SIG Health - Healthcare Informatics and Health Info Technology

Loading...

Media is loading
 

Paper Type

Complete

Paper Number

1697

Description

The recent advancements under the umbrella of artificial intelligence (AI) open opportunities to tackle complex problems related to image analysis. Recently, the proliferation of COVID-19 brought multiple challenges to medical practitioners, such as precise analysis and classification of COVID-19 cases. Deep learning (DL) and transfer learning (TL) techniques appear to be attractive solutions. To provide the precise classification of COVID-19 cases, this article presents a customized Deep Convolutional Neural Network (DCNN) and pre-trained TL model approach. Our pipeline accommodated several popular pre-trained TL models, namely DenseNet121, ResNet50, InceptionV3, EfficientNetB0, and VGG16, to classify COVID-19 positive and negative cases. We evaluated and compared the performance of these models with a wide range of measures, including accuracy, precision, recall, and F1 score for classifying COVID-19 cases based on chest X-rays. The results demonstrate that our customized DCNN model performed well with randomly assigned weights, achieving 98.5% recall and 97.0% accuracy.

Comments

SIG Health

Share

COinS
 
Aug 10th, 12:00 AM

Classification of COVID-19 Cases: The Customized Deep Convolutional Neural Network and Transfer Learning Approach

The recent advancements under the umbrella of artificial intelligence (AI) open opportunities to tackle complex problems related to image analysis. Recently, the proliferation of COVID-19 brought multiple challenges to medical practitioners, such as precise analysis and classification of COVID-19 cases. Deep learning (DL) and transfer learning (TL) techniques appear to be attractive solutions. To provide the precise classification of COVID-19 cases, this article presents a customized Deep Convolutional Neural Network (DCNN) and pre-trained TL model approach. Our pipeline accommodated several popular pre-trained TL models, namely DenseNet121, ResNet50, InceptionV3, EfficientNetB0, and VGG16, to classify COVID-19 positive and negative cases. We evaluated and compared the performance of these models with a wide range of measures, including accuracy, precision, recall, and F1 score for classifying COVID-19 cases based on chest X-rays. The results demonstrate that our customized DCNN model performed well with randomly assigned weights, achieving 98.5% recall and 97.0% accuracy.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.