基于双侧TIC定量特征的乳腺肿瘤良恶性鉴别

基于DCE-MRI提出了一种利用双侧乳腺对称区域的TIC定量特征识别乳腺肿瘤良恶性的方法.使用三维区域生长算法提取乳腺的病灶区，基于病灶区及其对侧乳腺对应的ROI的TIC曲线分别提取29个特征，并定义双侧差异特征参数，经SFFS方法筛选后得到7个有效特征.使用SVM进行特征训练，基于交叉验证方法得到分类结果.本研究随机选取回顾性病例112例(良性67例，恶性45例)，得到肿瘤良恶性平均分类准确率为88.39%.实验结果表明:此方法对乳腺肿瘤的良恶性鉴别有较高的准确率，对辅助医生进行乳腺病变组织的良恶性鉴别具有重要价值.

Breast Tumor Classification Based on Bilateral TIC Quantitative Features

An effective bilateral quantitative analyzing method for breast tumor classification was proposed based on the time intensity curve(TIC) of dynamic contrast enhanced magnetic resonance imaging(DCE-MRI). The breast lesion region in DCE-MRI images was extracted using three-dimensional region growing algorithm， and 29 features were extracted based on the TIC curves of the ROI corresponding to the lesion area and its contralateral mammary gland. The two-sided difference feature parameters were defined， and 7 effective features after screening by the sequential floating forward selection(SFFS) method were obtained. The support vector machines(SVM) was used for classification， and the classification results were obtained on the basis of cross-validation method for feature training. A hundred and twelve retrospective cases(67 benign， 45 malignant) were chosen randomly， and the average classification accuracy is 88.39%.The experimental results showed that this method has a high accuracy rate for breast tumor classification， and is of great value for assisting doctors in differential diagnosis of breast tumor.