Filters








1,450 Hits in 9.5 sec

The Power of Proxy Data and Proxy Networks for Hyper-Parameter Optimization in Medical Image Segmentation [article]

Vishwesh Nath, Dong Yang, Ali Hatamizadeh, Anas A. Abidin, Andriy Myronenko, Holger Roth, Daguang Xu
2021 arXiv   pre-print
In this work, we focus on accelerating the estimation of hyper-parameters by proposing two novel methodologies: proxy data and proxy networks.  ...  Third, we show that the proposed approach of utilizing a proxy network can speed up an AutoML framework for hyper-parameter search by 3.3x, and by 4.4x if proxy data and proxy network are utilized together  ...  To the best of our knowledge, there is no previous work which has proposed proxy data and proxy networks for medical image segmentation.  ... 
arXiv:2107.05471v1 fatcat:vl5hkcju6vgt7hor7s2djxrw2m

Embedding Task Knowledge into 3D Neural Networks via Self-supervised Learning [article]

Jiuwen Zhu, Yuexiang Li, Yifan Hu, S. Kevin Zhou
2020 arXiv   pre-print
Deep learning highly relies on the amount of annotated data. However, annotating medical images is extremely laborious and expensive.  ...  The experimental results demonstrate the effectiveness of embedding lesion-related prior-knowledge into neural networks for 3D medical image classification.  ...  The TCPC with different settings of these two hyper-parameters is evaluated in Fig. 4 . Our TCPC is observed to be insensitive to the hyper-parameters.  ... 
arXiv:2006.05798v1 fatcat:jsf3hq6y6fd6vbu7lqrlgf3fg4

Weakly supervised multiple instance learning histopathological tumor segmentation [article]

Marvin Lerousseau, Maria Vakalopoulou, Marion Classe, Julien Adam, Enzo Battistella, Alexandre Carré, Théo Estienne, Théophraste Henry, Eric Deutsch, Nikos Paragios
2021 arXiv   pre-print
Histopathological image segmentation is a challenging and important topic in medical imaging with tremendous potential impact in clinical practice.  ...  In this paper, we propose a weakly supervised framework for whole slide imaging segmentation that relies on standard clinical annotations, available in most medical systems.  ...  For the rest of the paper this testing set is denoted as "In-distribution". Each selected configuration is trained using the training set, with hyper-parameters optimized on the validation set.  ... 
arXiv:2004.05024v4 fatcat:3wdvmquq4rhzji7yep6guozeue

Improving Auto-Augment via Augmentation-Wise Weight Sharing [article]

Keyu Tian, Chen Lin, Ming Sun, Luping Zhou, Junjie Yan, Wanli Ouyang
2020 arXiv   pre-print
This inspires us to design a powerful and efficient proxy task based on the Augmentation-Wise Weight Sharing (AWS) to form a fast yet accurate evaluation process in an elegant way.  ...  Comprehensive analysis verifies the superiority of this approach in terms of effectiveness and efficiency.  ...  Acknowledgments and Disclosure of Funding Wanli Ouyang is supoorted by the Australian Research Council Grant DP200103223, and Australian Medical Research Future Fund MRFAI000085.  ... 
arXiv:2009.14737v2 fatcat:uxhpy3uxqzhmplqysvn2ieuhca

A Survey of Automated Data Augmentation Algorithms for Deep Learning-based Image Classication Tasks [article]

Zihan Yang, Richard O. Sinnott, James Bailey, Qiuhong Ke
2022 arXiv   pre-print
This paper presents the major works in AutoDA field, discussing their pros and cons, and proposing several potential directions for future improvements.  ...  The goal of AutoDA models is to find the optimal DA policies that can maximize the model performance gains.  ...  Optimization of policy hyper-parameters and network weights are performed in strict order.  ... 
arXiv:2206.06544v1 fatcat:wxub4chlbrhbdpzt7g6ydjwu2u

Imbalance-Aware Self-Supervised Learning for 3D Radiomic Representations [article]

Hongwei Li, Fei-Fei Xue, Krishna Chaitanya, Shengda Luo, Ivan Ezhov, Benedikt Wiestler, Jianguo Zhang, Bjoern Menze
2021 arXiv   pre-print
In this work, we address the challenge of learning representations of 3D medical images for an effective quantification under data imbalance.  ...  Radiomic representations can quantify properties of regions of interest in medical image data.  ...  Within each fold, we employ 20% of training data to optimize the hyper-parameters. Results Quantitative comparison.  ... 
arXiv:2103.04167v2 fatcat:gd4fo5dirvff3mcyl2rq7fpaae

Learning Deformable Image Registration from Optimization: Perspective, Modules, Bilevel Training and Beyond [article]

Risheng Liu, Zi Li, Xin Fan, Chenying Zhao, Hao Huang, Zhongxuan Luo
2021 arXiv   pre-print
and image segmentation.  ...  We also apply our framework to challenging multi-modal image registration, and investigate how our registration to support the down-streaming tasks for medical image analysis including multi-modal fusion  ...  Medical image segmentation is also crucial and highly relevant in medical image analysis.  ... 
arXiv:2004.14557v3 fatcat:22p4ui6zjjh5fkxtpzolln2eqy

Models Genesis [article]

Zongwei Zhou, Vatsal Sodha, Jiaxuan Pang, Michael B. Gotway, Jianming Liang
2020 arXiv   pre-print
Transfer learning from natural image to medical image has been established as one of the most practical paradigms in deep learning for medical image analysis.  ...  of Models Genesis for 3D medical imaging.  ...  Guo in evaluating the performance of Medical-Net (Chen et al., 2019b) ; M. M. Rahman Siddiquee for examining NiftyNet (Gibson et al., 2018b) with our Models Genesis; P.  ... 
arXiv:2004.07882v2 fatcat:iyb2x6y2qjertluhcs6dtjbtfy

Uncertainty-Aware Training of Neural Networks for Selective Medical Image Segmentation

Yukun Ding, Jinglan Liu, Xiaowei Xu, Meiping Huang, Jian Zhuang, Jinjun Xiong, Yiyu Shi
2020 International Conference on Medical Imaging with Deep Learning  
State-of-the-art deep learning based methods have achieved remarkable performance on medical image segmentation.  ...  Existing selective segmentation methods, however, ignore this unique property of selective segmentation and train their DNN models by optimizing accuracy on the entire dataset.  ...  Conclusions In this paper, we develop the first uncertainty-aware training method of neural networks for selective medical image segmentation.  ... 
dblp:conf/midl/DingL0HZXS20 fatcat:g2njbjpexvdonj3fhul22zw7sa

Unsupervised Domain Adaptation for Semantic Segmentation of NIR Images through Generative Latent Search [article]

Prashant Pandey, Aayush Kumar Tyagi, Sameer Ambekar, Prathosh AP
2020 arXiv   pre-print
We propose a method for target-independent segmentation where the 'nearest-clone' of a target image in the source domain is searched and used as a proxy in the segmentation network trained only on the  ...  However, Deep learning based state-of-the-art segmentation techniques demands large amounts of labelled data that is unavailable for the current problem.  ...  as a proxy in the segmentation network trained only on the source data.  ... 
arXiv:2006.08696v2 fatcat:iegnumzphbh7dffg7wynryorri

C2FNAS: Coarse-to-Fine Neural Architecture Search for 3D Medical Image Segmentation [article]

Qihang Yu, Dong Yang, Holger Roth, Yutong Bai, Yixiao Zhang, Alan L. Yuille, Daguang Xu
2020 arXiv   pre-print
some memory and time consuming tasks, such as 3D medical image segmentation.  ...  3D convolution neural networks (CNN) have been proved very successful in parsing organs or tumours in 3D medical images, but it remains sophisticated and time-consuming to choose or design proper 3D networks  ...  Besides, nnU-Net uses different networks and hyper-parameters for each task, while we use the same model and hyper-parameters for all task, showing that our model is not only more powerful but also much  ... 
arXiv:1912.09628v2 fatcat:zedl7ihv5zgqnk5og6rrgybi4m

C2FNAS: Coarse-to-Fine Neural Architecture Search for 3D Medical Image Segmentation

Qihang Yu, Dong Yang, Holger Roth, Yutong Bai, Yixiao Zhang, Alan L. Yuille, Daguang Xu
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
some memory and time-consuming tasks, such as 3D medical image segmentation.  ...  3D convolution neural networks (CNN) have been proved very successful in parsing organs or tumours in 3D medical images, but it remains sophisticated and timeconsuming to choose or design proper 3D networks  ...  Besides, nnU-Net uses different networks and hyper-parameters for each task, while we use the same model and hyper-parameters for all task, showing that our model is not only more powerful but also much  ... 
doi:10.1109/cvpr42600.2020.00418 dblp:conf/cvpr/YuYRBZYX20 fatcat:2aprshth2zdzpk43ejik5vlcjm

Label-Free Segmentation of COVID-19 Lesions in Lung CT

Qingsong Yao, Li Xiao, Peihang Liu, S. Kevin Zhou
2021 IEEE Transactions on Medical Imaging  
To alleviate the burden of data annotation, we herein present a label-free approach for segmenting COVID-19 lesions in CT via voxel-level anomaly modeling that mines out the relevant knowledge from normal  ...  Scarcity of annotated images hampers the building of automated solution for reliable COVID-19 diagnosis and evaluation from CT.  ...  As shown in Table IV , the performance drops rapidly when the HU threshold T = −700. b) Hyper-parameters of 'lesion' generator: For the sensitivity analyses, we change the choice of parameters separately  ... 
doi:10.1109/tmi.2021.3066161 pmid:33760731 fatcat:5sa3whfvbbh2xjcyhu6vxkrtru

Find it if You Can: End-to-End Adversarial Erasing for Weakly-Supervised Semantic Segmentation [article]

Erik Stammes, Tom F.H. Runia, Michael Hofmann, Mohsen Ghafoorian
2020 arXiv   pre-print
In contrast to previous adversarial erasing methods, we optimize two networks with opposing loss functions, which eliminates the requirement of certain suboptimal strategies; for instance, having multiple  ...  Recent advancements in the weakly-supervised setting show that reasonable performance can be obtained by using only image-level labels.  ...  The total loss function for the localizer then becomes: L total = L loc + αL am + βL reg , (8) where α and β are hyper-parameters to tune the importance of the adversarial and regularization losses respectively  ... 
arXiv:2011.04626v1 fatcat:dvxhoxxvpzbnfbihzlt7hcyohq

Maximum Entropy on Erroneous Predictions (MEEP): Improving model calibration for medical image segmentation [article]

Agostina Larrazabal, Cesar Martinez, Jose Dolz, Enzo Ferrante
2021 arXiv   pre-print
We benchmark the proposed strategy in two challenging medical image segmentation tasks: white matter hyperintensity lesions in magnetic resonance images (MRI) of the brain, and atrial segmentation in cardiac  ...  Modern deep neural networks have achieved remarkable progress in medical image segmentation tasks.  ...  Introduction Deep learning-based methods have become the de facto solution for many computer vision and medical imaging tasks, dominating the literature in image segmentation.  ... 
arXiv:2112.12218v1 fatcat:kegmdn6obrb5rfx3jyrhva3vka
« Previous Showing results 1 — 15 out of 1,450 results