CHS-Net: A Deep Learning Approach for Hierarchical Segmentation of COVID-19 via CT Images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonology

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
• Medical imaging, particularly CT scans, is crucial for diagnosing COVID-19.
• Manual analysis of CT scans is time-consuming for radiologists.

Purpose of the Study:

• To develop an automated deep learning model for segmenting COVID-19 infected regions in lung CT images.
• To improve the speed and accuracy of COVID-19 diagnosis using medical imaging.

Main Methods:

• Proposed a COVID-19 hierarchical segmentation network (CHS-Net) using two cascaded residual attention inception U-Net (RAIU-Net) models.
• RAIU-Net incorporates spectral spatial and depth attention with depthwise separable convolutions and hybrid pooling.
• Trained CHS-Net using a segmentation loss function combining binary cross-entropy and dice loss.

Main Results:

• CHS-Net effectively segmented COVID-19 infected regions in lung CT images.
• The proposed model demonstrated superior performance compared to existing approaches.
• Evaluated using metrics like accuracy, precision, recall, dice coefficient, and Jaccard similarity.

Conclusions:

• The automated CHS-Net model shows significant potential for aiding in the rapid and accurate diagnosis of COVID-19.
• Deep learning segmentation networks can enhance the efficiency of radiological analysis for infectious diseases.
• Further validation and clinical integration of CHS-Net are warranted.