Stepwise Feature Fusion: Local Guides Global

Wang, Jinfeng; Huang, Qiming; Tang, Feilong; Meng, Jian; Su, Jionglong; Song, Sifan

doi:10.48550/arxiv.2203.03635

Cited by 14 publications

(42 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…SFFormer-L [24] These transformer-based models have the perceptual field of the entire image and can take advantage of global contextual information These models are not as successful at acquiring local information as CNNs and do not take full advantage of the local information on different scales TransFuse-L [25] U-Net [14] These models take full advantage of global contextual information and enhance the perception of local contextual information These network models do not take into account the balance between global and multi-scale information and there is a large number of repetitive operations U-Net++ [33] ResUNet++ [17] PraNet [34] These two models fully exploit the edge information using the reverse attention mechanism, making the segmentation results appear with fine edges These two network models mainly focus on the edge information of polyps and do not fully utilize the contextual information of different regions CaraNet [32]…”

Section: Methods Strength Weaknessmentioning

confidence: 99%

“…Dice mIoU SFFormer-L [25] 0.9357 0.8905 PraNet [34] 0.898 0.849 TransFuse-L [26] 0.918 0.868 CaraNet [35] 0.918 0.865 ResUNet++ [17] 0.8133 0.793 U-Net++ [37] 0.821 0.722 U-Net [14] 0.818 0.742 PRAPNet 0.942 0.906 Table 4. Evaluation results of different models using the CVC-ClinicDB datase.…”

Section: Methodsmentioning

confidence: 99%

“…SFFormer-L [24] 0.9447 0.8995 PraNet [34] 0.899 0.849 TransFuse-L [25] 0.934 0.886 CaraNet [32] 0.936 0.887 U-Net++ [17] 0.794 0.729 U-Net [14] 0.818 0.746 PRAPNet 0.917 0.873…”

Section: Methods Dice Mioumentioning

confidence: 99%

“…After Vaswani et al [24] proposed transformer architecture in 2017, an increasing number of models based on transformer architecture have been proposed for medical image segmentation with proficient results in recent years [25,26]. However, the transformer loses the location information of the original image when converting the polyp image into a word vector for the analysis.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

PRAPNet: A Parallel Residual Atrous Pyramid Network for Polyp Segmentation

Han

et al. 2022

Sensors

View full text Add to dashboard Cite

In a colonoscopy, accurate computer-aided polyp detection and segmentation can help endoscopists to remove abnormal tissue. This reduces the chance of polyps developing into cancer, which is of great importance. In this paper, we propose a neural network (parallel residual atrous pyramid network or PRAPNet) based on a parallel residual atrous pyramid module for the segmentation of intestinal polyp detection. We made full use of the global contextual information of the different regions by the proposed parallel residual atrous pyramid module. The experimental results showed that our proposed global prior module could effectively achieve better segmentation results in the intestinal polyp segmentation task compared with the previously published results. The mean intersection over union and dice coefficient of the model in the Kvasir-SEG dataset were 90.4% and 94.2%, respectively. The experimental results outperformed the scores achieved by the seven classical segmentation network models (U-Net, U-Net++, ResUNet++, praNet, CaraNet, SFFormer-L, TransFuse-L).

show abstract

Section: Methods Strength Weaknessmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

“…SFFormer-L [24] 0.9447 0.8995 PraNet [34] 0.899 0.849 TransFuse-L [25] 0.934 0.886 CaraNet [32] 0.936 0.887 U-Net++ [17] 0.794 0.729 U-Net [14] 0.818 0.746 PRAPNet 0.917 0.873…”

Section: Methods Dice Mioumentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

PRAPNet: A Parallel Residual Atrous Pyramid Network for Polyp Segmentation

Han

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…To improve on the segmentation of polyps in colonoscopy images, a range of deep learning (DL) -based solutions [8,13,14,17,19,22,28,30,32,37] have been proposed. Such solutions are designed to automatically predict segmentation maps for colonoscopy images, in order to provide assistance to clinicians performing colonoscopy procedures.…”

Section: Introductionmentioning

confidence: 99%

FCN-Transformer Feature Fusion for Polyp Segmentation

Sanderson

Matuszewski

2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Colonoscopy is widely recognised as the gold standard procedure for the early detection of colorectal cancer (CRC). Segmentation is valuable for two significant clinical applications, namely lesion detection and classification, providing means to improve accuracy and robustness. The manual segmentation of polyps in colonoscopy images is time-consuming. As a result, the use of deep learning (DL) for automation of polyp segmentation has become important. However, DL-based solutions can be vulnerable to overfitting and the resulting inability to generalise to images captured by different colonoscopes. Recent transformer-based architectures for semantic segmentation both achieve higher performance and generalise better than alternatives, however typically predict a segmentation map of $$\frac{h}{4}\times \frac{w}{4}$$ h 4 × w 4 spatial dimensions for a $$h\times w$$ h × w input image. To this end, we propose a new architecture for full-size segmentation which leverages the strengths of a transformer in extracting the most important features for segmentation in a primary branch, while compensating for its limitations in full-size prediction with a secondary fully convolutional branch. The resulting features from both branches are then fused for final prediction of a $$h\times w$$ h × w segmentation map. We demonstrate our method’s state-of-the-art performance with respect to the mDice, mIoU, mPrecision, and mRecall metrics, on both the Kvasir-SEG and CVC-ClinicDB dataset benchmarks. Additionally, we train the model on each of these datasets and evaluate on the other to demonstrate its superior generalisation performance.Code available: https://github.com/CVML-UCLan/FCBFormer.

show abstract

Utilizing adaptive deformable convolution and position embedding for colon polyp segmentation with a visual transformer

Sikkandar,

Sundaram,

Alassaf

et al. 2024

Sci Rep

View full text Add to dashboard Cite

Polyp detection is a challenging task in the diagnosis of Colorectal Cancer (CRC), and it demands clinical expertise due to the diverse nature of polyps. The recent years have witnessed the development of automated polyp detection systems to assist the experts in early diagnosis, considerably reducing the time consumption and diagnostic errors. In automated CRC diagnosis, polyp segmentation is an important step which is carried out with deep learning segmentation models. Recently, Vision Transformers (ViT) are slowly replacing these models due to their ability to capture long range dependencies among image patches. However, the existing ViTs for polyp do not harness the inherent self-attention abilities and incorporate complex attention mechanisms. This paper presents Polyp-Vision Transformer (Polyp-ViT), a novel Transformer model based on the conventional Transformer architecture, which is enhanced with adaptive mechanisms for feature extraction and positional embedding. Polyp-ViT is tested on the Kvasir-seg and CVC-Clinic DB Datasets achieving segmentation accuracies of 0.9891 ± 0.01 and 0.9875 ± 0.71 respectively, outperforming state-of-the-art models. Polyp-ViT is a prospective tool for polyp segmentation which can be adapted to other medical image segmentation tasks as well due to its ability to generalize well.

show abstract

Stepwise Feature Fusion: Local Guides Global

Cited by 14 publications

References 18 publications

PRAPNet: A Parallel Residual Atrous Pyramid Network for Polyp Segmentation

PRAPNet: A Parallel Residual Atrous Pyramid Network for Polyp Segmentation

FCN-Transformer Feature Fusion for Polyp Segmentation

Utilizing adaptive deformable convolution and position embedding for colon polyp segmentation with a visual transformer

Contact Info

Product

Resources

About