Multi-Channel Input Deep Convolutional Neural Network for Mammogram Diagnosis
Ji Hoon Bae, Joon Hyeon Park, Jin Hyeok Park, Myung Hoon Sunwoo · 2020
Medical image diagnosis should consider the information contained in multiple images, not just a single image, such as natural image classification. Mammography is the most basic X-ray screening method for diagnosing breast cancer, and mammograms have four images per patient. Convolutional neural networks should be able to diagnose using these four images. This paper proposes a convolutional neural network to simultaneously concatenate four images to solve the multi-view problem. The proposed network was trained and validated with the digital database for screening mammography (DDSM) and achieved 0.952 area under the ROC curve (AUC) for the two-class problem (positive vs. negative). This paper also proposes a new approach to localize lesions without patch labels or mask labels.