Reader characteristics and mammogram features associated with breast imaging reporting scores

Area of Science:

• Radiology
• Medical Imaging
• Breast Cancer Detection

Background:

• Mammography is crucial for breast cancer screening.
• Accurate interpretation of mammograms relies on standardized reporting systems like Tabar Breast Imaging Reporting and Data System (BIRADS).
• Understanding radiologist performance and factors influencing their BIRADS classifications is essential for improving diagnostic accuracy.

Purpose of the Study:

• To evaluate radiologist performance in detecting breast cancer using the Tabar BIRADS classification.
• To identify factors associated with radiologists' reporting scores on mammograms.

Main Methods:

• 117 mammogram readings by Australian radiologists using the Tabar BIRADS lexicon (categories 1-5).
• Calculation of performance metrics (TP, FP, TN, FN) and analysis of reporting category distribution.
• Statistical examination (Mann-Whitney U, Kruskal-Wallis tests) of reader characteristics and case features influencing category assignment.

Main Results:

• Most cancer cases were reported using category 3 (38%), followed by category 4 (32.3%) and category 5 (16.2%).
• Female readers and those with BreastScreen or fellowship experience showed lower false-negative rates.
• Radiologists reading fewer cases per week were more likely to assign category 4, while high-volume readers favored category 3. Discrete masses and asymmetric densities were often reported as category 3, while spiculated masses received category 5.

Conclusions:

• Category 3 was the most common score for cancer mammograms among the participating radiologists.
• Radiologist gender, BreastScreen employment, fellowship completion, and weekly case volume significantly influenced BIRADS category selection.
• Certain mammographic abnormalities are more frequently assigned higher Tabar BIRADS scores, indicating a need for tailored training to enhance recognition of subtle findings.