Center for Research and Advanced Studies of the National Polytechnic Institute, ZIP 87130, Ciudad Victoria, Tamaulipas, Mexico.
Center for Research and Advanced Studies of the National Polytechnic Institute, ZIP 87130, Ciudad Victoria, Tamaulipas, Mexico.
Comput Methods Programs Biomed. 2018 Jan;153:33-40. doi: 10.1016/j.cmpb.2017.10.004. Epub 2017 Oct 3.
Conventional computer-aided diagnosis (CAD) systems for breast ultrasound (BUS) are trained to classify pathological classes, that is, benign and malignant. However, from a clinical perspective, this kind of classification does not agree totally with radiologists' diagnoses. Usually, the tumors are assessed by using a BI-RADS (Breast Imaging-Reporting and Data System) category and, accordingly, a recommendation is emitted: annual study for category 2 (benign), six-month follow-up study for category 3 (probably benign), and biopsy for categories 4 and 5 (suspicious of malignancy). Hence, in this paper, a CAD system based on BI-RADS categories weighted by pathological information is presented. The goal is to increase the classification performance by reducing the common class imbalance found in pathological classes as well as to provide outcomes quite similar to radiologists' recommendations.
The BUS dataset considers 781 benign lesions and 347 malignant tumors proven by biopsy. Moreover, every lesion is associated to one BI-RADS category in the set {2, 3, 4, 5}. Thus, the dataset is split into three weighted classes: benign, BI-RADS 2 in benign lesions; probably benign, BI-RADS 3 and 4 in benign lesions; and malignant, BI-RADS 4 and 5 in malignant lesions. Thereafter, a random forest (RF) classifier, denoted by RF, is trained to predict the weighted BI-RADS classes. In addition, for comparison purposes, a RF classifier is trained to predict pathological classes, denoted as RF.
The ability of the classifiers to predict the pathological classes is measured by the area under the ROC curve (AUC), sensitivity (SEN), and specificity (SPE). The RF classifier obtained AUC=0.872,SEN=0.826, and SPE=0.919, whereas the RF classifier reached AUC=0.868,SEN=0.808, and SPE=0.929. According to a one-way analysis of variance test, the RF classifier statistically outperforms (p < 0.001) the RF classifier in terms of the AUC and SEN. Moreover, the classification performance of RF to predict weighted BI-RADS classes is given by the Matthews correlation coefficient that obtained 0.614.
The division of the classification problem into three classes reduces the imbalance between benign and malignant classes; thus, the sensitivity is increased without degrading the specificity. Therefore, the CAD based on weighted BI-RADS classes improves the classification performance of the conventional CAD systems. Additionally, the proposed approach has the advantage of being capable of providing a multiclass outcome related to radiologists' recommendations.
传统的乳腺超声（BUS）计算机辅助诊断（CAD）系统经过训练，可以对病理类别进行分类，即良性和恶性。然而，从临床角度来看，这种分类并不完全符合放射科医生的诊断。通常，肿瘤使用 BI-RADS（乳腺影像报告和数据系统）类别进行评估，并据此发出建议：类别 2（良性）进行年度研究，类别 3（可能良性）进行六个月随访研究，类别 4 和 5（疑似恶性）进行活检。因此，本文提出了一种基于 BI-RADS 类别的 CAD 系统，该系统根据病理信息进行加权。目标是通过减少病理类别中常见的类别不平衡来提高分类性能，并提供与放射科医生建议非常相似的结果。
BUS 数据集包含 781 个良性病变和 347 个经活检证实的恶性肿瘤。此外，每个病变都与集合 {2,3,4,5} 中的一个 BI-RADS 类别相关联。因此，数据集分为三个加权类别：良性，良性病变中的 BI-RADS 2；可能良性，良性病变中的 BI-RADS 3 和 4；恶性，恶性病变中的 BI-RADS 4 和 5。此后，训练一个随机森林（RF）分类器，记为 RF，以预测加权 BI-RADS 类别。此外，为了进行比较，还训练了一个 RF 分类器来预测病理类别，记为 RF。
通过 ROC 曲线下面积（AUC）、灵敏度（SEN）和特异性（SPE）来衡量分类器预测病理类别的能力。RF 分类器获得 AUC=0.872，SEN=0.826，SPE=0.919，而 RF 分类器达到 AUC=0.868，SEN=0.808，SPE=0.929。根据单因素方差分析检验，RF 分类器在 AUC 和 SEN 方面均显著优于（p<0.001）RF 分类器。此外，RF 预测加权 BI-RADS 类别的分类性能由马修斯相关系数给出，其值为 0.614。
将分类问题分为三类可以减少良性和恶性类别之间的不平衡，从而提高灵敏度而不降低特异性。因此，基于加权 BI-RADS 类别的 CAD 提高了传统 CAD 系统的分类性能。此外，该方法的优点是能够提供与放射科医生建议相关的多类别结果。
