基于乳腺X线图像的微钙化点区域自动检测算法研究

Research on Algorithm of Automatic Detection of Region of Microcalcification Based on Mammograph

目的:乳腺癌是女性最常见和多发的恶性肿瘤之一。恶性乳腺肿瘤早期重要表征是伴有微钙化现象。人工识别乳腺X射线图像微钙化点和微小肿块等病变遗漏率较高。探讨乳腺X线影像中微钙化点区域的自动提取算法,以提高诊断速度和准确率。方法:在综合运用各种算法的基础上,基于乳腺X线图像微钙化点特征提出应用小波变换等研究乳腺X影像中微钙化点感兴趣区域的自动提取方法,并仿真验证算法的有效性。结果:应用ROC模型对医院临床80个病例的150副图像对提出的小波改进算法和相关算法进行了测试和比较。基于小波的微钙化点区域自动提取算法,可以获得高达98.8%的检出率和5%的误检率,实现了乳腺图像中含钙化点感兴趣区域的自动提取,验证了算法对提高检出率的可行性。结论:应用小波改进算法对乳腺X线图像中含钙化点感兴趣区域的自动提取是可行的,具有较高的微钙化点检出率,有利于实现乳腺影像可疑病灶区域的自动定位和早期发现,有利于加强对导致乳腺癌危险因素的研究和积极干预。

Objeetive： Breast cancer is the most common and multiple malignancies of the women.Important characterization of early malignant breast tumors associated with microcalcifications phenomenon. Artificial to identifing microcalcifications and tiny lumps lesions of breast X-ray image with omission rate higher. In order to improve diagnostic speed and accuracy, automatic extraction algorithm to explore the area of microcalcifications in the breast X-ray images. Methods： Comprehensive use of various algorithms based on the characteristics of the breast X-ray image microcalcifications based application of wavelet transform the automatic extraction study of breast X microcalcifications regions of interest in the image, and verify the validity of the algorithm with simulation, Results： Used ROC model tested and improved wavelet algorithm and related algorithms proposed with 150 images of 80 cases of hospital clinical. Automatic extraction algorithm can be as high as 98.8% detection rate and false detection rate of 5% based on wavelet microcalcifications area, calcium-point region of interest in the breast image automatically extracted, validated algorithm to improve the detectionrate of viability. Conclusions： Application of wavelet improved algorithm of automatic extraction of calcium-based point region of interest in the breast X-ray images is feasible, with higher microcalcifications detection rate, conducive to the realization of the automatic positioning of the area of breast imaging suspicious lesions and early detection,conducive to strengthening research and active intervention lead to breast cancer risk factors.