Accuracy and variation of biparametric prostate MRI reporting across a range of reader experience: The global BooMeR Study - initial results

R Miller, J.P. Radtke , Boris A. Hadaschik, N. Hansen , J. Teoh, Francesco Giganti, Clare Allen, Caroline M. Moore, R. Miano, Saoud R, Scott Eggener, A Whan , Geoffrey A. Sonn, Samir Taneja, T. Barrett , Antonio C. Westphalen, Ivo G. Schoots, Jeremy Grummet

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    ntroduction & Objectives: AUA, EAU and NICE now recommend multiparametric (mp)MRI prior to all prostate biopsies, including initial. This
    creates a massive increase in demand on the limited resource of access to MRI. Biparametric (bp)MRI, without dynamic contrast enhancement
    (DCE) series, is rapidly gaining interest as a faster, cheaper, less invasive way of performing prostate MRI with the goal of maintaining diagnostic
    accuracy. Using an online training tool, the global BooMeR Study assesses accuracy and variation in reading of bpMRI.
    Materials & Methods: A free bpMRI version of the online mpMRI training program MRI PRO (prostatemristudy.com) was promoted via email,
    Twitter and LinkedIn from August to October 2019 to target radiologists and urologists around the world. MRI PRO is an interactive program which
    matches 300 prostate mpMRIs acquired and reported to PIRADS v.2 standard to wholemount radical prostatectomy in positive MRIs and template
    transperineal biopsy histology for negative MRIs. The bpMRI version matches 50 cases, without any DCE series. We designated true PIRADS 4 or
    5 as true positive and true PIRADS 1 or 2 as true negative. True PIRADS 3 (equivocal) cases were excluded from analysis. MRI PRO’s proprietary
    analysis tool was used to compare users’ responses.
    Results: 59 prostate MRI readers registered for the study. 33% were radiologists, 67% were urologists and all respondents were consultants or
    fellows. 59%, 12%, 9%, 20% were from Europe, Asia-Pacific, North America, or elsewhere, respectively. 33% had previously read over 100 prostate
    mpMRIs. A total of 1,090 cases were completed, for a mean of 18.4 cases per reader. 15 readers completed all 50 cases. The overall specificity
    and sensitivity was 78% and 71% respectively. Cohorts of experienced vs inexperienced readers were then compared where readers who had
    performed more than 4 cases were included. Readers with over 100 previous cases of experience [n=13] had specificity of 77.7% (95% CI 70.9 -
    84.4) and sensitivity of 77.2% (95% CI 70.2, 84.2). Readers with less than 100 previous cases of experience [n=18] had a specificity of 53.8%, 95%
    CI [41.9%, 65.7%] and sensitivity of 64.4%, 95% CI [59.9%, 68.8%]. The difference in sensitivity (p=0.044) and specificity (p=0.003) between the
    two cohorts were statistically significant). PIRADS 3 reporting was 3.9% vs 8.2% in the experienced vs inexperienced groups.
    Conclusions: Preliminary BooMeR Study results suggest variation in bpMRI reporting accuracy and likelihood of reporting PIRADS 3 are
    associated with reader experience. Adequate training and quality assurance in reporting bpMRI is essential.
    Abstracts EAU20 Virtual Congress and Theme Week
    Eur Urol Open Sci 2020;19(Suppl 2):e473
    Original languageEnglish (US)
    Title of host publicationEuropean Urology Open Science
    PublisherElsevier
    StatePublished - 2020

    Fingerprint

    Dive into the research topics of 'Accuracy and variation of biparametric prostate MRI reporting across a range of reader experience: The global BooMeR Study - initial results'. Together they form a unique fingerprint.

    Cite this