Related Experiment Video
Updated: Sep 21, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Evolving deep convolutional neural networks by IP-based marine predator algorithm for COVID-19 diagnosis using chest
Bing Liu1, Xuan Nie1, Zhongxian Li1
1School of Software, Northwestern Polytechnical University, Xi'an, Shaanxi Province China.
Journal of Ambient Intelligence and Humanized Computing
|June 1, 2022
Summary
A novel deep convolutional neural network (DCNN) optimized with the marine predator algorithm (MPA) accurately identifies COVID-19 from CT scans. This DCNN-IPMPA model achieved high accuracy, aiding radiologists in diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computational Biology
Background:
- Accurate and rapid diagnosis of COVID-19 is crucial for patient management and public health.
- Deep convolutional neural networks (DCNNs) show promise in analyzing medical images for disease detection.
- Existing DCNN models require optimization for enhanced performance in COVID-19 diagnosis using CT scans.
Purpose of the Study:
- To develop an optimal structured deep convolutional neural network (DCNN) for automated COVID-19 diagnosis using CT scans.
- To enhance the marine predator algorithm (MPA) for optimizing DCNN architecture and parameters.
- To create a novel automatic diagnosis platform to assist radiologists in differentiating COVID-19 from non-COVID-19 patients.
Main Methods:
- Proposed a modified marine predator algorithm (MPA) incorporating a novel Internet Protocol (IP) address-based encoding scheme.
- Introduced an Enfeebled layer to construct a variable-length DCNN, enabling flexible network architectures.
- Implemented a learning process that divides large datasets into smaller, randomly evaluated chunks for efficient training.
Main Results:
- The developed DCNN-IPMPA model achieved high diagnostic accuracy: 97.21% on the SARS-CoV-2 dataset and 97.94% on the COVID-CT dataset.
- Performance comparison against standard DCNN and seven variable-length models demonstrated superior results for DCNN-IPMPA.
- Timing analysis showed competitive processing time for DCNN-IPMPA compared to standard DCNN, indicating efficiency.
Conclusions:
- The proposed DCNN-IPMPA offers a highly accurate and efficient approach for automated COVID-19 diagnosis from CT scans.
- The modified MPA and variable-length DCNN architecture contribute to improved diagnostic performance.
- This novel platform has the potential to significantly aid radiologists in clinical decision-making for COVID-19 detection.
