Academic Journal of Computing & Information Science, 2024, 7(9); doi: 10.25236/AJCIS.2024.070908.
Yucheng Liu1, Shuchang Huang2, Zijun Wang3
1College of Mathematics and Physics, Chengdu University of Technology, Yibin, China
2College of Energy, Chengdu University of Technology, Chengdu, China
3College of Materials and Chemistry & Chemical Engineering, Chengdu University of Technology, Yibin, China
In this study, a new deep learning model-Multi-scale Cascade Segmentation Network (MCSnet)-is proposed for the automatic analysis of lung radiographs and pneumonia detection. MCSnet combines an encoder, an ASPP module, and a decoder to efficiently extract the multiscale semantic information and realize the accurate recognition of lung abnormalities. The experimental results on the qata_v2 dataset show that the MCSnet model has excellent performance, with an average accuracy of 92.81% and an mIoU of 86.33%, which is a significant enhancement compared to the traditional methods. With the multi-scale segmentation technique, the method in this study is able to comprehensively capture the details of lung lesions, providing reliable support for the diagnosis of pneumonia and bringing new possibilities for clinical diagnosis.
Medical Image Segmentation, Deep Learning, Multiscale Features
Yucheng Liu, Shuchang Huang, Zijun Wang. Medical Image Recognition Based on Multiscale Cascade Segmentation Network MCSnet. Academic Journal of Computing & Information Science (2024), Vol. 7, Issue 9: 60-66. https://doi.org/10.25236/AJCIS.2024.070908.
[1] Mei X, Lee H C, Diao K, et al. Artificial intelligence–enabled rapid diagnosis of patients with COVID-19[J]. Nature medicine, 2020, 26(8): 1224-1228.
[2] Shorten C, Khoshgoftaar T M. A survey on image data augmentation for deep learning [J]. Journal of big data, 2019, 6(1): 1-48.
[3] Oliveira G L. Encoder-decoder methods for semantic segmentation: Efficiency and robustness aspects [D]. Albert-Ludwigs-Universität Freiburg, 2019.
[4] Li Y, Li K, Chen C, et al. Modeling temporal patterns with dilated convolutions for time-series forecasting [J]. ACM Transactions on Knowledge Discovery from Data (TKDD), 2021, 16(1): 1-22.
[5] Wang K, Yang J, Yuan S, et al. A lightweight network with attention decoder for real-time semantic segmentation [J]. The Visual Computer, 2022, 38(7): 2329-2339.
[6] Tharwat A. Classification assessment methods [J]. Applied computing and informatics, 2021, 17(1): 168-192.