Towards Interpretable Attention Networks for Cervical Cancer Analysis

Ruiqi Wang,Mohammad Ali Armin,Simon Denman,Lars Petersson,David Ahmedt-Aristizabal

DOI: https://doi.org/10.1109/EMBC46164.2021.9629604

2021-05-27

Abstract:Recent advances in deep learning have enabled the development of automated frameworks for analysing medical images and signals, including analysis of cervical cancer. Many previous works focus on the analysis of isolated cervical cells, or do not offer sufficient methods to explain and understand how the proposed models reach their classification decisions on multi-cell images. Here, we evaluate various state-of-the-art deep learning models and attention-based frameworks for the classification of images of multiple cervical cells. As we aim to provide interpretable deep learning models to address this task, we also compare their explainability through the visualization of their gradients. We demonstrate the importance of using images that contain multiple cells over using isolated single-cell images. We show the effectiveness of the residual channel attention model for extracting important features from a group of cells, and demonstrate this model's efficiency for this classification task. This work highlights the benefits of channel attention mechanisms in analyzing multiple-cell images for potential relations and distributions within a group of cells. It also provides interpretable models to address the classification of cervical cells.

Computer Vision and Pattern Recognition,Machine Learning,Image and Video Processing

What problem does this paper attempt to address?

This paper aims to solve the problem of multi - cell image classification in cervical cancer analysis. Specifically, the paper focuses on how to use deep - learning models, especially the attention mechanism, to improve the accuracy and interpretability of classifying images containing multiple cervical cells. Traditional methods usually focus on the analysis of single cells and ignore the relationship and distribution information between different cells in multi - cell images, which limits the performance and interpretability of the models. In addition, although existing deep - learning models perform well in classification tasks, they often lack sufficient interpretability and transparency, which hinders the application of these models in clinical practice.

To meet these challenges, the paper proposes the following goals:

1. **Develop interpretable deep - learning models**: By introducing the attention mechanism, especially the residual channel attention model, improve the accuracy and interpretability of the model for multi - cell cervical image classification.

2. **Verify the advantages of multi - cell images**: Prove that using images containing multiple cells can capture the relationship and distribution information between cells better than using single - cell images, thereby improving classification performance.

3. **Provide model interpretability**: By visualizing the gradients of the model, show how the attention mechanism works and how it helps the model focus on useful feature areas in the image and ignore background and other noise information.

In summary, the main contribution of this paper lies in exploring and verifying the effectiveness and interpretability of the attention mechanism in multi - cell cervical image classification, providing new ideas and technical means for improving the automation and reliability of cervical cancer analysis.