Unlocking Fine-Grained Details with Wavelet-based High-Frequency Enhancement in Transformers (2023-08-25T00:00:00.000000Z)

TL;DR

This paper addresses the local feature deficiency of the Transformer model by carefully re-designing the self-attention map to produce accurate dense prediction in medical images and proposes a multi-scale context enhancement block within skip connections to adaptively model inter-scale dependencies to overcome the semantic gap among stages of the encoder and decoder modules.

Abstract

Medical image segmentation is a critical task that plays a vital role in diagnosis, treatment planning, and disease monitoring. Accurate segmentation of anatomical structures and abnormalities from medical images can aid in the early detection and treatment of various diseases. In this paper, we address the local feature deficiency of the Transformer model by carefully re-designing the self-attention map to produce accurate dense prediction in medical images. To this end, we first apply the wavelet transformation to decompose the input feature map into low-frequency (LF) and high-frequency (HF) subbands. The LF segment is associated with coarse-grained features while the HF components preserve fine-grained features such as texture and edge information. Next, we reformulate the self-attention operation using the efficient Transformer to perform both spatial and context attention on top of the frequency representation. Furthermore, to intensify the importance of the boundary information, we impose an additional attention map by creating a Gaussian pyramid on top of the HF components. Moreover, we propose a multi-scale context enhancement block within skip connections to adaptively model inter-scale dependencies to overcome the semantic gap among stages of the encoder and decoder modules. Throughout comprehensive experiments, we demonstrate the effectiveness of our strategy on multi-organ and skin lesion segmentation benchmarks. The implementation code will be available upon acceptance. \href{https://github.com/mindflow-institue/WaveFormer}{GitHub}.

Authors

Reza Azad

4 papers

A. Kazerouni

2 papers

Ehsan Khodapanah Aghdam

2 papers

TL;DR

Abstract

Authors

References32 items

DermoSegDiff: A Boundary-aware Segmentation Diffusion Model for Skin Lesion Delineation

Implicit Neural Representation in Medical Imaging: A Comparative Survey

Self-supervised Few-shot Learning for Semantic Segmentation: An Annotation-free Approach

Advances in Medical Image Analysis with Vision Transformers: A Comprehensive Review

MISSFormer: An Effective Transformer for 2D Medical Image Segmentation

HiFormer: Hierarchical Multi-scale Representations Using Transformers for Medical Image Segmentation

Wave-ViT: Unifying Wavelet and Transformers for Visual Representation Learning

Beyond Fixation: Dynamic Window Visual Transformer

Anti-Oversmoothing in Deep Vision Transformers via the Fourier Domain Analysis: From Theory to Practice

Contextual Attention Network: Transformer Meets U-Net

Learning to Merge Tokens in Vision Transformers

A-ViT: Adaptive Tokens for Efficient Vision Transformer

FAT-Net: Feature adaptive transformers for automated skin lesion segmentation

LeViT-UNet: Make Faster Encoders with Transformer for Medical Image Segmentation

Transclaw U-Net: Claw U-Net With Transformers for Medical Image Segmentation

PVT v2: Improved baselines with Pyramid Vision Transformer

SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers

Swin-Unet: Unet-like Pure Transformer for Medical Image Segmentation

Multiscale Vision Transformers

Medical Transformer: Gated Axial-Attention for Medical Image Segmentation

TransUNet: Transformers Make Strong Encoders for Medical Image Segmentation

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Multi-level Context Gating of Embedded Collective Knowledge for Medical Image Segmentation

Bi-Directional ConvLSTM U-Net with Densley Connected Convolutions

Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)

Efficient Attention: Attention with Linear Complexities

Attention Gated Networks: Learning to Leverage Salient Regions in Medical Images

Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation

U-Net: Convolutional Networks for Biomedical Image Segmentation

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

Miccai multi-atlas labeling beyond the cranial vault–workshop and challenge

Frequency Enhanced Transformer

Field of Study

Journal Information

Name

Volume