计算机辅助法计算乳腺密度分类的可行性评估

Feasibility Evaluation of Computer Aided Calculation of Breast Density Classification

目的探讨计算机辅助算法在计算乳腺X线摄影乳腺密度分类的可行性。方法回顾性分析符合纳入标准的400例女性乳腺图像。通过小波融合技术和自适应直方图均衡技术增强乳腺图像对比度特征;然后利用迭代阈值和边缘检测技术分割目标腺体并且计算其乳腺密度;根据乳腺影像报告和数据系统(BI-RADS)分类的规则将乳房腺体组织分为四类。最后根据乳腺密度分类实验结果与医师分类结果进行了比较。结果接触乳腺X线摄影诊断一年的低年资住院医师1的准确率为75%,接触乳腺X线摄影诊断约5年的低年资主治医师2的准确率为85%,计算机辅助分类结果的准确率达到了94%。结论计算机辅助算法计算乳腺密度方法准确率更高,为临床提供一种更可靠的计算乳腺密度的方法。

Objective To investigate the feasibility of computer-aided algorithm in calculating breast density classification of mammography.Methods Retrospective analysis included 400 female standard breast images of the.Breast image contrast characteristics enhanced by wavelet fusion technique and adaptive histogram equalization technique.The target breast is then segmented using the iterative threshold and edge detection techniques and its breast density is calculated.The breast fibroglandular tissue is divided into four categories according to the rules of breast imaging report and data system classification.Finally,according to the results of breast density classification experiment,the accuracy of radiologist classification was compared.Results The accuracy rate of low-grade resident 1 who has been exposed to mammography interpretation for one year was 75%,and the accuracy of a low-grade attendant physician 2 who has been exposed to mammography interpretation for about 5 years was 85%.The accuracy of computer-aided classification results reached 94%.Conclusion Computer-aided algorithms for calculating breast density are more accurate and provide a more reliable method for calculating breast density.