Hybrid segmentation of mass in mammograms using template matching and dynamic programming

Area of Science:

• Medical Imaging
• Computer-Aided Diagnosis
• Image Segmentation

Background:

• Accurate breast mass segmentation is crucial for computer-aided diagnosis (CADx) systems.
• Existing methods may lack robustness in segmenting complex mass features on mammograms.

Purpose of the Study:

• To develop a robust, automated method for breast mass segmentation on mammograms.
• To extract feasible features for enhancing CADx systems.

Main Methods:

• A hybrid approach combining template matching and dynamic programming was developed.
• Template matching initially located mass regions, followed by dynamic programming for optimal contour extraction.
• Performance was evaluated using area and boundary-based similarity measures against manual annotations.

Main Results:

• The proposed hybrid method achieved a mean overlap percentage of 0.727 ± 0.127.
• This significantly outperformed three other segmentation algorithms (P < .001).
• The method demonstrated superior accuracy in mass segmentation compared to existing techniques.

Conclusions:

• A novel hybrid segmentation method using template matching and dynamic programming was successfully developed.
• The proposed method offers improved accuracy for breast mass segmentation on mammograms.
• This technique holds significant potential for enhancing the performance of CADx systems in mammogram interpretation.