Segmentation of microcalcifications in mammograms.

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Accurate detection and segmentation of microcalcifications in mammograms are critical for early breast cancer diagnosis.
• Minimizing false positives and false negatives is essential for reliable diagnostic outcomes.
• Existing methods may struggle with preserving the precise size and shape of individual calcifications.

Purpose of the Study:

• To develop and evaluate a systematic, two-stage method for the detection and segmentation of microcalcifications.
• To ensure the preservation of individual calcification size and shape with high fidelity.
• To establish a reproducible segmentation technique for efficient mammography screening programs.

Main Methods:

• A two-stage algorithm combining a weighted difference of Gaussians filter for spot detection and a morphological filter for shape extraction.
• The first stage identifies topology and number of spots, while the second stage defines their shape.
• Conditional thickening operation integrates results from both filters for comprehensive analysis.

Main Results:

• The algorithm demonstrated effective noise-invariant and size-specific detection of microcalcification spots.
• Accurate reproduction of calcification shapes was achieved.
• Testing on real mammograms yielded highly encouraging results when compared to expert radiological assessments.

Conclusions:

• The proposed method offers a reproducible approach to microcalcification segmentation.
• This technique supports the necessary preconditions for an efficient breast cancer screening program.
• The algorithm's ability to preserve calcification morphology is vital for improving diagnostic accuracy.