Inter-reader Agreement for Prostate Cancer Detection Using Micro-ultrasound: A Multi-institutional Study

Area of Science:

• Urology
• Medical Imaging
• Oncology

Background:

• Micro-ultrasound (MUS) offers superior resolution for differentiating prostate cancer from normal tissue.
• Preliminary studies suggest MUS sensitivity is comparable to MRI, but user consistency requires evaluation.

Purpose of the Study:

• To assess the inter-reader agreement in interpreting micro-ultrasound (MUS) images for prostate cancer detection.
• To evaluate the consistency of MUS interpretation across multiple urologists with varying experience levels.

Main Methods:

• Prospective collection of MUS images from 57 patients referred for prostate biopsy.
• Interpretation of MUS images by six urologists from four institutions, blinded to MRI and clinical data.
• Measurement of reader agreement using Light's κ and positive percent agreement (PPA); secondary outcome was sensitivity for grade group (GG) ≥2 prostate cancer.

Main Results:

• Inter-reader agreement (κ) was 0.30, with positive percent agreement (PPA) at 33%.
• Mean sensitivity for GG ≥2 prostate cancer was 0.66, but significantly lower for anterior lesions.
• Higher agreement (κ 0.41, PPA 45%) and detection likelihood (12x) were observed for higher-grade cancers (GG ≥3).

Conclusions:

• Inter-reader agreement for lesion location on MUS is lower than previously reported for MRI.
• MUS demonstrates a lack of sensitivity for detecting prostate cancer in the anterior gland.
• Current MUS performance inconsistencies suggest it may not yet replace MRI for prostate cancer diagnosis.