scholarly journals Deep learning–assisted prostate cancer detection on bi-parametric MRI: minimum training data size requirements and effect of prior knowledge

2021 ◽  
Author(s):  
Matin Hosseinzadeh ◽  
Anindo Saha ◽  
Patrick Brand ◽  
Ilse Slootweg ◽  
Maarten de Rooij ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Deep Learning ◽  
Prior Knowledge ◽  
Training Data ◽  
Expert Performance ◽  
Free Response ◽  
Algorithm Performance ◽  
Prostate Imaging ◽  
Key Points ◽  
Clinically Significant

Abstract Objectives To assess Prostate Imaging Reporting and Data System (PI-RADS)–trained deep learning (DL) algorithm performance and to investigate the effect of data size and prior knowledge on the detection of clinically significant prostate cancer (csPCa) in biopsy-naïve men with a suspicion of PCa. Methods Multi-institution data included 2734 consecutive biopsy-naïve men with elevated PSA levels (≥ 3 ng/mL) that underwent multi-parametric MRI (mpMRI). mpMRI exams were prospectively reported using PI-RADS v2 by expert radiologists. A DL framework was designed and trained on center 1 data (n = 1952) to predict PI-RADS ≥ 4 (n = 1092) lesions from bi-parametric MRI (bpMRI). Experiments included varying the number of cases and the use of automatic zonal segmentation as a DL prior. Independent center 2 cases (n = 296) that included pathology outcome (systematic and MRI targeted biopsy) were used to compute performance for radiologists and DL. The performance of detecting PI-RADS 4–5 and Gleason > 6 lesions was assessed on 782 unseen cases (486 center 1, 296 center 2) using free-response ROC (FROC) and ROC analysis. Results The DL sensitivity for detecting PI-RADS ≥ 4 lesions was 87% (193/223, 95% CI: 82–91) at an average of 1 false positive (FP) per patient, and an AUC of 0.88 (95% CI: 0.84–0.91). The DL sensitivity for the detection of Gleason > 6 lesions was 85% (79/93, 95% CI: 77–83) @ 1 FP compared to 91% (85/93, 95% CI: 84–96) @ 0.3 FP for a consensus panel of expert radiologists. Data size and prior zonal knowledge significantly affected performance (4%, $$p<0.05$$ p < 0.05 ). Conclusion PI-RADS-trained DL can accurately detect and localize Gleason > 6 lesions. DL could reach expert performance using substantially more than 2000 training cases, and DL zonal segmentation. Key Points • AI for prostate MRI analysis depends strongly on data size and prior zonal knowledge. • AI needs substantially more than 2000 training cases to achieve expert performance.

Comparison of prostate imaging reporting and data system v2.1 and 2 in transition and peripheral zones: evaluation of interreader agreement and diagnostic performance in detecting clinically significant prostate cancer

2021 ◽  
pp. 20201434
Author(s):  
Yasuyo Urase ◽  
Yoshiko Ueno ◽  
Tsutomu Tamada ◽  
Keitaro Sofue ◽  
Satoru Takahashi ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Diagnostic Performance ◽  
Area Under The Curve ◽  
Data System ◽  
Transition Zones ◽  
Prostate Imaging ◽  
Institutional Review ◽  
3.0 T ◽  
Clinically Significant ◽  
Interreader Agreement

Objectives: To evaluate the interreader agreement and diagnostic performance of the Prostate Imaging Reporting and Data System (PI-RADS) v2.1, in comparison with v2. Methods: Institutional review board approval was obtained for this retrospective study. Seventy-seven consecutive patients who underwent a prostate multiparametric magnetic resonance imaging at 3.0 T before radical prostatectomy were included. Four radiologists (two experienced uroradiologists and two inexperienced radiologists) independently scored eight regions [six peripheral zones (PZ) and two transition zones (TZ)] using v2.1 and v2. Interreader agreement was assessed using κ statistics. To evaluate diagnostic performance for clinically significant prostate cancer (csPC), area under the curve (AUC) was estimated. Results 228 regions were pathologically diagnosed as positive for csPC. With a cutoff ≥3, the agreement among all readers was better with v2.1 than v2 in TZ, PZ, or both zones combined (κ-value: TZ, 0.509 vs 0.414; PZ, 0.686 vs 0.568; both zones combined, 0.644 vs 0.531). With a cutoff ≥4, the agreement among all readers was also better with v2.1 than v2 in the PZ or both zones combined (κ-value: PZ, 0.761 vs 0.701; both zones combined, 0.756 vs 0.709). For all readers, AUC with v2.1 was higher than with v2 (TZ, 0.826–0.907 vs 0.788–0.856; PZ, 0.857–0.919 vs 0.853–0.902). Conclusions: Our study suggests that the PI-RADS v2.1 could improve the interreader agreement and might contribute to improved diagnostic performance compared with v2. Advances in knowledge: PI-RADS v2.1 has a potential to improve interreader variability and diagnostic performance among radiologists with different levels of expertise.

scholarly journals Comparison of Likert and PI-RADS version 2 MRI scoring systems for the detection of clinically significant prostate cancer

2020 ◽  
Vol 93 (1112) ◽  
pp. 20200298 ◽  
Author(s):  
Jeries P Zawaideh ◽  
Evis Sala ◽  
Maria Pantelidou ◽  
Nadeem Shaida ◽  
Brendan Koo ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Area Under The Curve ◽  
Scoring Systems ◽  
Detection Rates ◽  
Prostate Imaging ◽  
Prostate Biopsies ◽  
Threshold Score ◽  
Lesion Level ◽  
Clinically Significant ◽  
Histological Results

Objective: To compare the performance of Likert and Prostate Imaging–Reporting and Data System (PI-RADS) multiparametric (mp) MRI scoring systems for detecting clinically significant prostate cancer (csPCa). Methods: 199 biopsy-naïve males undergoing prostate mpMRI were prospectively scored with Likert and PI-RADS systems by four experienced radiologists. A binary cut-off (threshold score ≥3) was used to analyze histological results by three groups: negative, insignificant disease (Gleason 3 + 3; iPCa), and csPCa (Gleason ≥3 +4). Lesion-level results and prostate zonal location were also compared. Results: 129/199 (64.8%) males underwent biopsy, 96 with Likert or PI-RADS score ≥3, and 21 with negative MRI. A further 12 patients were biopsied during follow-up (mean 507 days). Prostate cancer was diagnosed in 87/199 (43.7%) patients, 65 with (33.6%) csPCa. 30/92 (32.6%) patients with negative MRI were biopsied, with an NPV of 83.3% for cancer and 86.7% for csPCa. Likert and PI-RADS score differences were observed in 92 patients (46.2%), but only for 16 patients (8%) at threshold score ≥3. Likert scoring had higher specificity than PI-RADS (0.77 vs 0.66), higher area under the curve (0.92 vs 0.87, p = 0.002) and higher PPV (0.66 vs 0.58); NPV and sensitivity were the same. Likert had more five score results (58%) compared to PI-RADS (36%), but with similar csCPa detection (81.0 and 80.6% respectively). Likert demonstrated lower proportion of false positive in the predominately AFMS-involving lesions. Conclusion: Likert and PI-RADS systems both demonstrate high cancer detection rates. Likert scoring had a higher AUC with moderately higher specificity and lower positive call rate and could potentially help to reduce the number of unnecessary biopsies performed. Advances in knowledge: This paper illustrates that the Likert scoring system has potential to help urologists reduce the number of prostate biopsies performed.

scholarly journals Impact of PI-RADS Category 3 lesions on the diagnostic accuracy of MRI for detecting prostate cancer and the prevalence of prostate cancer within each PI-RADS category: A systematic review and meta-analysis

2020 ◽  
pp. 20191050
Author(s):  
Akshay Wadera ◽  
Mostafa Alabousi ◽  
Alex Pozdnyakov ◽  
Mohammed Kashif Al-Ghita ◽  
Ali Jafri ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Meta Analysis ◽  
Test Group ◽  
Positive Test ◽  
Test Accuracy ◽  
Prostate Imaging ◽  
Lower Specificity ◽  
Meta Regression ◽  
Clinically Significant ◽  
Low Prevalence

Objective: To evaluate Prostate Imaging Reporting and Data System (PI-RADS) category 3 lesions’ impact on the diagnostic test accuracy (DTA) of MRI for prostate cancer (PC) and to derive the prevalence of PC within each PI-RADS category. Methods: MEDLINE and Embase were searched until April 10, 2020 for studies reporting on the DTA of MRI by PI-RADS category. Accuracy metrics were calculated using a bivariate random-effects meta-analysis with PI-RADS three lesions treated as a positive test, negative test, and excluded from the analysis. Differences in DTA were assessed utilizing meta-regression. PC prevalence within each PI-RADS category was estimated with a proportional meta-analysis. Results: In total, 26 studies reporting on 12,913 patients (4,853 with PC) were included. Sensitivities for PC in the positive, negative, and excluded test groups were 96% (95% confidence interval [CI] 92–98), 82% (CI 75-87), and 95% (CI 91-97), respectively. Specificities for the positive, negative, and excluded test groups were 33% (CI 23-44), 71% (CI 62-79), and 52% (CI 37-66), respectively. Meta-regression demonstrated higher sensitivity (p < 0.001) and lower specificity (p < 0.001) in the positive test group compared to the negative group. Clinically significant PC prevalences were 5.9% (CI 0-17.1), 11.4% (CI 6.5–17.3), 24.9% (CI 18.4–32.0), 55.7% (CI 47.8–63.5), and 81.4% (CI 75.9–86.4) for PI-RADS categories 1, 2, 3, 4 and 5, respectively. Conclusion: PI-RADS category 3 lesions can significantly impact the DTA of MRI for PC detection. A low prevalence of clinically significant PC is noted in PI-RADS category 1 and 2 cases. Advances in knowledge: Inclusion or exclusion of PI-RADS category 3 lesions impacts the DTA of MRI for PC detection.

scholarly journals Sensitivity of prostate cancer foci using 1.5 T multiparametric magnetic resonance imaging

2020 ◽  
Author(s):  
Suguru Ito ◽  
SEI NAITO ◽  
Takafumi Narisawa ◽  
Mayu Yagi ◽  
Yuta Kurota ◽  
...  
Keyword(s):  
Magnetic Resonance Imaging ◽  
Prostate Cancer ◽  
Magnetic Resonance ◽  
False Negative ◽  
Resonance Imaging ◽  
Prostate Imaging ◽  
Multiparametric Magnetic Resonance Imaging ◽  
Negative Results ◽  
False Negative Results ◽  
Clinically Significant

Abstract Background The detection of prostate cancer (CaP) has increasingly being carried out by multiparametric magnetic resonance imaging (mpMRI). Despite many previous studies, the sensitivity for clinically significant CaP (csCaP) was high, information on mpMRI false-negative lesions is limited. Therefore, the aim of this study was to evaluate the use and limitations of mpMRI in CaP. Methods A total of 228 CaP foci in 100 patients who underwent 1.5 T mpMRI and radical prostatectomy between December 2015 and June 2017 were retrospectively analyzed. The sensitivities of CaP foci, csCaP, and index tumors (ITs) were measured. Clinically significant CaP was defined into two categories based on the Gleason score (GS): csCaP/GS ≥ 3 + 4 (GS ≥ 3 + 4 or diameter > 10 mm) and csCaP/GS ≥ 4 + 3 (GS ≥ 4 + 3 or diameter > 10 mm). In addition, the characteristics of false-negative lesions were identified. The Prostate Imaging Reporting and Data System version 2 was used to determine an mpMRI positive lesion, defined as a lesion having a score of ≥ 3. Results The sensitivity of all legions, csCaP/GS ≥ 3 + 4, csCaP/GS ≥ 4 + 3, and ITs were 61.4%, 75.8%, 83.0%, and 91%, respectively. There were 91 lesions that were mpMRI false, 40% of which were csCaP/GS ≥ 3 + 4. There were three lesions with a GS of ≥ 8 and ≥ 10 mm in the false-negative results. Conclusions mpMRI can highly detect ITs and csCaP/GS ≥ 4 + 3; however, a few large and high-GS CaPs constitute undetectable lesions in 1.5 T mpMRI.

Sign in / Sign up

Export Citation Format

Share Document