<i>PolypSeg</i>+: A Lightweight Context-Aware Network for Real-Time Polyp Segmentation

Wu, Huisi; Zhen, Zhao; Zhong, Jiafu; Wang, Wei; Wen, Zhenkun; Qin, Jing

doi:10.1109/tcyb.2022.3162873

Cited by 37 publications

(14 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The postprocessing costs extra 2.76 milliseconds, resulting average of 14 FPS. The speed of our approach is close to the fastest [20], while S3 cost additional 1.13 milliseconds achieving 13% more on polyp IoU. Moreover, its inference is based on batch size 1 with single CPU threading for preprocessing and post-processing.…”

Section: Segmentation Evaluation Resultsmentioning

confidence: 79%

“…Based on the best combination, the proposed method (LinkNet+ Densenet121) achieves the high accuracy on both CVC-EndoSceneStill [10] and Kvasir-SEG [11] test sets. The latest research [20] proposed a lightweight PolypSeg+. This model includes an adaptive scale context (ASC) module with a lightweight attention mechanism, and feature pyramid fusion (FPF).…”

Section: Related Workmentioning

confidence: 99%

“…The image resolution ranges from 332 × 487 to 1920 × 1072 pixels. We followed the same strategy in [20] to split the dataset into training, validation and tests for comparison. More specifically, the total of 1000 images is split into 800 training images, 100 validation images and 100 testing images.…”

Section: A Experiments Materialsmentioning

confidence: 99%

See 2 more Smart Citations

Rethinking the Transfer Learning for FCN Based Polyp Segmentation in Colonoscopy

et al. 2023

View full text Add to dashboard Cite

Besides the complex nature of colonoscopy frames with intrinsic frame formation artefacts such as light reflections and the diversity of polyp types/shapes, the publicly available polyp segmentation training datasets are limited, small and imbalanced. In this case, the automated polyp segmentation using a deep neural network remains an open challenge due to the overfitting of training on small datasets. We proposed a simple yet effective polyp segmentation pipeline that couples the segmentation (FCN) and classification (CNN) tasks. We find the effectiveness of interactive weight transfer between dense and coarse vision tasks that mitigates the overfitting in learning. This motivates us to design a new training scheme within our segmentation pipeline. Our method is evaluated on CVC-EndoSceneStill and Kvasir-SEG datasets. It achieves 4.34% and 5.70% Polyp-IoU improvements compared to the state-of-the-art methods on the EndoSceneStill and Kvasir-SEG datasets, respectively and achieves real-time performance in inference. The model and code are available at: https://github.com/MELSunny/Keras-FCN INDEX TERMS Colonoscopy, real-time polyp segmentation, transfer learning, convolutional neural network

show abstract

Section: Segmentation Evaluation Resultsmentioning

confidence: 79%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Rethinking the Transfer Learning for FCN Based Polyp Segmentation in Colonoscopy

et al. 2023

View full text Add to dashboard Cite

show abstract

“…PolypSeg+ [16] As shown in Fig. 9, mDice score gradually increase and eventually converges to a certain value about 0.947 and loss score decrease and eventually converges to a certain value about 0.481.…”

Section: Resultsmentioning

confidence: 83%

“…Although the cross-entropy loss function is often used in semantic segmentation tasks, the polyp region is relatively smaller to the background region and has a serious category imbalance problem during the experiment [16] .…”

Section: Loss Function Design Lossmentioning

confidence: 99%

The application research of deep neural networks in colonic polyp segmentation

Wang

Yang

et al. 2022

Preprint

View full text Add to dashboard Cite

Background: In the study of medical images segmentation, U-Net is a good choice with better performance. But it will cause some problems such as gradient problems or information loss. Methods: In this paper, we introduce a new model based on such an encoder-decoder structure. We mixed a full connection and encoder output in per layer and we add an axial attention to solve gradient problem and keep the connection with pixel from long distance. After pooling, we add a module which has three atrous Convolution processes with different expansion rate and a non-local Self attention to increase the feeling field and fix the problem of losing spatial information. In decoder part, we put a dual channel gate to point out 'what' and 'where' feature we need by the union of Channel Attention and Spatial Attention. Results: We use 4 public dataSets are CVC-ClinicDB, Kvasir-SEG, CVC-ColonDB and CVC-T. We compared more than 10 groups other different models with these 4 group dataSets and marked some evaluating indicators such as mIou, mDice and so on. We also did some ablation experiments to verify that our model structure is reasonable and effective. Conclusions: The comparative training of 12 models in 4 different datasets verifies the feasibility and the good effectiveness of the proposed ideas. We also did some ablation experiments to verify that our model structure is reasonable and effective. The basic structure of the encoder decoder in semantic segmentation has been greatly improved.

show abstract

CafeNet: A Novel Multi‐Scale Context Aggregation and Multi‐Level Foreground Enhancement Network for Polyp Segmentation

Ji,

Li,

Wang

et al. 2024

Int J Imaging Syst Tech

View full text Add to dashboard Cite

The detection of polyps plays a significant role in colonoscopy examinations, cancer diagnosis, and early patient treatment. However, due to the diversity in the size, color, and shape of polyps, as well as the presence of low image contrast with the surrounding mucosa and fuzzy boundaries, precise polyp segmentation remains a challenging task. Furthermore, this task requires excellent real‐time performance to promptly and efficiently present predictive results to doctors during colonoscopy examinations. To address these challenges, a novel neural network, called CafeNet, is proposed in this paper for rapid and accurate polyp segmentation. CafeNet utilizes newly designed multi‐scale context aggregation (MCA) modules to adapt to the extensive variations in polyp morphology, covering small to large polyps by fusing simplified global contextual information and local information at different scales. Additionally, the proposed network utilizes newly designed multi‐level foreground enhancement (MFE) modules to compute and extract differential features between adjacent layers and uses the prediction output from the adjacent lower‐layer decoder as a guidance map to enhance the polyp information extracted by the upper‐layer encoder, thereby improving the contrast between polyps and the background. The polyp segmentation performance of the proposed CafeNet network is evaluated on five benchmark public datasets using six evaluation metrics. Experimental results indicate that CafeNet outperforms the state‐of‐the‐art networks, while also exhibiting the least parameter count along with excellent real‐time operational speed.

show abstract

PolypSeg+: A Lightweight Context-Aware Network for Real-Time Polyp Segmentation

Cited by 37 publications

References 38 publications

Rethinking the Transfer Learning for FCN Based Polyp Segmentation in Colonoscopy

Rethinking the Transfer Learning for FCN Based Polyp Segmentation in Colonoscopy

The application research of deep neural networks in colonic polyp segmentation

CafeNet: A Novel Multi‐Scale Context Aggregation and Multi‐Level Foreground Enhancement Network for Polyp Segmentation

Contact Info

Product

Resources

About