How AI Is Changing Prostate MRI Reporting and Diagnosis

Reading Time: 5 minutes

Medically Approved by Dr Aqua Asif (May 1st 2025)

Written by Brian Lynch

Artificial intelligence (AI) is rapidly moving from futuristic concept to practical tool in healthcare, especially in diagnosing prostate cancer. AI is changing how prostate Magnetic Resonance Imaging (MRI) scans are read, making the process faster, more accurate, and more consistent between different hospitals.

The use of MRI in the prostate cancer pathway grew significantly after major trials like PROMIS and PRECISION showed MRI before biopsy could find more significant cancers and help many men avoid invasive procedures. This led to guidelines recommending upfront MRI, causing a surge in scan numbers and straining radiology departments due to workload and variability in interpretations. AI offers solutions to these challenges.

Almar van Loon, Director of Customer Success at Quantib, a company developing AI tools like Quantib Prostate, works to help clinicians improve workflows. Quantib Prostate supports radiologists reading complex prostate MRI scans, aiming to enhance patient care by addressing the demands created by increased MRI use.

What Is AI in MRI?

AI in prostate MRI refers to software using machine learning (ML) and deep learning (DL) algorithms trained on thousands of prostate MRI images with known outcomes. These AI tools learn to recognise patterns associated with prostate cancer.

Key functions include:

• Gland segmentation: Automatically outlining the prostate gland and its different zones. This is essential for calculating prostate volume and PSA density (PSAD), a key risk indicator when considered alongside the PSA level.
• Lesion detection: Identifying potentially suspicious areas on the MRI images, sometimes highlighting them visually (e.g. using bounding boxes).
• PI-RADS support: Assisting radiologists in consistently applying the Prostate Imaging Reporting and Data System (PI-RADS) score (a 1-to-5 scale indicating cancer likelihood). This might involve automating measurements or guiding structured reporting templates.

It’s important to understand that current AI tools are designed to complement, not replace, the radiologist. AI acts as an assistant, automating time-consuming tasks like volume calculation and helping to standardise the application of PI-RADS criteria, which can otherwise vary between different readers. This aims to lead to more reproducible results and frees up radiologists’ time for complex interpretation and decision-making.

How AI improves prostate MRI

Integrating AI into the radiology workflow offers several potential benefits. Significant time savings can be achieved by automating manual tasks like calculating prostate volume. Radiologists using tools like Quantib Prostate have reported decreased reporting times, allowing them to handle more cases each day.

Faster reporting means quicker results for patients and the doctors referring to them, potentially shortening waiting times for biopsy or treatment planning, a valuable gain in busy healthcare systems like the NHS. AI can also enhance the assessment process by automatically highlighting suspicious areas and providing tools for rapid measurements.

Structured, visual reports generated with AI assistance can improve communication between radiology and urology departments, making multidisciplinary team meetings more efficient. Standardised workflows also promote consistency, reducing the variability in reads between different radiologists and hospitals. This efficiency helps manage increasing workloads and can reduce diagnostic backlogs, shortening the overall time it takes to get a diagnosis.

Key benefits of AI in scans

The potential advantages AI brings to prostate MRI include:

• Improved workflow efficiency: Automating tasks like segmentation and calculation speeds up assessment and reporting, helping manage scan volumes and reduce backlogs.
• Enhanced communication: Standardised visual reports can clarify findings for the clinical team, aiding collaboration, MDT meetings, and planning for targeted biopsies or treatment.
• Increased diagnostic confidence: AI can act as a ‘second reader’, potentially improving accuracy and boosting radiologists’ confidence, especially with challenging scans or for those with less experience.

Limitations and concerns

Despite the potential, the adoption of AI faces challenges. Over-reliance on AI output without critical appraisal (‘automation bias’) is a concern; radiologists must always evaluate the AI’s suggestions within the full clinical context. Robust validation of AI tools across diverse patient populations, different MRI scanner types, and various hospital settings is paramount. Studies like PAIR-1, testing AI on real-world NHS data, are vital to ensure these tools work reliably in everyday practice.

Integrating AI into existing hospital IT systems (like PACS, i.e. Picture Archiving and Communication System) can be complex, involving issues of data governance, patient privacy, and the need for staff training. The ‘black box’ nature of some AI models – where the exact reasoning process isn’t transparent – can sometimes hinder trust among clinicians. Furthermore, regulatory approval (like the CE mark in Europe or FDA clearance in the US) and demonstrating cost-effectiveness are key hurdles for widespread adoption.

AI use in clinics today

AI is steadily moving from the research lab into clinical practice. Commercial solutions like Quantib Prostate are being used in hospitals in Europe and the US. Studies, such as those involving Sapienza University in Rome, have shown potential benefits, including improved detection sensitivity and assistance for less experienced readers. RadNet’s acquisition of Quantib signals an intent for larger-scale integration in imaging centres.

In the UK, several NHS trusts are actively evaluating AI tools. The PAIR-1 study provided important validation for Lucida Medical’s Pi software across multiple diverse NHS sites. Trusts like Somerset and Leeds are piloting AI tools with the aim of improving efficiency and speeding up the diagnostic pathway.

Why radiologists still matter

It’s essential to reiterate that AI assists, it does not replace, the radiologist. As Almar van Loon emphasised, the goal is synergy between human expertise and machine intelligence. The radiologist’s clinical judgement remains irreplaceable. They synthesise the MRI findings with the patient’s clinical history, PSA level, examination results, and other data. They interpret complex or ambiguous patterns within the broader clinical context in a way AI currently cannot.

Furthermore, healthcare requires human connection. AI cannot replicate the empathy, nuanced communication, and reassurance that clinicians provide when discussing complex findings with patients, an element highlighted as vital in real patient testimonials. Radiologists also play a critical role in overseeing AI quality control and ensuring its outputs are used appropriately. By automating routine tasks, AI frees up radiologists to focus on complex interpretation, quality assurance, teaching, research, and communication, making their expertise potentially more vital than ever.

The future of AI in imaging

Current AI applications in prostate MRI are just the beginning. Future roles are envisioned across the entire prostate cancer care pathway:

• Early detection and screening: AI could help analyse scans more quickly and consistently, potentially making MRI-based screening programmes more feasible. As Dr Christos Mikropoulos notes, “Early diagnosis is key in prostate cancer”.
• Risk stratification: AI may become better at predicting cancer aggressiveness by analysing subtle imaging features (‘radiomics’) combined with clinical and genomic data, helping to distinguish indolent from significant disease more accurately.
• Predicting treatment response: AI might help predict how likely a tumour is to respond to specific therapies like radiotherapy or hormone therapy, guiding more personalised treatment choices.
• Monitoring and surveillance: AI could assist in tracking subtle changes in tumours over time for men on active surveillance or after treatment, potentially improving the reliability of monitoring.

AI’s ability to provide precise lesion segmentation and characterisation directly supports the planning of targeted treatments. Detailed information from AI regarding tumour location and boundaries are particularly valuable for selecting your focal therapy approach. This information is essential for determining suitable candidates and accurately targeting treatments like HIFU or NanoKnife, which treat only the cancerous areas within the prostate, aiming to minimise side effects.

The future likely involves AI integrating multi-modal data, combining information from imaging, pathology reports, genetics, and clinical history, to provide highly personalised risk assessments, outcome predictions, and tailored management strategies for each individual patient.

References

Ahmed, H.U., et al. (PROMIS Study Group). (2017). Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study. The Lancet. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)32401-1/fulltext

Giganti, F., et al. (2025). AI-powered prostate cancer detection: a multi-centre, multi-scanner validation study (PAIR-1). Presentation at European Congress of Radiology (ECR) 2025.

Kasivisvanathan, V., et al. (PRECISION Study Group Collaborators). (2018). MRI-Targeted or Standard Biopsy for Prostate-Cancer Diagnosis. New England Journal of Medicine. https://www.nejm.org/doi/full/10.1056/NEJMoa1801993

Lucida Medical. (Accessed 2024). Pi: Expert Performance with NHS Patient Data (PAIR-1 Study Summary). https://lucidamedical.com/pi-expert-performance-with-nhs-patient-data/

NHS AI Lab. (Accessed 2024). AI in imaging. https://transform.england.nhs.uk/ai-lab/ai-lab-programmes/ai-in-imaging/

Forookhi, A. et al. (2023). Bridging the experience gap in prostate multiparametric magnetic resonance imaging using artificial intelligence: A prospective multi-reader comparison study on inter-reader agreement in PI-RADS v2.1, image quality and reporting time between novice and expert readers. European journal of radiology, 161, 110749. https://doi.org/10.1016/j.ejrad.2023.110749 

Quantib. (Accessed 2024). Quantib® Prostate product overview. https://www.quantib.com/en/solutions/quantib-prostate

RadNet, Inc. (2023). RadNet’s Quantib B.V. Subsidiary Receives FDA Clearance for its Quantib Prostate™ 3.0 Software. https://www.radnet.com/about-radnet/news/quantib-fda-clearance-prostate-ai

Saha, A., et al. (PI-CAI Consortium). (2024). Artificial intelligence and radiologists in prostate cancer detection on MRI (PI-CAI): an international, paired, non-inferiority, confirmatory study. The Lancet Oncology. https://doi.org/10.1016/S1470-2045(24)00220-1

Van Loon, A. (2024). AI: The future of prostate MRI reporting? The Focal Therapy Clinic Blog. https://www.thefocaltherapyclinic.co.uk/focal-therapy/medical-suitability/mri-scans/ai-the-future-of-prostate-mri-reporting/

About UsBook Consultation