The Computer-Aided Diagnostic (CAD) system is an important tool that helps radiologists to provide a second opinion for the early detection of breast cancer and therefore, aids to reduce the mortality rates. In this work, we try to develop a new (CAD) system to classify mammograms into benign or malignant. The proposed system consists of three main steps. The preprocessing stage consists of noise filtering, elimination of unwanted objects and suppressing the pectoral muscle. The Seeded Region Growing (SRG) segmentation technique is applied in a triangular region that contains the pectoral muscle to localize it and extract the region of interest (ROI). The features extraction step is performed by applying the discrete wavelet transform (DWT) to each obtained ROI, and the most discriminating coefficients are selected using the discrimination power analysis (DPA) method. Finally, the classification is carried out by the support vector machine (SVM), artificial neural networks (ANN), random forest (RF) and Naive Bayes (NB) classifiers. The evaluation of the proposed system on the mini-MIAS database shows its effectiveness compared to other recently published CAD systems, and a classification accuracy of about 99.41% with the SVM classifier was obtained.

中文翻译：

---

利用小波系数和支持向量机的判别力分析乳腺癌分类的新 CAD 系统

计算机辅助诊断 (CAD) 系统是一种重要工具，可帮助放射科医生为早期发现乳腺癌提供第二意见，从而有助于降低死亡率。在这项工作中，我们尝试开发一种新的 (CAD) 系统来将乳房 X 线照片分类为良性或恶性。建议的系统包括三个主要步骤。预处理阶段包括噪声过滤、消除不需要的物体和抑制胸肌。种子区域生长 (SRG) 分割技术应用于包含胸肌的三角形区域，以对其进行定位并提取感兴趣区域 (ROI)。通过对每个获得的 ROI 应用离散小波变换 (DWT) 来执行特征提取步骤，并使用辨别力分析（DPA）方法选择最具辨别力的系数。最后，通过支持向量机（SVM）、人工神经网络（ANN）、随机森林（RF）和朴素贝叶斯（NB）分类器进行分类。该系统在 mini-MIAS 数据库上的评估显示其与其他最近发布的 CAD 系统相比的有效性，并且使用 SVM 分类器获得了约 99.41% 的分类准确度。