Mohammed A. Al-masni1, Woo-Ram Kim2, Eung Yeop Kim3, Young Noh4, and Dong-Hyun Kim1
1Department of Electrical and Electronic Engineering, College of Engineering, Yonsei University, Seoul, Korea, Republic of, 2Neuroscience Research Institute, Gachon University, Incheon, Korea, Republic of, 3Department of Radiology, Gachon University College of Medicine, Gachon University, Incheon, Korea, Republic of, 4Department of Neurology, Gachon University College of Medicine, Gachon University, Incheon, Korea, Republic of
1Department of Electrical and Electronic Engineering, College of Engineering, Yonsei University, Seoul, Korea, Republic of, 2Neuroscience Research Institute, Gachon University, Incheon, Korea, Republic of, 3Department of Radiology, Gachon University College of Medicine, Gachon University, Incheon, Korea, Republic of, 4Department of Neurology, Gachon University College of Medicine, Gachon University, Incheon, Korea, Republic of
This study proposes a 3D multi-scale ResNet for lacunes detection. The network conducts multiple parallel paths using different input scales from two image modalities. This enables to learn global features of the lacunes’ anatomical location and hence achieve better detection performance.
Figure 1. A detailed structure of the proposed 3D
multi-scale ResNet.
Figure 4. Some exemplar results
of the proposed 3D multi-scale ResNet. The drawn orange and blue circles refer
to the truly detected lacune and non-lacune lesions, respectively.
