The analysis revealed an excellent agreement between bpMRI and mpMRI for prostate volume measurement, with an ICC value of 0.9963 for single measures, indicating near-perfect reliability. This finding suggests that bpMRI provides highly consistent and reproducible prostate volume assessments, comparable to mpMRI.

Fig 2: The distribution of prostate ellipsoid volume measured using both mpMRI and bpMRI is visualized through a combination of violin plots, box plots, and individual data points. Each violin plot illustrates the density of values within the respective group, with wider sections indicating a higher concentration of data. Within each violin, a box plot highlights the interquartile range, with a central line representing the median value and whiskers extending to depict data variability. Outliers appear as individual points beyond the whiskers. Additionally, individual data points are superimposed for clarity, with orange circles representing the first group, purple squares for the second group, and blue dots for the third group.

For PI-RADS scoring, substantial agreement was observed between bpMRI and mpMRI, with an ICC value of 0.9070. However, a slightly higher degree of variability was noted in PI-RADS assignments using bpMRI compared to mpMRI.

Fig 3: The distribution of PI-RADS scores using both mpMRI and bpMRI is visualized through a combination of violin plots, box plots, and individual data points. Each violin plot depicts the density of values within each group, with wider sections indicating a higher concentration of data points. Within each violin, a box plot highlights the interquartile range, with a central line representing the median value and whiskers extending to illustrate the data spread. Outliers appear as individual points beyond the whiskers. Additionally, individual data points are overlaid for enhanced clarity, with orange circles representing the first group, purple squares for the second group, and blue dots for the third group.

In particular, mpMRI was more likely to classify lesions as PI-RADS 3, suggesting a superior ability to detect intermediate-risk lesions.

Fig 4: The inter-rater agreement for prostate ellipsoid volume and PI-RADS scores measured using bpMRI was assessed to evaluate consistency between observers. Agreement was quantified using intraclass correlation coefficients (ICC) for continuous variables such as prostate ellipsoid volume, and Cohen’s kappa statistics for categorical variables like PI-RADS scoring. Higher ICC values indicate strong concordance in prostate volume measurements, while substantial kappa values reflect consistency in PI-RADS classification. These measures provide insight into the reliability of bpMRI in clinical practice and its reproducibility across different radiologists.

This trend may be attributed to the additional information provided by the DCE sequence, which aids in characterizing vascularization patterns associated with malignancy.

In contrast, bpMRI demonstrated strong performance in identifying low-risk prostate cancer but exhibited limitations in detecting more complex disease characteristics, such as periprostatic fat infiltration and lymphadenopathy.

Fig 5: The graph depicts the distribution of PI-RADS scores assigned by Readers 1 and 2 using bpMRI (biparametric MRI). The y-axis displays the PI-RADS categories (ranging from 1 to 5), while the x-axis represents the frequency of cases in each category. Different colors are used to distinguish PI-RADS scores: pink for PI-RADS 1, yellow for PI-RADS 2, green for PI-RADS 3, blue for PI-RADS 4, and orange for PI-RADS 5.