Deep convolutional neural networks (CNNs) have been widely used in various medical imaging tasks. However, due to the intrinsic locality of convolution operations, CNNs generally cannot model long-range dependencies well, which are important for accurately identifying or mapping corresponding breast lesion features computed from unregistered multiple mammograms. This motivated us to leverage the architecture of Multi-view Vision Transformers to capture long-range relationships of multiple mammograms from the same patient in one examination. For this purpose, we employed local transformer blocks to separately learn patch relationships within four mammograms acquired from two-view (CC/MLO) of two-side (right/left) breasts. The outputs from different views and sides were concatenated and fed into global transformer blocks, to jointly learn patch relationships between four images representing two different views of the left and right breasts. To evaluate the proposed model, we retrospectively assembled a dataset involving 949 sets of mammograms, which included 470 malignant cases and 479 normal or benign cases. We trained and evaluated the model using a five-fold cross-validation method. Without any arduous preprocessing steps (e.g., optimal window cropping, chest wall or pectoral muscle removal, two-view image registration, etc.), our four-image (two-view-two-side) transformer-based model achieves case classification performance with an area under ROC curve (AUC = 0.818 ± 0.039), which significantly outperforms AUC = 0.784 ± 0.016 achieved by the state-of-the-art multi-view CNNs (p = 0.009). It also outperforms two one-view-two-side models that achieve AUC of 0.724 ± 0.013 (CC view) and 0.769 ± 0.036 (MLO view), respectively. The study demonstrates the potential of using transformers to develop high-performing computer-aided diagnosis schemes that combine four mammograms.

PURPOSE: To develop and test a deep learning based computer-aided diagnosis (CAD) scheme of mammograms for classifying between malignant and benign masses. METHODS: An image dataset involving 560 regions of interest (ROIs) extracted from digital mammograms was used. After down-sampling each ROI from 512×512 to 64×64 pixel size, we applied an 8 layer deep learning network that involves 3 pairs of convolution-max-pooling layers for automatic feature extraction and a multiple layer perceptron (MLP) classifier for feature categorization to process ROIs. The 3 pairs of convolution layers contain 20, 10, and 5 feature maps, respectively. Each convolution layer is connected with a max-pooling layer to improve the feature robustness. The output of the sixth layer is fully connected with a MLP classifier, which is composed of one hidden layer and one logistic regression layer. The network then generates a classification score to predict the likelihood of ROI depicting a malignant mass. A four-fold cross validation method was applied to train and test this deep learning network. RESULTS: The results revealed that this CAD scheme yields an area under the receiver operation characteristic curve (AUC) of 0.696±0.044, 0.802±0.037, 0.836±0.036, and 0.822±0.035 for fold 1 to 4 testing datasets, respectively. The overall AUC of the entire dataset is 0.790±0.019. CONCLUSIONS: This study demonstrates the feasibility of applying a deep learning based CAD scheme to classify between malignant and benign breast masses without a lesion segmentation, image feature computation and selection process.

Background: The accurate classification between malignant and benign breast lesions detected on mammograms is a crucial but difficult challenge for reducing false-positive recall rates and improving the efficacy of breast cancer screening. Objective: This study aims to optimize a new deep transfer learning model by implementing a novel attention mechanism in order to improve the accuracy of breast lesion classification. Methods: ResNet50 is selected as the base model to develop a new deep transfer learning model. To enhance the accuracy of breast lesion classification, we propose adding a convolutional block attention module (CBAM) to the standard ResNet50 model and optimizing a new model for this task. We assembled a large dataset with 4280 mammograms depicting suspicious soft-tissue mass-type lesions. A region of interest (ROI) is extracted from each image based on lesion center. Among them, 2480 and 1800 ROIs depict verified benign and malignant lesions, respectively. The image dataset is randomly split into two subsets with a ratio of 9:1 five times to train and test two ResNet50 models with and without using CBAM. Results: Using the area under ROC curve (AUC) as an evaluation index, the new CBAM-based ResNet50 model yields AUC = 0.866 ± 0.015, which is significantly higher than that obtained by the standard ResNet50 model (AUC = 0.772 ± 0.008) (p < 0.01). Conclusion: This study demonstrates that although deep transfer learning technology attracted broad research interest in medical-imaging informatic fields, adding a new attention mechanism to optimize deep transfer learning models for specific application tasks can play an important role in further improving model performances.