Fuzzy technique for microcalcifications clustering in digital mammograms

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Breast Cancer Research

Background:

• Mammography is crucial for detecting breast lesions, but microcalcifications are challenging due to their size and low contrast.
• Computer-Aided Detection (CAD) systems can aid in identifying these subtle findings.

Purpose of the Study:

• To develop and evaluate a novel method for enhancing microcalcification clusters in digital mammograms.
• To improve the accuracy and efficiency of microcalcification detection using advanced clustering techniques.

Main Methods:

• A segmentation approach using a form filter derived from the LoG filter was employed.
• A Fuzzy C-means with Features (FCM-WF) clustering algorithm was developed and tested.
• The method was validated on simulated microcalcification clusters and the MIAS database.

Main Results:

• FCM-WF demonstrated superior microcalcification clustering compared to standard FCM, with a 5-10% improvement in Merit Figure.
• The method achieved a 10-22% reduction in false positives.
• High performance metrics were reported: Sensitivity (82-93%), Accuracy (94-95%), and Precision (62-65%).

Conclusions:

• The FCM-WF method accurately clusters microcalcifications, with 70% of injected clusters remaining unaffected in private databases.
• Testing on MIAS databases confirmed the algorithm's segmentation capabilities, with 80% of pathological clusters unaffected.