3D Densely Convolutional Networks for Volumetric Segmentation

Bui, Toan Duc; Shin, Jitae; Moon, Taesup

doi:10.48550/arxiv.1709.03199

Cited by 12 publications

(25 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They reduced the computational and memory costs, which was quite a severe issue for 3D CNN, via small kernels with a deeper network. Bui et al proposed a deep densely convolutional network for volumetric brain segmentation [22]. This architecture provided a dense connection between layers.…”

Section: Related Workmentioning

confidence: 99%

RA-UNet: A Hybrid Deep Attention-Aware Network to Extract Liver and Tumor in CT Scans

Jin

Meng

Sun

et al. 2020

Front. Bioeng. Biotechnol.

265

125

View full text Add to dashboard Cite

Automatic extraction of liver and tumor from CT volumes is a challenging task due to their heterogeneous and diffusive shapes. Recently, 2D deep convolutional neural networks have become popular in medical image segmentation tasks because of the utilization of large labeled datasets to learn hierarchical features. However, few studies investigate 3D networks for liver tumor segmentation. In this paper, we propose a 3D hybrid residual attention-aware segmentation method, i.e., RA-UNet, to precisely extract the liver region and segment tumors from the liver. The proposed network has a basic architecture as U-Net which extracts contextual information combining low-level feature maps with high-level ones. Attention residual modules are integrated so that the attention-aware features change adaptively. This is the first work that an attention residual mechanism is used to segment tumors from 3D medical volumetric images. We evaluated our framework on the public MICCAI 2017 Liver Tumor Segmentation dataset and tested the generalization on the 3DIRCADb dataset. The experiments show that our architecture obtains competitive results.

show abstract

Section: Related Workmentioning

confidence: 99%

RA-UNet: A Hybrid Deep Attention-Aware Network to Extract Liver and Tumor in CT Scans

Jin

Meng

Sun

et al. 2020

Front. Bioeng. Biotechnol.

265

125

View full text Add to dashboard Cite

show abstract

“…The performance of JAS-GAN had further been demonstrated by comparing it with the widely used methods and the state-of-the-art methods. For the segmentations of LA and atrial scars, we compared the segmentation performance of JAS-GAN to the 2D U-Net [26], 3D U-Net [40], 3D DenseNet [41], SegNet [42], the method (MVTT) proposed by yang et al [43] and two methods aiming to tackle the imbalance issue (Tversky loss [44] and surface loss [45]). We also compare the LA segmentation performance of JAS-GAN to the method (MTL) proposed by Chen et al [46].…”

Section: G Analysis Of Joint Discriminative Networkmentioning

confidence: 99%

JAS-GAN: Generative Adversarial Network Based Joint Atrium and Scar Segmentations on Unbalanced Atrial Targets

Yang¹,

Khan

Zhang

et al. 2021

Preprint

View full text Add to dashboard Cite

Automated and accurate segmentations of left atrium (LA) and atrial scars from late gadolinium-enhanced cardiac magnetic resonance (LGE CMR) images are in high demand for quantifying atrial scars. The previous quantification of atrial scars relies on a two-phase segmentation for LA and atrial scars due to their large volume difference (unbalanced atrial targets). In this paper, we propose an inter-cascade generative adversarial network, namely JAS-GAN, to segment the unbalanced atrial targets from LGE CMR images automatically and accurately in an end-to-end way. Firstly, JAS-GAN investigates an adaptive attention cascade to automatically correlate the segmentation tasks of the unbalanced atrial targets. The adaptive attention cascade mainly models the inclusion relationship of the two unbalanced atrial targets, where the estimated LA acts as the attention map to adaptively focus on the small atrial scars roughly. Then, an adversarial regularization is applied to the segmentation tasks of the unbalanced atrial targets for making a consistent optimization. It mainly forces the estimated joint distribution of LA and atrial scars to match the real ones. We evaluated the performance of our JAS-GAN on a 3D LGE CMR dataset with 192 scans. Compared with the state-of-theart methods, our proposed approach yielded better segmentation performance (Average Dice Similarity Coefficient (DSC) values of 0.946 and 0.821 for LA and atrial scars, respectively), which indicated the effectiveness of our proposed approach for segmenting unbalanced atrial targets.

show abstract

“…Deep convolutional neural networks have shown great potential in medical imaging on account of dominance over traditional methods in applications such as segmentation of neuroanatomy (Bui et al, 2017;Moeskops et al, 2016;Zhang et al, 2015;, lesions (Valverde et al, 2017;Brosch et al, 2015;Kamnitsas et al, 2017;, and tumors (Havaei et al, 2017;Pereira et al, 2016;Wachinger et al, 2017) using voxelwise networks (Moeskops et al, 2016;Havaei et al, 2017;Salehi et al, 2017, 3D voxelwise networks Kamnitsas et al, 2017) and Fully Convolutional Networks (FCNs) (C ¸içek et al, 2016;Milletari et al, 2016;Salehi et al, 2017. FCNs have shown better performance while also being faster in training and testing than voxelwise methods (Salehi et al, 2017.…”

Section: Introductionmentioning

confidence: 99%

“…Among these, the densely connected networks, referred to as DenseNets (Huang et al, 2017) and a few of its extensions, such as a 3D version called DenseSeg (Bui et al, 2017) and a fully convolutional two-path edition (FC-DenseNet) (Jegou et al, 2017), have shown promising results in image segmentation tasks (Dolz et al, 2018). For example the DenseSeg showed top performance in the 2017 MICCAI isointense infant brain MRI segmentation (iSeg) grand challenge 1 , which is considered a very difficult image segmentation task for both traditional and deep learning approaches.…”

Section: Introductionmentioning

confidence: 99%

“…In this work, we aimed to further improve image segmentation under these challenging conditions. While the top performing methods in the iSeg challenge relied on the power of DenseNets and used conventional training strategies based on cross-entropy loss function (Bui et al, 2017;Dolz et al, 2018), in this work we focused on the training part and developed new strategies that helped us achieve the best performance currently reported on the iSeg challenge among all first 2 and second round submissions 3 . We built our technique over a deep 3D two-channel fully convolutional DenseNet; and trained it purposefully using our proposed exclusive multi-label multi-class method of training, with exclusively adjusted similarity loss functions on large overlapping 3D image patches.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exclusive Independent Probability Estimation using Deep 3D Fully Convolutional DenseNets: Application to IsoIntense Infant Brain MRI Segmentation

Hashemi,

Prabhu,

Warfield

et al. 2018

Preprint

View full text Add to dashboard Cite

The most recent fast and accurate image segmentation methods are built upon fully convolutional deep neural networks. In particular, densely connected convolutional neural networks (DenseNets) have shown excellent performance in detection and segmentation tasks. In this paper, we propose new deep learning strategies for DenseNets to improve segmenting images with subtle differences in intensity values and features. In particular, we aim to segment brain tissue on infant brain MRI at about 6 months of age where white matter and gray matter of the developing brain show similar T1 and T2 relaxation times, thus appear to have similar intensity values on both T1-and T2-weighted MRI scans. Brain tissue segmentation at this age is, therefore, very challenging. To this end, we propose an exclusive multi-label training strategy to segment the mutually exclusive brain tissues with similarity loss functions that automatically balance the training based on class prevalence. Using our proposed training strategy based on similarity loss functions and patch prediction fusion we decrease the number of parameters in the network, reduce the complexity of the training process focusing the attention on less number of tasks, while mitigating the effects of data imbalance between labels and inaccuracies near patch borders. By taking advantage of these strategies we were able to perform fast image segmentation (less than 90 seconds per 3D volume), using a network with less parameters than many state-of-the-art networks (1.4 million parameters), overcoming issues such as 3D vs 2D training and large vs small patch size selection, while achieving the top performance in segmenting brain tissue among all methods tested in first and second round submissions of the isointense infant brain MRI segmentation (iSeg) challenge according to the official challenge test results. Our proposed strategy improves the training process through balanced training and by reducing its complexity while providing a trained model that works for any size input image, and is fast and more accurate than many state-of-the-art methods.

show abstract

3D Densely Convolutional Networks for Volumetric Segmentation

Cited by 12 publications

References 0 publications

RA-UNet: A Hybrid Deep Attention-Aware Network to Extract Liver and Tumor in CT Scans

RA-UNet: A Hybrid Deep Attention-Aware Network to Extract Liver and Tumor in CT Scans

JAS-GAN: Generative Adversarial Network Based Joint Atrium and Scar Segmentations on Unbalanced Atrial Targets

Exclusive Independent Probability Estimation using Deep 3D Fully Convolutional DenseNets: Application to IsoIntense Infant Brain MRI Segmentation

Contact Info

Product

Resources

About