복부 CT 영상에서의 신장 및 신장종양에 대한 U-Net 기반의 자동 분할 기법
Automated Segmentation of Kidney and Kidney Tumor inAbdominal CT Images with U-Net Based Models
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 연구에서는 복부 CT 영상으로부터 신장과 신장에서 발생한 종양을 U-Net 기반의 컨볼루션 신경망으로 자동으로 분할하는 방법을 제안한다. 본 논문에서 제안하는 모델은 3차원으로 촬영된 영상의 시멘틱 분할을 위하여 Encoder-Decoder 구조의 컨볼루션 신경망 구조를 활용하며, 여러 방식의 skip connection을 활용하여 구성하고 비교했다. 아울러 의료 영상 분할에서 자주 나타나는 학습 과정 중 클래스 불균형 문제를 해결하기 위한 손실 함수를 제시한다. 신장 및 신장 내의 종양에 대한 학습은 KiTS19 데이터셋을 사용했으며, 회전 변환, elastic deformation 등의 다양한 변환을 활용하여 데이터를 증강했다. 본 논문에서 제안하는 방법은 skip connection과 Dice Loss 및 Focal Loss를 활용함으로써, 이를 도입하기 전과 비교하여 최대 7.28%의 성능 향상을 보였다.
Abstract
This study proposes a method for automatically segmenting kidney and kidney tumors from abdominal CT images using a U-Net-based convolutional neural network. The model proposed in this paper utilizes the convolutional neural network structure of the encoder-decoder structure for the semantic segmentation of 3D images. We built multiple models by introducing various types of skip connections and compared the models' performance. In addition, we present a loss function to solve the class imbalance problem during the learning process, which is common in medical image segmentation. We used the KiTS19 dataset to train kidneys and tumors and augmented the data by utilizing various rotations and elastic deformations. The method proposed in this paper introduces skip connections, Dice Loss, and Focal Loss to show a performance improvement of up to 7.28%.
참고문헌 (23)
- G. S. Lodwick, "Computer-aided analysis of radiographic images", Journal of Chronic Diseases, 19(4), pp.485~496 (1966)
- G. H. Jeong, "Trends in artificial intelligencebased medical image analysis technology", Institute of Information & Communications Technology Planning & Evaluation (IITP)Weekly ICT Trends, (1863), pp.2~13 (2018)
- 황지원, "지방간 질환 자동 진단을 위한 3D 간 CT 영상의 분할 및 강체 정합 기법", 한국차세대컴퓨팅학회 논문지, 18(3), pp.28~39 (2022), 한국차세대컴퓨팅학회[DOI]
- 챠이트라 다야난다, "Skin lesion segmentation based on spatial group convolutions in dermoscopic images", 한국차세대컴퓨팅학회 논문지, 17(5), pp.54~63 (2021), 한국차세대컴퓨팅학회[DOI]
- 이민경, "컨볼루션 신경망의 신뢰도 규제를 이용한 CT 영상에서 악성 난소 종양 판별 방법", 한국차세대컴퓨팅학회 논문지, 16(2), pp.80~87 (2020), 한국차세대컴퓨팅학회[DOI]
- 정원준, "국내 인공지능(AI) 의료기기 현황 및 규제 이슈", Institute of Information & Communications Technology Planning & Evaluation (IITP)Weekly ICT Trends, (1831), pp.2~15 (2018)
- "National Cancer Information Center" (2018)[URL]
- K. Simonyan, "Very deep convolutional networks for large-scale image recognition" (2014)
- K. He, "Deep residual learning for image recognition", Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pp.770~778 (2016)
- J. Long, "Fully convolutional networks for semantic segmentation", Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pp.3431~3440 (2015)
- V. Badrinarayanan, "SegNet : A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation", IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(12), pp.2481~2495 (2017)
- 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. Zhao, "Pyramid Scene Parsing Network", Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pp.6230~6239 (2017)
- L. C. Chen, "Encoder-decoder with atrous separable convolution for semantic image segmentation", Proc. of the European Conference on Computer Vision, 11211, pp.833~851 (2018)
- N. Heller, "The KiTS19 Challenge Data: 300 Kidney Tumor Cases with Clinical Context, CT Semantic Segmentations, and Surgical Outcomes" (2019)
- P. Y. Simard, "Best practices for convolutional neural networks applied to visual document analysis", Proc. of International Conference on Document Analysis and Recognition, pp.958~963 (2003)
- 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)
- G. Huang, "Densely Connected Convolutional Networks", Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, pp.2261~2269 (2017)
- F. Isensee, "An attempt at beating the 3D U-Net" (2019)
- T. Lin, "Focal Loss for Dense Object Detection", Proc. of the IEEE International Conference on Computer Vision, pp.2999~3007 (2017)
- F. Milletari, "V-Net : Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation", Proc. of International Conference on 3D Vision, pp.565~571 (2016)
- D. Muller, "MIScnn: a frame work for medical image segmentation with convolutional neural networks and deep learning", BMC Medical Imaging, 21(12) (2021)
- Xiaoshuai Hou, "Cascaded Semantic Segmentation for Kidney and Tumor", Submissions to the 2019 Kidney Tumor Segmentation Challenge: KiTS19 (2019), University of Minnesota Libraries Publishing[DOI]