Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BCDNet: A Fast Residual Neural Network For Invasive Ductal Carcinoma Detection

Yujia Lin, Aiwei Lian, Mingyu Liao, Shuangjie Yuan

TL;DR

BCDNet is proposed, which firstly upsamples the input image by the residual block and use smaller convolutional block and a special MLP to learn features and is proofed to effectively detect IDC in histopathological RGB images.

Abstract

It is of great significance to diagnose Invasive Ductal Carcinoma (IDC) in early stage, which is the most common subtype of breast cancer. Although the powerful models in the Computer-Aided Diagnosis (CAD) systems provide promising results, it is still difficult to integrate them into other medical devices or use them without sufficient computation resource. In this paper, we propose BCDNet, which firstly upsamples the input image by the residual block and use smaller convolutional block and a special MLP to learn features. BCDNet is proofed to effectively detect IDC in histopathological RGB images with an average accuracy of 91.6% and reduce training consumption effectively compared to ResNet 50 and ViT-B-16.

BCDNet: A Fast Residual Neural Network For Invasive Ductal Carcinoma Detection

TL;DR

BCDNet is proposed, which firstly upsamples the input image by the residual block and use smaller convolutional block and a special MLP to learn features and is proofed to effectively detect IDC in histopathological RGB images.

Abstract

It is of great significance to diagnose Invasive Ductal Carcinoma (IDC) in early stage, which is the most common subtype of breast cancer. Although the powerful models in the Computer-Aided Diagnosis (CAD) systems provide promising results, it is still difficult to integrate them into other medical devices or use them without sufficient computation resource. In this paper, we propose BCDNet, which firstly upsamples the input image by the residual block and use smaller convolutional block and a special MLP to learn features. BCDNet is proofed to effectively detect IDC in histopathological RGB images with an average accuracy of 91.6% and reduce training consumption effectively compared to ResNet 50 and ViT-B-16.
Paper Structure (14 sections, 6 equations, 3 figures, 1 table)

This paper contains 14 sections, 6 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Conventional Diagnosis Methods
  4. Deep Learning-Based Methods
  5. Method
  6. Residual Block
  7. Convolutional Block
  8. Multi-Layer Perceptron (MLP)
  9. Experiments
  10. Dataset
  11. Data Augmentation
  12. Evalution Metrics
  13. Results
  14. Conclusion

Figures (3)

  • Figure 1: The architecture of BCDNet. We combine residual blocks, convolutional blocks, and MLPs to extract features from the input image. The detailed structure of each block is alsoshown in the figure. The blocks are generally composed of convolutional layers, pooling layers, activation layers, normalization layers, and dropout layers in different orders.
  • Figure 2: The accuracy and training time of BCDNet, ResNet 50 and ViT-B-16. Left: The results of the three models on BreaKHis v1 dataset. Right: The results of the three models on IDC regular dataset.
  • Figure 3: The curves of the accuracy and loss of BCDNet, ResNet 50 and ViT-B-16 (from the first row to the third row). Left: The curves of the three models on IDC regular dataset. Right: The curves of the three models on BreaKHis v1 dataset.