Three-Dimensional Computer-Aided Detection of Microcalcification Clusters in Digital Breast Tomosynthesis

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Digital breast tomosynthesis (DBT) is crucial for breast cancer screening.
• Microcalcification (MC) clusters are key indicators of early breast cancer.
• Accurate detection of MC clusters in DBT images remains challenging.

Purpose of the Study:

• To develop and evaluate a computer-aided detection (CADe) algorithm for microcalcification (MC) clusters in DBT images.
• To improve the accuracy and efficiency of MC cluster detection.
• To reduce false positives in DBT analysis.

Main Methods:

• Proposed a CADe algorithm with prescreening, MC detection, clustering, and false-positive (FP) reduction steps.
• Utilized multiscale Hessian-based 3D objectness and connected-component segmentation for MC seed extraction.
• Employed signal-to-noise ratio (SNR) enhancement for individual MC candidate detection and prescreening.

Main Results:

• The algorithm achieved a sensitivity of 83.3% for MC cluster detection.
• The false-positive (FP) rate was reduced to an average of 2.47 FPs per DBT volume.
• The method successfully prescreened MC-like objects and clustered individual MC candidates.

Conclusions:

• The proposed CADe algorithm demonstrates significant potential for accurate MC cluster detection in DBT.
• The developed method effectively balances sensitivity and specificity, reducing FP rates.
• This algorithm can aid radiologists in improving diagnostic performance for breast cancer screening using DBT.