Filters








10,292 Hits in 6.4 sec

Efficient Medical Image Segmentation Based on Knowledge Distillation

Dian Qin, Jia-Jun Bu, Zhe Liu, Xin Shen, Sheng Zhou, Jing-Jun Gu, Zhi-Hua Wang, Lei Wu, Hui-Fen Dai
2021 IEEE Transactions on Medical Imaging  
To deal with this problem, we propose an efficient architecture by distilling knowledge from well-trained medical image segmentation networks to train another lightweight network.  ...  We further devise a novel distillation module tailored for medical image segmentation to transfer semantic region information from teacher to student network.  ...  Only a few researches have studied the efficiency of segmentation for medical imaging problems and utilized the knowledge distillation technology in recent years.  ... 
doi:10.1109/tmi.2021.3098703 pmid:34283713 fatcat:76uzt3bqcratloemi4fq6gmuzy

Graph Flow: Cross-layer Graph Flow Distillation for Dual Efficient Medical Image Segmentation [article]

Wenxuan Zou, Muyi Sun
2022 arXiv   pre-print
To tackle these problems, we propose Graph Flow, a comprehensive knowledge distillation framework, for both network-efficiency and annotation-efficiency medical image segmentation.  ...  Moreover, we demonstrate the effectiveness of our Graph Flow through a novel semi-supervised paradigm for dual efficient medical image segmentation. Our code will be available at Graph Flow.  ...  In the field of image segmentation, most knowledge distillation studies focus on relation-based knowledge. Liu et al.  ... 
arXiv:2203.08667v5 fatcat:s5n5nsqejrf35avrdh4ij7nhkq

ACT-Net: Asymmetric Co-Teacher Network for Semi-supervised Memory-efficient Medical Image Segmentation [article]

Ziyuan Zhao, Andong Zhu, Zeng Zeng, Bharadwaj Veeravalli, Cuntai Guan
2022 arXiv   pre-print
While deep models have shown promising performance in medical image segmentation, they heavily rely on a large amount of well-annotated data, which is difficult to access, especially in clinical practice  ...  Extensive experimental results demonstrate that ACT-Net outperforms other knowledge distillation methods and achieves lossless segmentation performance with 250x fewer parameters.  ...  success with limited labels on medical image segmentation.  ... 
arXiv:2207.01900v1 fatcat:ea33nyqwwjfadcbqzi6m3e37ly

Robust and Efficient Segmentation of Cross-domain Medical Images [article]

Xingqun Qi, Zhuojie Wu, Min Ren, Muyi Sun, Zhenan Sun
2022 arXiv   pre-print
method for robust and efficient segmentation of cross-domain medical images.  ...  Efficient medical image segmentation aims to provide accurate pixel-wise prediction for the medical images with the lightweight implementation framework.  ...  Recently, knowledge distillation-based approaches show better potential for the efficient medical image segmentation tasks comparing with the previous two types.  ... 
arXiv:2207.12995v1 fatcat:guftgzvqavdq5o77zmt4gbsjb4

CoCo DistillNet: a Cross-layer Correlation Distillation Network for Pathological Gastric Cancer Segmentation [article]

Wenxuan Zou, Muyi Sun
2021 arXiv   pre-print
To tackle this problem, we propose CoCo DistillNet, a novel Cross-layer Correlation (CoCo) knowledge distillation network for pathological gastric cancer segmentation.  ...  In recent years, deep convolutional neural networks have made significant advances in pathology image segmentation.  ...  In the next, a few researchers utilized knowledge distillation to deal with the problem of efficiency in medical image analysis. Ho et al.  ... 
arXiv:2108.12173v2 fatcat:ctl6lpmgnnfbveh4aztvbqe2hi

MT-UDA: Towards Unsupervised Cross-modality Medical Image Segmentation with Limited Source Labels [chapter]

Ziyuan Zhao, Kaixin Xu, Shumeng Li, Zeng Zeng, Cuntai Guan
2021 Lecture Notes in Computer Science  
We aim to investigate how to efficiently leverage unlabeled data from the source and target domains with limited source annotations for cross-modality image segmentation.  ...  However, annotating medical images is laborious, expensive, and requires human expertise, which induces the label scarcity problem.  ...  Introduction Deep convolutional neural networks (DCNNs) have obtained promising performance on medical image segmentation tasks [17, 18] , which further promotes the development of automated medical image  ... 
doi:10.1007/978-3-030-87193-2_28 fatcat:cmqrhlvgpvba5kape4qawz5eti

PA-Seg: Learning from Point Annotations for 3D Medical Image Segmentation using Contextual Regularization and Cross Knowledge Distillation [article]

Shuwei Zhai, Guotai Wang, Xiangde Luo, Qiang Yue, Kang Li, Shaoting Zhang
2022 arXiv   pre-print
The success of Convolutional Neural Networks (CNNs) in 3D medical image segmentation relies on massive fully annotated 3D volumes for training that are time-consuming and labor-intensive to acquire.  ...  In this paper, we propose to annotate a segmentation target with only seven points in 3D medical images, and design a two-stage weakly supervised learning framework PA-Seg.  ...  Differently from typical KD methods [22] , [23] that focus on image classification problems and only distill the knowledge from the teacher to the student, our CKD lets two networks distill knowledge  ... 
arXiv:2208.05669v1 fatcat:bzmjzvq4s5c6pcvrokmt6oijrm

MISSU: 3D Medical Image Segmentation via Self-distilling TransUNet [article]

Nan Wang, Shaohui Lin, Xiaoxiao Li, Ke Li, Yunhang Shen, Yue Gao, Lizhuang Ma
2022 arXiv   pre-print
This motivates us to design the efficiently Transformer-based UNet model and study the feasibility of Transformer-based network architectures for medical image segmentation tasks.  ...  To this end, we propose to self-distill a Transformer-based UNet for medical image segmentation, which simultaneously learns global semantic information and local spatial-detailed features.  ...  To our best knowledge, it is unexploited for medical image segmentation, except KD-ResUNet++ [36] .  ... 
arXiv:2206.00902v1 fatcat:earitthqrvgntoxmunjvlj22mi

Experimenting with Knowledge Distillation techniques for performing Brain Tumor Segmentation [article]

Ashwin Nalwade, Jackie Kisa
2021 arXiv   pre-print
Hence, we also want to experiment with Knowledge Distillation techniques.  ...  Our primary focus is on working with different approaches to perform the segmentation of brain tumors in multimodal MRI scans.  ...  , the validation of segmentation tasks in medical imaging.  ... 
arXiv:2105.11486v1 fatcat:hmu36fdsrje4be4dyboi4yhzpq

Unpaired Multi-modal Segmentation via Knowledge Distillation

Qi Dou, Quande Liu, Pheng Ann Heng, Ben Glocker
2020 IEEE Transactions on Medical Imaging  
We have extensively validated our approach on two multi-class segmentation problems: i) cardiac structure segmentation, and ii) abdominal organ segmentation.  ...  To effectively train such a highly compact model, we introduce a novel loss term inspired by knowledge distillation, by explicitly constraining the KL-divergence of our derived prediction distributions  ...  In medical imaging, the potential of the knowledge distillation technique is promising yet relatively under-explored as far as we know. Wang et al.  ... 
doi:10.1109/tmi.2019.2963882 pmid:32012001 fatcat:htw4dwhhsbbcbjnqdbwcvrkaxe

Towards Efficient Instance Segmentation with Hierarchical Distillation

Ziwei Deng, Quan Kong, Tomokazu Murakami
2019 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)  
) framework for structure information distillation on multitask learning based instance segmentation.  ...  In particular, we present channel-wise distillation for the segmentation head to achieve instance-level mask knowledge transfer.  ...  Method We distill knowledge from a heavy and cumbersome teacher network to teach a light but efficient student network.  ... 
doi:10.1109/iccvw.2019.00405 dblp:conf/iccvw/DengKM19 fatcat:ew5flhryjnchnkp7kqehc4mac4

SimCVD: Simple Contrastive Voxel-Wise Representation Distillation for Semi-Supervised Medical Image Segmentation [article]

Chenyu You, Yuan Zhou, Ruihan Zhao, Lawrence Staib, James S. Duncan
2022 arXiv   pre-print
However, most existing learning-based approaches usually suffer from limited manually annotated medical data, which poses a major practical problem for accurate and robust medical image segmentation.  ...  Automated segmentation in medical image analysis is a challenging task that requires a large amount of manually labeled data.  ...  In contrast, considering that medical image semantic segmentation is a structured prediction problem, we present a novel structured knowledge pair-wise distillation, which further use the structural knowledge  ... 
arXiv:2108.06227v4 fatcat:5n3tqaivifbvzpl4z3q3c27vry

Light-weight Deformable Registration using Adversarial Learning with Distilling Knowledge [article]

Minh Q. Tran, Tuong Do, Huy Tran, Erman Tjiputra, Quang D. Tran, Anh Nguyen
2021 arXiv   pre-print
Most recent learning-based methods focus on improving the accuracy by optimizing the non-linear spatial correspondence between the input images.  ...  Deformable registration is a crucial step in many medical procedures such as image-guided surgery and radiation therapy.  ...  Medical image registration is a popular research topic in medical imaging [1] - [6] , [23] - [25] .  ... 
arXiv:2110.01293v1 fatcat:dr4qbylnsbasjbe5gwyuxcwkne

SlimSeg: Slimmable Semantic Segmentation with Boundary Supervision [article]

Danna Xue, Fei Yang, Pei Wang, Luis Herranz, Jinqiu Sun, Yu Zhu, Yanning Zhang
2022 arXiv   pre-print
More specifically, we employ parametrized channel slimming by stepwise downward knowledge distillation during training.  ...  In this paper, we propose a simple but effective slimmable semantic segmentation (SlimSeg) method, which can be executed at different capacities during inference depending on the desired accuracy-efficiency  ...  medical imaging [25, 39] .  ... 
arXiv:2207.06242v1 fatcat:gyph6fjo5ranrclwwkwp7oe3da

WORD: A large scale dataset, benchmark and clinical applicable study for abdominal organ segmentation from CT image [article]

Xiangde Luo, Wenjun Liao, Jianghong Xiao, Tao Song, Xiaofan Zhang, Kang Li, Dimitris N. Metaxas, Guotai Wang, Shaoting Zhang
2022 arXiv   pre-print
Recently, deep learning-based medical image segmentation has shown the potential to reduce manual delineation efforts, but it still requires a large-scale fine annotated dataset for training.  ...  We further evaluate the scribble-based annotation-efficient learning on this dataset, as the pixel-wise manual annotation is time-consuming and expensive.  ...  Recently, knowledge distillation has achieved success in several 2D natural image tasks inspiring us that knowledge distillation may have the potential to handle the 3D medical image segmentation tasks  ... 
arXiv:2111.02403v3 fatcat:c3qcnobmfbhhlj5nq5bgmwkyli
« Previous Showing results 1 — 15 out of 10,292 results