DCAN: Deep Contour-Aware Networks for Accurate Gland Segmentation

Chen, Hao; Qi, Xiaojuan; Yu, Lequan; Heng, Pheng‐Ann

doi:10.1109/cvpr.2016.273

Cited by 476 publications

(328 citation statements)

References 35 publications

Supporting

Mentioning

326

Contrasting

Order By: Relevance

“…No obvious improvement was observed for Core and Enhancing tasks. We further performed a comparison by replacing the combination stage with the threshold-based fusion method in [15]. This resulted in Dice dropping by 15% for the Complete tumor task (from 88 to 75), which indicates the combination stage was beneficial.…”

Section: Results On Brats15 Datasetmentioning

confidence: 99%

“…x n,i is the i-th voxel in the n-th image used for training, and P t refers to the predicted probability of the voxel x n,i belonging to class l t . Similarly to [15], we extract boundaries from radiologists' region annotations and dilate them with a disk filter.…”

Section: Variant Of Fcnmentioning

confidence: 99%

“…All these methods operated at the patch level. Fully convolutional networks (FCNs) recently achieved promising results for natural image segmentation [11,12] as well as medical image segmentation [13][14][15]. In Fig.…”

Section: Introductionmentioning

confidence: 99%

“…Our model is a variant of [14][15][16]. Instead of using 3 auxiliary classifiers for each upsampling path for regularization as in [14], we extract multi-level contextual information by concatenating features from each upsampling path before the classification layer.…”

Section: Introductionmentioning

confidence: 99%

“…This also differs from [16] which performed only one upsampling in the region task. Instead of either applying threshold-based fusion [15] or a deep fusion stage based on a pooling-upsampling FCN [16] to help separate glands, we designed a simple combination stage consisting of three convolutional layers without pooling, aiming at improving tumour boundary segmentation accuracy. Moreover, our network enables multi-task joint training while [16] has to train different tasks separately, followed by a fine-tuning of the entire network.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Boundary-Aware Fully Convolutional Network for Brain Tumor Segmentation

Shen

Wang

Zhang

et al. 2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. We propose a novel, multi-task, fully convolutional network (FCN) architecture for automatic segmentation of brain tumor. This network extracts multi-level contextual information by concatenating hierarchical feature representations extracted from multimodal MR images along with their symmetric-difference images. It achieves improved segmentation performance by incorporating boundary information directly into the loss function. The proposed method was evaluated on the BRATS13 and BRATS15 datasets and compared with competing methods on the BRATS13 testing set. Segmented tumor boundaries obtained were better than those obtained by single-task FCN and by FCN with CRF. The method is among the most accurate available and has relatively low computational cost at test time.

show abstract

Section: Results On Brats15 Datasetmentioning

confidence: 99%

Section: Variant Of Fcnmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Boundary-Aware Fully Convolutional Network for Brain Tumor Segmentation

Shen

Wang

Zhang

et al. 2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

Metastasis detection from whole slide images using local features and random forests

et al. 2017

View full text Add to dashboard Cite

Digital pathology has led to a demand for automated detection of regions of interest, such as cancerous tissue, from scanned whole slide images. With accurate methods using image analysis and machine learning, significant speed-up, and savings in costs through increased throughput in histological assessment could be achieved. This article describes a machine learning approach for detection of cancerous tissue from scanned whole slide images. Our method is based on feature engineering and supervised learning with a random forest model. The features extracted from the whole slide images include several local descriptors related to image texture, spatial structure, and distribution of nuclei. The method was evaluated in breast cancer metastasis detection from lymph node samples. Our results show that the method detects metastatic areas with high accuracy (AUC = 0.97-0.98 for tumor detection within whole image area, AUC = 0.84-0.91 for tumor vs. normal tissue detection) and that the method generalizes well for images from more than one laboratory. Further, the method outputs an interpretable classification model, enabling the linking of individual features to differences between tissue types. © 2017 International Society for Advancement of Cytometry.

show abstract

Automated Grading of Lumbar Disc Degeneration Using a Push‐Pull Regularization Network Based on MRI

Gao

Shui

Zhang

et al. 2020

Magnetic Resonance Imaging

View full text Add to dashboard Cite

Background Lower back pain is one of the most widely experienced health problems and is strongly associated with intervertebral disc (IVD) degeneration. The quantification of the degeneration is valuable for estimation of the material properties of the IVDs and then to perform biomechanical simulation of the spinal conditions and treatments. The MR image characteristics of relatively high intraclass variability and small interclass differences pose challenges for the classification algorithm to perform automatic grading of degenerated IVD. Purpose To assess the feasibility and improvement of a computer‐assisted IVD degeneration grading method based on proposed push‐pull regularization (PPR) strategy. Study Type Retrospective. Population In total, 500 subjects (350 for training, 70 for validation, and 80 for test in a 10‐time 10‐fold cross validation setting) with varied lumbar disorders were included. Field Strength/Sequence 3.0T; T2‐weighted spin echo sequence. Assessment IVD degeneration grading was taken as a classification task of five classes according to the Pfirrmann grading system in this study. The classification results of deep‐learning models with and without PPR were compared with the classifications made by three experienced spinal radiologists. Statistical Tests Paired t‐tests. Results The classification results show that in four classical CNN models of VGG‐M, VGG‐16, GoogleNet, and ResNet‐34, by embedding a PPR strategy, the accuracies of grade II and III IVD classification were improved by more than 10% (P < 0.05), and that the overall accuracy (grades I to V) was improved by over 8% (P < 0.05). Data Conclusion The embedded PPR significantly improved the classification performance, which enhanced CNN representation capability for IVD degeneration grading. Level of Evidence 3 Technical Efficacy Stage 1

show abstract

DCAN: Deep Contour-Aware Networks for Accurate Gland Segmentation

Cited by 476 publications

References 35 publications

Boundary-Aware Fully Convolutional Network for Brain Tumor Segmentation

Boundary-Aware Fully Convolutional Network for Brain Tumor Segmentation

Metastasis detection from whole slide images using local features and random forests

Automated Grading of Lumbar Disc Degeneration Using a Push‐Pull Regularization Network Based on MRI

Contact Info

Product

Resources

About