Small lesion classification on abbreviated breast MRI: training can improve diagnostic performance and inter-reader...

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Abbreviated breast MRI (AB-MRI) offers a faster alternative to full diagnostic protocol MRI (FDP-MRI).
• Assessing the impact of training on AB-MRI interpretation is crucial for its clinical adoption.
• Standardizing interpretation improves diagnostic consistency and patient outcomes.

Purpose of the Study:

• To evaluate if training enhances diagnostic performance and inter-reader agreement for small lesion classification on AB-MRI.
• To compare the diagnostic capabilities of trained AB-MRI interpretation against FDP-MRI.

Main Methods:

• A retrospective analysis of 1165 breast lesions (≤ 2 cm) from 1165 MRI examinations.
• Twelve radiologists were divided into trained and non-trained groups, interpreting AB-MRI twice.
• Diagnostic accuracy and inter-reader agreement were assessed using BI-RADS categories and multirater k analysis, comparing pre- and post-training reads and AB-MRI versus FDP-MRI.

Main Results:

• Post-training, AB-MRI diagnostic accuracy increased from 77.6% to 84.4%, with improved inter-reader agreement (k=0.410 to 0.579).
• The trained group showed significantly better performance than the non-trained group (accuracy 84.4% vs 78.0%; k=0.579 vs 0.461).
• Despite improvements, post-training AB-MRI performance (accuracy 84.4%, k=0.579) remained lower than FDP-MRI (accuracy 92.8%, k=0.602).

Conclusions:

• Structured training significantly improves diagnostic performance and reduces inter-observer variability in AB-MRI for small breast lesions.
• Training is particularly beneficial for junior radiologists, enhancing their interpretation skills.
• While AB-MRI shows promise, FDP-MRI currently offers superior diagnostic accuracy and agreement for small lesion classification.