English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The Transformer has been successfully used in medical image segmentation dueto its excellent long-range modeling capabilities. However, patch segmentationis necessary when building a Transformer class model. This process may disruptthe tissue structure in medical images, resulting in the loss of relevantinformation. In this study, we proposed a Heterogeneous Swin Transformer withMulti-Receptive Field (HST-MRF) model based on U-shaped networks for medicalimage segmentation. The main purpose is to solve the problem of loss ofstructural information caused by patch segmentation using transformer by fusingpatch information under different receptive fields. The heterogeneous SwinTransformer (HST) is the core module, which achieves the interaction ofmulti-receptive field patch information through heterogeneous attention andpasses it to the next stage for progressive learning. We also designed atwo-stage fusion module, multimodal bilinear pooling (MBP), to assist HST infurther fusing multi-receptive field information and combining low-level andhigh-level semantic information for accurate localization of lesion regions. Inaddition, we developed adaptive patch embedding (APE) and soft channelattention (SCA) modules to retain more valuable information when acquiringpatch embedding and filtering channel features, respectively, thereby improvingmodel segmentation quality. We evaluated HST-MRF on multiple datasets for polypand skin lesion segmentation tasks. Experimental results show that our proposedmethod outperforms state-of-the-art models and can achieve superiorperformance. Furthermore, we verified the effectiveness of each module and thebenefits of multi-receptive field segmentation in reducing the loss ofstructural information through ablation experiments.

HST-MRF: Heterogeneous Swin Transformer with Multi-Receptive Field for Medical Image Segmentation