Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AWGUNET: Attention-Aided Wavelet Guided U-Net for Nuclei Segmentation in Histopathology Images

Ayush Roy, Payel Pramanik, Dmitrii Kaplun, Sergei Antonov, Ram Sarkar

TL;DR

This work tackles nuclei segmentation in histopathology, where uncertain cell boundaries and staining variability hinder automated analysis. It introduces AWGUNET, a U-Net-based framework with a DenseNet-121 backbone augmented by a Wavelet-guided Channel Attention Module (WGCAM) and a learnable weighted Global Average Pooling (lw-GAP), plus a decoder that performs noise-reduced upsampling via Gaussian and Lanczos filtering and multi-kernel feature extraction. Ablation studies and experiments on the MonuSeg and TNBC datasets demonstrate consistent performance gains, with the full model achieving the best Dice and IoU scores, validating the effectiveness of edge-aware attention and the refined decoder. The approach promises improved boundary delineation and robustness to staining variations, potentially enhancing computer-aided histopathology analysis and cancer diagnosis, with available code at GitHub for reproducibility and broader evaluation.

Abstract

Accurate nuclei segmentation in histopathological images is crucial for cancer diagnosis. Automating this process offers valuable support to clinical experts, as manual annotation is time-consuming and prone to human errors. However, automating nuclei segmentation presents challenges due to uncertain cell boundaries, intricate staining, and diverse structures. In this paper, we present a segmentation approach that combines the U-Net architecture with a DenseNet-121 backbone, harnessing the strengths of both to capture comprehensive contextual and spatial information. Our model introduces the Wavelet-guided channel attention module to enhance cell boundary delineation, along with a learnable weighted global attention module for channel-specific attention. The decoder module, composed of an upsample block and convolution block, further refines segmentation in handling staining patterns. The experimental results conducted on two publicly accessible histopathology datasets, namely Monuseg and TNBC, underscore the superiority of our proposed model, demonstrating its potential to advance histopathological image analysis and cancer diagnosis. The code is made available at: https://github.com/AyushRoy2001/AWGUNET.

AWGUNET: Attention-Aided Wavelet Guided U-Net for Nuclei Segmentation in Histopathology Images

TL;DR

This work tackles nuclei segmentation in histopathology, where uncertain cell boundaries and staining variability hinder automated analysis. It introduces AWGUNET, a U-Net-based framework with a DenseNet-121 backbone augmented by a Wavelet-guided Channel Attention Module (WGCAM) and a learnable weighted Global Average Pooling (lw-GAP), plus a decoder that performs noise-reduced upsampling via Gaussian and Lanczos filtering and multi-kernel feature extraction. Ablation studies and experiments on the MonuSeg and TNBC datasets demonstrate consistent performance gains, with the full model achieving the best Dice and IoU scores, validating the effectiveness of edge-aware attention and the refined decoder. The approach promises improved boundary delineation and robustness to staining variations, potentially enhancing computer-aided histopathology analysis and cancer diagnosis, with available code at GitHub for reproducibility and broader evaluation.

Abstract

Accurate nuclei segmentation in histopathological images is crucial for cancer diagnosis. Automating this process offers valuable support to clinical experts, as manual annotation is time-consuming and prone to human errors. However, automating nuclei segmentation presents challenges due to uncertain cell boundaries, intricate staining, and diverse structures. In this paper, we present a segmentation approach that combines the U-Net architecture with a DenseNet-121 backbone, harnessing the strengths of both to capture comprehensive contextual and spatial information. Our model introduces the Wavelet-guided channel attention module to enhance cell boundary delineation, along with a learnable weighted global attention module for channel-specific attention. The decoder module, composed of an upsample block and convolution block, further refines segmentation in handling staining patterns. The experimental results conducted on two publicly accessible histopathology datasets, namely Monuseg and TNBC, underscore the superiority of our proposed model, demonstrating its potential to advance histopathological image analysis and cancer diagnosis. The code is made available at: https://github.com/AyushRoy2001/AWGUNET.
Paper Structure (8 sections, 3 equations, 4 figures, 2 tables)

This paper contains 8 sections, 3 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Wavelet-guided Channel Attention Module
  4. The Decoder Module
  5. Results
  6. Experimental Setup
  7. Ablation study
  8. Conclusion

Figures (4)

  • Figure 1: Our segmentation model reinforces decoder features with edge information-enhanced features using WGCAM.
  • Figure 2: Wavelet-guided Channel Attention Module
  • Figure 3: A block diagram of the Decoder module. The two key components are the upsample and convolution blocks
  • Figure 4: Segmentation results on MonuSeg and TNBC datasets. GT represents ground truth, ELS, BOT, and DLS are the Encoder Last Layer, BOTtleneck, and Decoder Last Layer, respectively. DWT-1, DWT-2, and DWT-3 are the wavelet features of WGCAM between the second, third and fourth encoder and decoder layers, respectively.