PraNet: Parallel Reverse Attention Network for Polyp Segmentation

By Deng-Ping Fan, Ge-Peng Ji, Tao Zhou, Geng Chen, Huan Fu, Jianbing Shen, Ling Shao. International Conference on Medical Image Computing and Computer-Assisted Intervention, 2020

1,833 citations recorded by the source corpus.

Colonoscopy is an effective technique for detecting colorectal polyps, which are highly related to colorectal cancer. In clinical practice, segmenting polyps from colonoscopy images is of great importance since it provides valuable information for diagnosis and surgery. However, accurate polyp segmentation is a challenging task, for two major reasons: (i) the same type of polyps has a diversity of size, color and texture; and (ii) the boundary between a polyp and its surrounding mucosa is not sharp. To address these challenges, we propose a parallel reverse attention network (PraNet) for accurate polyp segmentation in colonoscopy images. Specifically, we first aggregate the features in high-level layers using a parallel partial decoder (PPD). Based on the combined feature, we then generate a global map as the initial guidance area for the following components. In addition, we mine the boundary cues using the reverse attention (RA) module, which is able to establish the relationship between areas and boundary cues. Thanks to the recurrent cooperation mechanism between areas and boundaries, our PraNet is capable of calibrating some misaligned predictions, improving the segmentation accuracy. Quantitative and qualitative evaluations on five challenging datasets across six metrics show that our PraNet improves the segmentation accuracy significantly, and presents a number of advantages in terms of generalizability, and real-time segmentation efficiency (\(\varvec{\sim }\)50 fps).

Subfields: Computer Vision and Pattern Recognition, Oncology

View Paper: https://www.semanticscholar.org/paper/89c6badea0d7bf834d4c069517116dd99c4cc0fd

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Oncology

Citations in the corpus

TransFuse: Fusing Transformers and CNNs for Medical Image Segmentation (2021, methodology): “In polyp segmentation experiments, no data augmentation was used except for multi-scale training, as in [8,11].”
Simultaneously Localize, Segment and Rank the Camouflaged Objects (2021, background): “Further, an effective camouflaged object detection model has potential to be applied in the field of agriculture for insect control, or in medical image segmentation to detect an infection or tumor area [11, 12].”
Inf-Net: Automatic COVID-19 Lung Infection Segmentation From CT Images (2020, methodology): “In the future, we plan to apply our Inf-Net to other related tasks, such as polyp segmentation [84], product defects detection, camouflaged animal detection [85].”
Distribution-Free, Risk-Controlling Prediction Sets (2021, methodology): “We used PraNet [54] as our base segmentation model.”
TransAttUnet: Multi-Level Attention-Guided U-Net With Transformer for Medical Image Segmentation (2021, background): “…include three broad approaches as baselines, including attention-guided models such as Attention U-Net [10], Attention R2U-Net [38], Channel-UNet [8], XLSor [23], FANet [39], and PraNet [40], which have been proposed to improve segmentation performance by incorporating attention mechanisms.”
Camouflaged Object Detection (2020, background): “…in the ﬁelds of computer vision (for search-and-rescue work, or rare species discovery), medical image segmentation ( e.g ., polyp segmentation [14], lung infection segmentation [18,67]), agriculture ( e.g ., locust detection to prevent invasion), and art ( e.g ., for photo-realistic blending…”
Context-aware Cross-level Fusion Network for Camouflaged Object Detection (2021, background): “[Fan et al., 2020] proposed a parallel reverse attention network (PraNet), which first predicts rough regions and then refines the boundaries.”
Camouflaged Object Segmentation with Distraction Mining (2021, methodology): “…RCNN [17], HTC [2], and MSRCNN [22]; shadow detection methods DSC [20] and BDRAR [72]; medical image segmentation methods UNet++ [71] and PraNet [13]; salient object detection methods PiCANet [33], BASNet [44], CPD [55], PFANet [69], EGNet [67], F3Net [53], GC-PANet [5], and MINet-R [40]; and…”
RGB-D salient object detection: A survey (2020, background): “SOD plays a key role in a range of real-world applications, such as stereo matching [11], image understanding [12], co-saliency detection [13], action recognition [14], video detection and segmentation [15]–[18], semantic segmentation [19], [20], medical image segmentation [21]–[23], object tracking”
Cross-level Feature Aggregation Network for Polyp Segmentation (2023, methodology): “Because our model [17] , UNet++ [10] , SFA [11] , PraNet [5] , MSNet [59] , and C2FNet [60] .”
Medical Image Segmentation via Cascaded Attention Decoding (2023, methodology): “Convolutional neural networks (CNNs) have been widely used for medical image segmentation tasks [24, 37, 15, 22, 23, 10].”
Polyp-PVT: Polyp Segmentation with Pyramid Vision Transformers (2021, methodology): “PraNet [5] employs a twostage segmentation approach, adopting a parallel decoder to predict rough regions, and an attention mechanism to restore the edges and internal structure of a polyp for fine-grained segmentation.”
Concealed Object Detection (2021, methodology): “We select 12 deep learning baselines [7], [12], [25], [63], [75], [76], [77], [78], [79], [80], [81], [82] according to the following criteria: a) classical architectures, b) recently published, and c) achieve SOTA performance in a specific field.”
Pattern Recognition (2018, positive): “This setting often makes intra-subject variations larger than inter-subject variations since the gallery view and probe view are likely to be very different.”
UACANet: Uncertainty Augmented Context Attention for Polyp Segmentation (2021, methodology): “Parallel Reverse Attention Network (PraNet) [6] achieved state-of-the-art performance on five different polyp segmentation benchmark by adopting the majority of the network design and techniques from [5].”
DS-TransUNet: Dual Swin Transformer U-Net for Medical Image Segmentation (2021, methodology): “The split and training settings of these datasets are different in [38], [24] and [39], so we conduct experiments according to these three articles respectively.”
MISSFormer: An Effective Transformer for 2D Medical Image Segmentation (2022, background): “…various attention mechanisms were effective to promote segmentation performance, reverse attention [25] was applied to accurate polyp segmentation [26], squeeze-and-excitation attention [27] was integrated into the module to reﬁne the channel information to segment vessels in retina images [28],…”
MISSFormer: An Effective Medical Image Segmentation Transformer (2021, methodology): “2018) was applied to accurate polyp segmentation(Fan et al. 2020), squeeze-and-excitation attention(Hu, Shen, and Sun 2018) was integrated into module to refine the channel information to segment vessel in retinal images(Zhang et al.”