Prostate Brachytherapy UK & Ireland 11-12 June 2026

Proffered Presentations Archive

UK & Ireland Prostate Brachytherapy Conference 2024

Abstract number 24 

Innovative Approaches to Prostate Biopsy Decision-Making: A Machine Learning and Optimisation Framework

Sara Saadatmand, Dr Banu Lokman, Professor Ashraf Labib, Professor Dylan Jones,
University of Portsmouth 

Mr Dominic Hodgson, Dr Tomas Austin
Portsmouth Hospitals University NHS Trust 

Purpose: In the UK, around 190,000 prostate biopsies are performed every year [1], and the annual costs of pre-biopsy MRIs and biopsies for men with suspected cancer are over £32 million [2]. However, there does not exist a framework that guides healthcare professionals in personalising biopsy decisions for an individual patient. The principal objective of this study is to develop a decision support system (DSS) for prostate cancer (PCa) diagnosis to increase the detection of significant PCa while mitigating the risks associated with overdiagnosis. It involves an extensive analysis of data from prostate biopsies at Portsmouth Hospitals University NHS Trust (PHU) with the aim of developing patient-specific biopsy decision rules to assist healthcare professionals. 

Methods: The retrospective data of +2000 patients who underwent a prostate biopsy at PHU between 2018 and 2023 were extracted. The dataset comprises diverse features, including PSA, number of lesions, location of the lesions, PI-RADS score, number of cores taken from each lesion, and result. Aiming to classify the patients and support prostate biopsy-decisions, different machine learning algorithms are applied to the data set to identify relevant patient features and corresponding threshold values. In addition, an optimisation model has been developed to optimise the biopsy sampling plan, which maximises the detection rate whilst minimising the number of cores. The trade-offs between the detection rate and number of cores are then analysed to identify the optimal number of biopsy cores for each patient class. 

Results: A patient classification and decision framework have been proposed to facilitate the prostate biopsy decision-making process and generate patient-specific biopsy plans. 

Conclusion: A data-driven decision support system will support healthcare professionals in generating patient-specific biopsy plans that optimise the diagnosis of PCa and improve overall efficiency. 

References: 

  1. Prostate Cancer UK. Guidance for Clinical Commissioning Groups: Commissioning multi-parametric MRI before first biopsy for men with suspected prostate cancer [Internet]. 2017. 
  2. NICE (2019). Prostate cancer: diagnosis and management (NICE guideline [NG131]). National Institute for Health and Care Excellence (NICE).