Análisis de técnicas para segmentación y claificación de imágenes

Abstract

The problem is based on the reduction of computational calculation using techniques to identify masses and microcalcifications in digital mammographic images. Our work used 243 images from the database INbreast, the techniques were developed in C/C++, for masses were used Median filter, global thresholding, Canny and mathematical morphology are used; for microcalcifications, the same techniques were used except global thresholding. The results obtained in masses were AUC 63.34% and in the average computational calculation both segmentation and classification was 2.16s, 14.60% in CPU and 28406.44kb in RAM. Also, for microcalcifications, AUC 89.65% was obtained and in the average computational calculation both segmentation and classification was 9.68s, 14.90% in CPU and 74079.30kb in RAM. Finally, it was concluded that the combination of techniques improves the results obtained, but there aren’t still optimal algorithms that emit with accuracy the detection of breast cancer.