Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Simple 2D Convolutional Neural Network-based Approach for COVID-19 Detection

Chih-Chung Hsu, Chia-Ming Lee, Yang Fan Chiang, Yi-Shiuan Chou, Chih-Yu Jiang, Shen-Chieh Tai, Chi-Han Tsai

TL;DR

The paper tackles COVID-19 detection from CT scans with variable slice counts by introducing SSFL++, a morphology-based, unsupervised RoI and spatial-slice filtering method, and Kernel-Density-based Slice Sampling (KDS) to maintain global sequence information during training. These components are integrated with a simple 2D EfficientNet backbone (E2D), enabling robust performance even with limited data. The approach achieves high macro F1-scores and demonstrates data efficiency, reducing redundancy while preserving diagnostic information, as validated on the COVID-19-CT-DB dataset from the DEF-AI-MIA workshop. The work offers practical improvements for explainable, efficient CT-based COVID-19 detection and suggests strong potential for few-shot and cross-device generalization.

Abstract

This study explores the use of deep learning techniques for analyzing lung Computed Tomography (CT) images. Classic deep learning approaches face challenges with varying slice counts and resolutions in CT images, a diversity arising from the utilization of assorted scanning equipment. Typically, predictions are made on single slices which are then combined for a comprehensive outcome. Yet, this method does not incorporate learning features specific to each slice, leading to a compromise in effectiveness. To address these challenges, we propose an advanced Spatial-Slice Feature Learning (SSFL++) framework specifically tailored for CT scans. It aims to filter out out-of-distribution (OOD) data within the entire CT scan, allowing us to select essential spatial-slice features for analysis by reducing data redundancy by 70\%. Additionally, we introduce a Kernel-Density-based slice Sampling (KDS) method to enhance stability during training and inference phases, thereby accelerating convergence and enhancing overall performance. Remarkably, our experiments reveal that our model achieves promising results with a simple EfficientNet-2D (E2D) model. The effectiveness of our approach is confirmed on the COVID-19-CT-DB datasets provided by the DEF-AI-MIA workshop.

Simple 2D Convolutional Neural Network-based Approach for COVID-19 Detection

TL;DR

The paper tackles COVID-19 detection from CT scans with variable slice counts by introducing SSFL++, a morphology-based, unsupervised RoI and spatial-slice filtering method, and Kernel-Density-based Slice Sampling (KDS) to maintain global sequence information during training. These components are integrated with a simple 2D EfficientNet backbone (E2D), enabling robust performance even with limited data. The approach achieves high macro F1-scores and demonstrates data efficiency, reducing redundancy while preserving diagnostic information, as validated on the COVID-19-CT-DB dataset from the DEF-AI-MIA workshop. The work offers practical improvements for explainable, efficient CT-based COVID-19 detection and suggests strong potential for few-shot and cross-device generalization.

Abstract

This study explores the use of deep learning techniques for analyzing lung Computed Tomography (CT) images. Classic deep learning approaches face challenges with varying slice counts and resolutions in CT images, a diversity arising from the utilization of assorted scanning equipment. Typically, predictions are made on single slices which are then combined for a comprehensive outcome. Yet, this method does not incorporate learning features specific to each slice, leading to a compromise in effectiveness. To address these challenges, we propose an advanced Spatial-Slice Feature Learning (SSFL++) framework specifically tailored for CT scans. It aims to filter out out-of-distribution (OOD) data within the entire CT scan, allowing us to select essential spatial-slice features for analysis by reducing data redundancy by 70\%. Additionally, we introduce a Kernel-Density-based slice Sampling (KDS) method to enhance stability during training and inference phases, thereby accelerating convergence and enhancing overall performance. Remarkably, our experiments reveal that our model achieves promising results with a simple EfficientNet-2D (E2D) model. The effectiveness of our approach is confirmed on the COVID-19-CT-DB datasets provided by the DEF-AI-MIA workshop.
Paper Structure (13 sections, 11 equations, 4 figures, 4 tables)

This paper contains 13 sections, 11 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Region of Interests for Computed Tomography
  4. COVID-19 Recognition
  5. Methodology
  6. Model Architecture
  7. Spatial-Slice Feature Learning
  8. Density-aware Slice Sampling
  9. Experiment
  10. Model Details and Performance Evaluation
  11. Ablation Study
  12. Submission for COVID-19 Detection Challenge
  13. Conclusion

Figures (4)

  • Figure 1: Our model architecture (E2D) and pipeline for COVID19-Detection Challenge.
  • Figure 2: The illustration of proposed SSFL++. It aims to remove redundancy both in spatial and slice dimension by simple computed-morphology methods, thereby improving performance and data-efficiency.
  • Figure 3: The GradCAM++ gradcam++ visualization before and after proposed SSFL++.
  • Figure 4: The illustration of ensemble strategy for our submission.