Design And Implementation Of Processing And Analysis In Mammograms Based On OpenCV

Breast cancer is one of the most common malignancy of women, which seriously affect women’s health and even life-threatening. Micro-calcifications are the key characteristic of early breast cancer, but only3%of the information in mammograms can be seen, and early breast micro-calcifications is very small, irregular, difference of shape and distribution, so the information by artificial diagnosis is easily overlooked or misdiagnosed. Therefore, using computer vision technology for efficient diagnosis in mammograms abroad has been a research scholar in the field direction.This paper studies the processing and analysis algorithms in mammograms, and the design and implementation of the medical system. The paper is focused on the following aspects:1) Image preprocessing. Since the original image has a high resolution, the distribution of gray-scale value is too concentrated and the contrasts is low, firstly make the histogram and the gray-scale value equalization and normalized, and use the adaptive median filtering to remove the noise and increase the contrast of the image. Then remove background and extract breast area by using iterative thresholding method in order to greatly reduce the workload and improve efficacy by increasing the accuracy of detection.2) Extraction of micro-calcifications region of interest. According to the medical definition of the micro-calcifications region of interest, identify the possible anomalies in the input of the mammograms and extract the regions of interest. The specific algorithms include:the Laplacian sharpening, the image smooth, the subtraction processing, the method of carpet covering by fractal dimension.3) Detection of micro-calcifications. Improve the traditional LOG algorithm, and present a novel1LOG₂LOG cascade algorithm which combines with morphological filtering. Firstly, enhance the image of micro-calcifications region of interest by using morphological Top-hat method, then use the morphological opening operation to remove the false calcification and get the candidate region of micro-calcifications. Finally, use1LOG₂LOG algorithm to further confirm the candidate micro-calcifications. Through experiments show that, compared to other conventional algorithms, this method not only greatly improve the detection speed, but also is able to extract accurately the location and edge of the micro-calcifications, and this has important application value for the realization of the automatic diagnosis of breast cancer.In this paper, all of the codes of algorithms are based on OpenCV open source code. Since the OpenCV source code entirely open, in the paper design and implementation of the system of processing and analysis in mammograms are all based on OpenCV source code by the Visual C++6.0integrated environment. By the data manipulation of image, dynamic data structures, mathematical morphology operations, structural analysis and the other functions in the OpenCV, realize these functions:the preprocessing of mammograms, extraction of micro-calcifications region of interest and detection the micro-calcifications.The experiments show that compared the results of processing to the marked area in the mammograms of mini-MIAS database data, in this paper the design and development of this system can be better to complete the detection of micro calcifications work in mammograms, and has good robustness and versatility.