Comparison of diagnostic performance between two prostate imaging reporting and data system versions: A systematic review

PI-RADS v2.1 Combined With Prostate-Specific Antigen Density for Detection of Prostate Cancer in Peripheral Zone

Purpose

To evaluate the diagnostic performance of combining the Prostate Imaging Reporting and Data System (PI-RADS) scoring system v2.1 with prostate-specific antigen density (PSAD) to detect prostate cancer (PCa).

Methods

A total of 266 participants with suspicion of PCa underwent multiparametric magnetic resonance imaging (mpMRI) in our hospital, after at least 4 weeks all patients underwent subsequent systematic transrectal ultrasound (TRUS)-guided biopsy or MRI-TRUS fusion targeted biopsy. All mpMRI images were scored in accordance with the PI-RADS v2.1, and univariate and multivariate logistic regression analyses were performed to determine significant predictors of PCa.

Results

A total of 119 patients were diagnosed with PCa in the biopsy, of them 101 patients were diagnosed with clinically significant PCa. The multivariate analysis revealed that PI-RADS v2.1 and PSAD were independent predictors for PCa. For peripheral zone (PZ), the area under the ROC curve (AUC) for the combination of PI-RADS score and PSAD was 0.90 (95% CI 0.83-0.96), which is significantly superior to using PI-RADS score (0.85, 95% CI 0.78-0.93, P=0.031) and PSAD alone (0.83, 95% CI 0.75-0.90, P=0.037). For transition zone (TZ), however, the combination model was not significantly superior to PI-RADS alone, with AUC of 0.94 (95% CI 0.89-0.99) vs. 0.93 (95% CI 0.88-0.97, P=0.186).

Conclusion

The combination of PI-RADS v2.1 with PSAD could significantly improve the diagnostic performance of PCa in PZ. Nevertheless, no significant improvement was observed regarding PCa in TZ.

Applications of artificial intelligence in prostate cancer imaging

PURPOSE OF REVIEW

The purpose of this review was to identify the most recent lines of research focusing on the application of artificial intelligence (AI) in the diagnosis and staging of prostate cancer (PCa) with imaging.

RECENT FINDINGS

The majority of studies focused on the improvement in the interpretation of bi-parametric and multiparametric magnetic resonance imaging, and in the planning of image guided biopsy. These initial studies showed that AI methods based on convolutional neural networks could achieve a diagnostic performance close to that of radiologists. In addition, these methods could improve segmentation and reduce inter-reader variability. Methods based on both clinical and imaging findings could help in the identification of high-grade PCa and more aggressive disease, thus guiding treatment decisions. Though these initial results are promising, only few studies addressed the repeatability and reproducibility of the investigated AI tools. Further, large-scale validation studies are missing and no diagnostic phase III or higher studies proving improved outcomes regarding clinical decision making have been conducted.

SUMMARY

AI techniques have the potential to significantly improve and simplify diagnosis, risk stratification and staging of PCa. Larger studies with a focus on quality standards are needed to allow a widespread introduction of AI in clinical practice.

Multicenter Multireader Evaluation of an Artificial Intelligence-Based Attention Mapping System for the Detection of Prostate Cancer With Multiparametric MRI

OBJECTIVE. The purpose of this study was to evaluate in a multicenter dataset the performance of an artificial intelligence (AI) detection system with attention mapping compared with multiparametric MRI (mpMRI) interpretation in the detection of prostate cancer. MATERIALS AND METHODS. MRI examinations from five institutions were included in this study and were evaluated by nine readers. In the first round, readers evaluated mpMRI studies using the Prostate Imaging Reporting and Data System version 2. After 4 weeks, images were again presented to readers along with the AI-based detection system output. Readers accepted or rejected lesions within four AI-generated attention map boxes. Additional lesions outside of boxes were excluded from detection and categorization. The performances of readers using the mpMRI-only and AI-assisted approaches were compared. RESULTS. The study population included 152 case patients and 84 control patients with 274 pathologically proven cancer lesions. The lesion-based AUC was 74.9% for MRI and 77.5% for AI with no significant difference (p = 0.095). The sensitivity for overall detection of cancer lesions was higher for AI than for mpMRI but did not reach statistical significance (57.4% vs 53.6%, p = 0.073). However, for transition zone lesions, sensitivity was higher for AI than for MRI (61.8% vs 50.8%, p = 0.001). Reading time was longer for AI than for MRI (4.66 vs 4.03 minutes, p < 0.001). There was moderate interreader agreement for AI and MRI with no significant difference (58.7% vs 58.5%, p = 0.966). CONCLUSION. Overall sensitivity was only minimally improved by use of the AI system. Significant improvement was achieved, however, in the detection of transition zone lesions with use of the AI system at the cost of a mean of 40 seconds of additional reading time.

Effect of observation size and apparent diffusion coefficient (ADC) value in PI-RADS v2.1 assessment category 4 and 5 observations compared to adverse pathological outcomes

OBJECTIVE

To compare observation size and apparent diffusion coefficient (ADC) values in Prostate Imaging Reporting and Data System (PI-RADS) v2.1 category 4 and 5 observations to adverse pathological features.

MATERIALS AND METHODS

With institutional review board approval, 267 consecutive men with 3-T MRI before radical prostatectomy (RP) between 2012 and 2018 were evaluated by two blinded radiologists who assigned PI-RADS v2.1 scores. Discrepancies were resolved by consensus. A third blinded radiologist measured observation size and ADC (ADC.mean, ADC.min [lowest ADC within an observation], ADC.ratio [ADC.mean/ADC.peripheral zone {PZ}]). Size and ADC were compared to pathological stage and Gleason score (GS) using t tests, ANOVA, Pearson correlation, and receiver operating characteristic (ROC) analysis.

RESULTS

Consensus review identified 267 true positive category 4 and 5 observations representing 83.1% (222/267) PZ and 16.9% (45/267) transition zone (TZ) tumors. Inter-observer agreement for PI-RADS v2.1 scoring was moderate (K = 0.45). Size was associated with extra-prostatic extension (EPE) (19 ± 8 versus 14 ± 6 mm, p < 0.001) and seminal vesicle invasion (SVI) (24 ± 9 versus 16 ± 7 mm, p < 0.001). Size ≥ 15 mm optimized the accuracy for EPE with area under the ROC curve (AUC) and sensitivity/specificity of 0.68 (CI 0.62-0.75) and 63.2%/65.6%. Size ≥ 19 mm optimized the accuracy for SVI with AUC/sensitivity/specificity of 0.75 (CI 0.66-0.83)/69.4%/70.6%. ADC metrics were not associated with pathological stage. Larger observation size (p = 0.032), lower ADC.min (p = 0.010), and lower ADC.ratio (p = 0.010) were associated with higher GS. Size correlated better to higher Gleason scores (p = 0.002) compared to ADC metrics (p = 0.09-0.11).

CONCLUSION

Among PI-RADS v2.1 category 4 and 5 observations, size was associated with higher pathological stage whereas ADC metrics were not. Size, ADC.minimum, and ADC.ratio differed in tumors stratified by Gleason score.

KEY POINTS

• Among PI-RADS category 4 and 5 observations, size but not ADC can differentiate between tumors by pathological stage. • An observation size threshold of 15 mm and 19 mm optimized the accuracy for diagnosis of extra-prostatic extension and seminal vesicle invasion. • Among PI-RADS category 4 and 5 observations, size, ADC.minimum, and ADC.ratio differed comparing tumors by Gleason score.

Assessment of DCE Utility for PCa Diagnosis Using PI-RADS v2.1: Effects on Diagnostic Accuracy and Reproducibility

The role of dynamic contrast-enhanced-MRI (DCE-MRI) for Prostate Imaging-Reporting and Data System (PI-RADS) scoring is a controversial topic. In this retrospective study, we aimed to measure the added value of DCE-MRI in combination with T2-weighted (T2W) and diffusion-weighted imaging (DWI) using PI-RADS v2.1, in terms of reproducibility and diagnostic accuracy, for detection of prostate cancer (PCa) and clinically significant PCa (CS-PCa, for Gleason Score ≥ 7). 117 lesions in 111 patients were identified as suspicion by multiparametric MRI (mpMRI) and addressed for biopsy. Three experienced readers independently assessed PI-RADS score, first using biparametric MRI (bpMRI, including DWI and T2W), and then multiparametric MRI (also including DCE). The inter-rater and inter-method agreement (bpMRI- vs. mpMRI-based scores) were assessed by Cohen's kappa (κ). Receiver operating characteristics (ROC) analysis was performed to evaluate the diagnostic accuracy for PCa and CS-PCa detection among the two scores. Inter-rater agreement was excellent for the three pairs of readers (κ ≥ 0.83), while the inter-method agreement was good (κ ≥ 0.73). Areas under the ROC curve (AUC) showed similar high-values (0.8 ≤ AUC ≤ 0.85). The reproducibility of PI-RADS v2.1 scoring was comparable and high among readers, without relevant differences, depending on the MRI protocol used. The inclusion of DCE did not influence the diagnostic accuracy.