Lesion type and reader experience affect the diagnostic accuracy of breast MRI: a multiple reader ROC study

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Breast MRI (BMRI) is a valuable tool for breast lesion evaluation.
• Diagnostic performance can be influenced by lesion characteristics and interpreter experience.
• Understanding these factors is crucial for accurate non-screening breast imaging interpretation.

Purpose of the Study:

• To assess how lesion type (mass vs. non-mass) and reader experience affect breast MRI diagnostic performance.
• To evaluate the impact of these factors in a diagnostic, non-screening setting.

Main Methods:

• Retrospective analysis of 259 histologically verified breast lesions from patients undergoing diagnostic BMRI.
• Six blinded readers with varying experience levels (high, intermediate, low) interpreted BMRI examinations.
• Statistical analysis using ROC curves and logistic regression to determine diagnostic accuracy.

Main Results:

• Area under the curve (AUC) ranged from 0.834 (low experience) to 0.935 (high experience).
• Non-mass lesions (P<0.0001) and reader experience (P=0.005) significantly impacted diagnostic performance.
• Malignant lesions were more common in mass-type lesions (103/169) compared to non-mass lesions (19/48).

Conclusions:

• Non-mass lesions are associated with lower diagnostic accuracy in breast MRI.
• Lower reader experience negatively affects breast MRI diagnostic performance.
• Optimizing interpretation strategies for non-mass lesions and enhancing reader training are important for improving diagnostic outcomes.