多参数磁共振成像（mpMRI）已经改变了前列腺癌的诊断路径，提高了风险分层和更有针对性的后续治疗。然而，人们关注图像的观察者间变异性以及长期使用该模型的适用性，尤其考虑到目前放射科医生短缺和人口老龄化趋势。为了解决这些问题，人工智能（AI）正在整合到临床实践中，以支持诊断和治疗成像分析。本报告详述了一个系统回顾和荟萃分析的方案，研究AI在预测mpMRI上的原发性前列腺癌的准确性。 将使用PubMed、MEDLINE、Embase和Cochrane数据库进行系统搜索。所有相关文章的出版时间应在2016年1月到2023年2月之间。为了纳入研究，文章必须使用AI来研究MRI前列腺图像以检测前列腺癌。所有纳入的文章必须为全文，并报告原始数据，且以英文撰写。该方案遵循《系统回顾和荟萃分析方案的首选报告项目2015年清单》。QUADAS-2评分将评估所选研究的质量和偏倚风险。 对于此系统回顾，不需要道德批准。研究结果将通过同行评议的出版物以及在国内外会议上进行演讲来进行传播。 CRD42021293745. ©作者（或其雇主）2023。在CC BY许可下允许再利用。由BMJ出版。

Multiparametric MRI (mpMRI) has transformed the prostate cancer diagnostic pathway, allowing for improved risk stratification and more targeted subsequent management. However, concerns exist over the interobserver variability of images and the applicability of this model long term, especially considering the current shortage of radiologists and the growing ageing population. Artificial intelligence (AI) is being integrated into clinical practice to support diagnostic and therapeutic imaging analysis to overcome these concerns. The following report details a protocol for a systematic review and meta-analysis investigating the accuracy of AI in predicting primary prostate cancer on mpMRI.A systematic search will be performed using PubMed, MEDLINE, Embase and Cochrane databases. All relevant articles published between January 2016 and February 2023 will be eligible for inclusion. To be included, articles must use AI to study MRI prostate images to detect prostate cancer. All included articles will be in full-text, reporting original data and written in English. The protocol follows the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols 2015 checklist. The QUADAS-2 score will assess the quality and risk of bias across selected studies.Ethical approval will not be required for this systematic review. Findings will be disseminated through peer-reviewed publications and presentations at both national and international conferences.CRD42021293745.© Author(s) (or their employer(s)) 2023. Re-use permitted under CC BY. Published by BMJ.