| Summary: | Nowadays, the segmentation process is very important in the detection of microcalcification in mammogram images. However, the microcalcification in breasts is difficult to detect because of the tiny size and subtle nature abnormalities. In this study, the hybrid method of Canny Edge Detection, Otsu Thresholding, and 2D Wavelet Transform was used by exploiting the strength of each respective methods. The hybrid method is an improvisation of the traditional Canny Edge Detection method that requires a manual setting of the threshold. The quality of the edge detection method will be seriously affected if the manual setting of the threshold is not precise. The proposed hybrid method has overcome this problem by determining the high and low thresholds of the Canny algorithm automatically. The hybrid method was tested with 30 mammogram images, whereby the microcalcifications were already confirmed by radiologists. The performance of the segmentation results was measured in terms of accuracy, sensitivity, F-measure, and error rate. All the experimental results generated outstanding results, where the hybrid method produced 97.50% for accuracy, 0.9280 for F-measure, and 0.1375 for error rate. Hence, the proposed hybrid method has proven its reliability to segment the microcalcification efficiently and is relevant to be used in the medical industry for the segmentation of the region of interest in the near future. © 2020 IEEE.
|
|---|