Abstract #3504

The ReIMAGINE consortium – establishing an infrastructure for the external validation of prostate MRI lesion classification models

Natasha Thorley^1,2, Tom Syer^1,3, Swetha Srikanthan⁴, Jacob Antunes⁴, Thomas Parry¹, Teresa Marsden⁵, Rosemary Clow¹, Aida Santaolalla⁶, Mrishta Brizmohun Appayya¹, Giorgio Brembilla¹, Chris Brew-Graves¹, Zhe Min⁷, Yipeng Hu⁷, David Atkinson¹, Sue Mallett¹, Steve Rodney⁴, Paul Jacobs⁴, Jonathan Piper⁴, Hashim U Ahmed^8,9, Mark Emberton⁵, and Shonit Punwani^1,2

¹Centre for Medical Imaging, University College London, London, United Kingdom, ²Imaging Department, University College London Hospital NHS Foundation Trust, London, United Kingdom, ³Department of Radiology, University of Cambridge, Cambridge, United Kingdom, ⁴MIM Software Inc, Cleveland, OH, United States, ⁵Division of Surgical and Interventional Science, University College London, London, United Kingdom, ⁶School of Cancer and Pharmaceutical Sciences, King's College London, London, United Kingdom, ⁷Dept of Med Phys & Biomedical Eng, University College London, London, United Kingdom, ⁸Division of Surgery, Imperial College London, London, United Kingdom, ⁹Imperial Urology, Imperial College Healthcare NHS Trust, London, United Kingdom

Synopsis

Keywords: Prostate, Prostate

Motivation: Multiparametric MRI is highly sensitive for identifying clinically significant prostate cancer (csPCa), but has a poorer specificity, meaning many men undergo unnecessary prostate biopsies.

Goal(s): To evaluate whether artificial intelligence (AI) could improve the diagnostic accuracy of MRI compared to current clinical methods, including Likert score and PSA density (PSAd).

Approach: We carried out independent evaluation of a prostate MRI lesion classifier model using a large multisite and multivendor prostate MRI dataset (1,039 patients).

Results: The AI model matched the sensitivity and specificity of Likert score plus PSAd cut-offs on data similar to the training set, but did not generalise to other data.

Impact: An infrastructure has been successfully established to allow robust and independent evaluation of prostate MRI lesion classification models to accelerate the development of such tools and to ensure adequate testing pre-deployment.

How to access this content:

For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.

After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.

After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.

Click here for more information on becoming a member.