Analysis of challenging mammographic cases demonstrates subtle reader group discrepancies

Area of Science:

• Radiology
• Medical Imaging Analysis
• Mammography Interpretation

Background:

• High-quality mammogram interpretation is crucial for early abnormality detection.
• Understanding challenging image characteristics aids reader education and AI development.
• This study compares challenges faced by radiography advanced practitioners (RAPs) and radiologists.

Purpose of the Study:

• To determine if RAPs and radiologists face similar challenges in mammogram interpretation.
• To identify specific mammographic characteristics that pose difficulties for each group.
• To inform educational strategies and AI tool development.

Main Methods:

• Prospective comparison study of radiographer and radiologist mammography readings.
• A test set of 60 mammograms (20 with cancer) was interpreted using a cloud-based platform.
• Difficulty indices were calculated based on error rates; Mann-Whitney and Spearman correlation analyses were used.

Main Results:

• Strong correlations (r=0.83 and r=0.73) in difficulty indices between RAPs and radiologists for cancer and normal cases.
• Soft tissue appearances and lack of prior images presented greater difficulty than calcifications or availability of prior images.
• No significant image characteristic differences were noted for radiologists alone.

Conclusions:

• A strong correlation exists between radiologists and radiographers in identifying difficult mammographic cases.
• Radiographers showed increased susceptibility to challenges from varied cancer appearances and absent prior images.
• Findings support tailored educational strategies and AI development for reader support.