딥러닝을 활용한 MRI 영상 내 뇌종양 영역 분할 방법 연구
A Study on Brain Tumor Area Segmentation Method in MRI Images Using Deep Learning
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
뇌종양은 치사율이 높은 질병으로 일반적으로 전문의의 MRI 판독을 통하여 진단된다. 그러나, 뇌종양의 크기, 위치가 다양하기 때문에, 전문의가 수행하는 MRI 판독은 많은 시간과 비용이 소요될 수 밖에 없다. 이를 해결하기 위해 MRI 영상에서 뇌종양 영역에 대한 자동 분할에 대한 요구가 증가하고 있으며, 최근 딥러닝 기술이 발달함에 따라, 딥러닝 기법을 활용한 뇌종양 분할 연구가 다양하게 진행되고 있다. 이에, 본 논문에서는 3D U-Net 기반의 변형된 모델을 통해 MRI 영상에서 보다 효과적으로 뇌종양을 검출하는 방법을 제시한다. 제안하는 모델은 기존의 3D U-Net에서 depthwise-Convolution을 적용하여 연산의 효율적을 높였으며, 잔차 블록을 사용하여 네트워크의 학습을 향상시켰다. 그 결과, 뇌종양 영역 WT, TC, ET Dice 점수는 각 90.08%, 86.95%, 79.48%로 기존의 3D U-Net보다 나은 성능을 보였다.
Abstract
Brain tumors are a lethal disease that is generally diagnosed through MRI readings by specialists. However, due to the diverse sizes and locations of brain tumors, MRI readings performed by specialists inevitably consume a great deal of time and cost. To resolve this, there is an increasing demand for automatic segmentation of brain tumor areas in MRI images, and with the recent development of deep learning technology, various studies on brain tumor segmentation using deep learning techniques are being conducted. In this paper, a modified 3D U-Net-based model is proposed to detect brain tumors in MRI images. The proposed model enhances computational efficiency by applying depthwise convolution in the traditional 3D U-Net and improves network training using residual blocks. As a result, the brain tumor area WT, TC, ET Dice scores were 90.08%, 86.95%, and 79.48% respectively, demonstrating better performance than the conventional 3D U-Net
참고문헌 (15)
- Kocarnik, J. M., "Cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life years for 29 cancer groups from 2010 to 2019", JAMA oncology, 8, pp.420~444 (2022)
- A.Mustaqeem, "An efficient brain tumor detection algorithm using watershed & thresholding based segmentation", International Journal of Image, Graphics and Signal Processing, 4(10), pp.34~39 (2012)
- T.Węgliński, "Brain tumor segmentation from MRI data sets using region growing approach", Perspective Technologies and Methods in MEMS Design, pp.185~188 (2011)
- X. Zhao, "A deep learning model integrating FCNNs and CRFs for brain tumor segmentation", Medical image analysis, 43, pp.98~111 (2018)
- L.Chen, "MRI tumor segmentation with densely connected 3D CNN", Medical Imaging 2018: Image Processing, 10574, pp.357~364 (2018)
- J.Long, "Fully convolutional networks for semantic segmentation", Proceedings of the IEEE conference on computer vision and pattern recognition, pp.3431~3440 (2015)
- O. Ronneberger, "U-net:Convolutional networks for biomedical image segmentation", Proc. of International Conference on Medical Image Computing and Computer Assisted Intervention, 9351, pp.234~241 (2015)
- H.Dong, "Automatic brain tumor detection and segmentation using U-Net based fully convolutional networks", MIUA 2017, 21, pp.506~517 (2017)
- O. Cicek, "3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation", Proc. of International Conference on Medical Image Computing and Computer-Assisted Intervention, 9901, pp.424~432 (2016)
- Z.Zhu, "Brain tumor segmentation based on the fusion of deep semantics and edge information in multimodal MRI", Information Fusion, 91, pp.376~387 (2023)
- W.Wu, "Inner Cascaded U²-Net: An Improvement to Plain Cascaded U-Net", CMES-Computer Modeling in Engineering & Sciences, 134(2), pp.1323~1335 (2023)
- F.Isensee, "nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation", Nature methods, 18(2) (2021)
- O.Oktay, "Attention u-net: Learning where to look for the pancreas" (2018)