AI-powered prostate cancer detection: a multi-centre, multi-scanner validation study - Scorecard - MDSpire

AI-powered prostate cancer detection: a multi-centre, multi-scanner validation study

  • By

  • Francesco Giganti

  • Nadia Moreira da Silva

  • Michael Yeung

  • Lucy Davies

  • Amy Frary

  • Mirjana Ferrer Rodriguez

  • Nikita Sushentsev

  • Nicholas Ashley

  • Adrian Andreou

  • Alison Bradley

  • Chris Wilson

  • Giles Maskell

  • Giorgio Brembilla

  • Iztok Caglic

  • Jakub Suchánek

  • Jobie Budd

  • Zobair Arya

  • Jonathan Aning

  • John Hayes

  • Mark De Bono

  • Nikhil Vasdev

  • Nimalan Sanmugalingam

  • Paul Burn

  • Raj Persad

  • Ramona Woitek

  • Richard Hindley

  • Sidath Liyanage

  • Sophie Squire

  • Tristan Barrett

  • Steffi Barwick

  • Mark Hinton

  • Anwar R. Padhani

  • Antony Rix

  • Aarti Shah

  • Evis Sala

  • February 28, 2025

  • 0 min

Share

Clinical Scorecard: Validation Study of AI-Enhanced Detection of Prostate Cancer Across Multiple Centers and Scanners

At a Glance

CategoryDetail
ConditionProstate Cancer (PCa), clinically significant defined as Grade Group ≥ 2
Key MechanismsMagnetic resonance imaging (MRI) with Prostate Imaging-Reporting and Data Systems (PI-RADS) scoring; Deep-learning-based computer-aided detection (DL-CAD) software providing risk scores for csPCa
Target PopulationPatients ≥ 21 years old referred for prostate MRI for suspected prostate cancer
Care SettingMulti-center UK National Health Service hospitals with varying MRI scanners and protocols

Key Highlights

  • MRI improves pre-biopsy assessment and early detection of prostate cancer, enabling timely intervention.
  • DL-CAD systems approach expert radiologist performance and may reduce invasive biopsies by improving detection accuracy of clinically significant PCa.
  • Multi-center, multi-vendor validation is critical to assess AI model generalisability beyond single-site retrospective studies.

Guideline-Based Recommendations

Diagnosis

  • Use multiparametric MRI interpreted by expert genitourinary radiologists with PI-RADS or Likert scoring for suspected PCa.
  • Confirm clinically significant PCa (GG ≥ 2) by histopathology from transperineal targeted and systematic biopsies.
  • Employ AI-assisted risk scoring as an adjunct to radiologist interpretation to improve detection accuracy.

Management

  • Perform biopsy following MRI findings based on multidisciplinary team (MDT) recommendations.
  • Consider AI outputs as concurrent or second-reader support without replacing clinical judgment.
  • Exclude patients with prior PCa diagnosis or poor-quality scans from AI-based assessment.

Monitoring & Follow-up

  • Monitor PSA density (<0.15 ng/mL²) to help rule out elevated risk of clinically significant PCa when biopsy is not performed.
  • Ensure quality control of MRI acquisition protocols and scanner consistency across centers.
  • Validate AI performance continuously across diverse populations and scanner models.

Risks

  • Variability in radiologist training and MRI protocols can affect diagnostic consistency.
  • Limited generalisability of AI models trained on single-site or single-vendor data.
  • Potential for false negatives if relying solely on AI without expert radiologist review.

Patient & Prescribing Data

Patients referred for prostate MRI with suspected prostate cancer across multiple UK NHS centers

AI-based risk scoring systems can assist radiologists in identifying lesions with GG ≥ 2, potentially reducing unnecessary biopsies and improving diagnostic accuracy.

Clinical Best Practices

  • Use expert genitourinary radiologists with extensive prostate MRI experience (>1000 cases) for image interpretation.
  • Apply multidisciplinary team discussions to guide biopsy decisions based on MRI and AI findings.
  • Incorporate multi-parametric MRI sequences (T2-weighted, diffusion-weighted, dynamic contrast-enhanced) for comprehensive lesion assessment.
  • Validate AI tools on multi-center, multi-vendor datasets to ensure robustness and generalisability.
  • Maintain rigorous quality assurance for MRI acquisition and biopsy procedures.

References

Original Source(s)

AI-powered prostate cancer detection: a multi-centre, multi-scanner validation study

  • by Francesco Giganti, Nadia Moreira da Silva, Michael Yeung, Lucy Davies, Amy Frary, Mirjana Ferrer Rodriguez, Nikita Sushentsev, Nicholas Ashley, Adrian Andreou, Alison Bradley, Chris Wilson, Giles Maskell, Giorgio Brembilla, Iztok Caglic, Jakub Suchánek, Jobie Budd, Zobair Arya, Jonathan Aning, John Hayes, Mark De Bono, Nikhil Vasdev, Nimalan Sanmugalingam, Paul Burn, Raj Persad, Ramona Woitek, Richard Hindley, Sidath Liyanage, Sophie Squire, Tristan Barrett, Steffi Barwick, Mark Hinton, Anwar R. Padhani, Antony Rix, Aarti Shah, Evis Sala

  • February 28, 2025

Related Content

Suspended Lead Suits Linked to Lower Radiation in LAAO

Study shows suspended lead suits reduce head-level radiation exposure during left atrial appendage occlusion procedures

AI Falls Short on Differential Dx

New PrIME-LLM benchmark shows strong diagnostic accuracy but persistent gaps in clinical reasoning across 21 large language models

NPs Top US Jobs List Again

The nurse practitioner profession claims the No. 1 spot across three categories in the U.S. News & World Report 2026 Best Jobs rankings for the third consecutive year.