarmoluk from Pixabay

(IN BRIEF) Researchers at The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust have developed AI software that automates the complex interpretation of whole-body diffusion-weighted MRI scans, which are increasingly used to evaluate advanced prostate cancer. The system integrates multiple AI models that rapidly segment anatomy, standardise scan data and detect metastatic lesions, reducing image-processing time from over an hour to mere seconds and improving consistency in disease assessment. Early performance results — including correct treatment-response identification in around 80% of trial cases — suggest the software could help clinicians make faster, more objective decisions about therapy and personalise treatment pathways. Published across two peer-reviewed studies, the approach supports precision oncology and could one day be used across hospitals to speed diagnosis, reduce reporting bottlenecks and improve outcomes for patients with metastatic disease.

(PRESS RELEASE) LONDON, 1-Dec-2025 — /EuropaWire/ — Two new research studies have shown that pairing advanced MRI technology with artificial intelligence could dramatically change how clinicians detect, track and treat advanced prostate cancer. Scientists at The Institute of Cancer Research, London, and The Royal Marsden NHS Foundation Trust have created software that integrates multiple AI models to automate the intricate tasks required to analyse whole-body diffusion-weighted MRI (WB-DWI) scans — an imaging technique increasingly used to assess metastatic prostate cancer.

The AI-driven system is designed to take over the slow and labour-intensive manual work currently required to evaluate WB-DWI scans, which can take expert radiologists more than an hour per scan. By rapidly segmenting anatomy, standardising images and identifying disease in bone, it may offer clinicians faster, more objective information about whether treatment is working and help guide decisions on whether to continue, modify or escalate therapy. In the long term, this approach could support personalised treatment planning and improve quality of life for patients with disease that has spread beyond the prostate.

WB-DWI is one of the most sensitive imaging tools available for mapping metastatic prostate cancer, providing detailed information on tumour cell density and total skeletal disease burden. However, traditional interpretation is time-consuming, subjective and not scalable for routine hospital use. The research team set out to resolve this by developing software that combines several AI models, each performing a different task in the imaging pipeline. One model can extract and outline the skeletal structure in under 25 seconds, another harmonises scan data so comparisons across patients and timepoints are accurate, while a third detects regions that appear to contain secondary cancer.

The first study, published in Computer Methods and Programs in Biomedicine, introduced an automated segmentation method trained on more than 500 scans, achieving near-human anatomical accuracy without requiring expert manual labelling. Regions such as the spine, internal organs and bone structures — areas at high risk for cancer spread — were outlined in minutes, drastically reducing processing time.

A second study, featured in Physics in Medicine & Biology, focused on automating disease-burden measurement. Instead of relying on manual review, the AI software identifies metastatic lesions, estimates total tumour volume, counts lesions and monitors how these change over time. When evaluated using scans from multiple hospitals and MRI settings, the approach correctly assessed treatment response in approximately 80% of cases. Researchers believe this could enable faster therapy adjustments, more reliable staging and standardised reporting across treatment centres.

Lead author Dr Antonio Candito (ICR) said the software aims to shift clinicians’ time from manual image marking to decision-making. “By localising key anatomical structures quickly and reproducibly, we can unlock the full potential of WB-DWI for cancer staging and treatment monitoring,” he commented. Senior author Dr Matthew Blackledge added that combining WB-DWI with AI is a key step toward precision oncology and could ultimately be applied to other cancers including breast cancer and myeloma.

While results are highly promising, the team notes that wider clinical rollout will require multi-centre testing, regulatory approval and safeguards to ensure transparency, bias control and ethical use. If validated, such technology could transform prostate-cancer management worldwide — enabling faster scan interpretation, earlier intervention and more personalised patient care.

Media Contact:

Tel: 0203 437 3502
email: mediaoffice@icr.ac.uk

SOURCE: The Institute of Cancer Research