Quantum Machine Learning Architecture for COVID-19 Classification Based on Synthetic Data Generation Using...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Quantum Computing

Background:

• COVID-19 significantly impacts the respiratory system, necessitating accurate diagnostic tools.
• Computed Tomography (CT) scans reveal characteristic lung opacities like ground-glass opacity and consolidation in COVID-19 patients.
• Early detection and analysis of COVID-19 through medical imaging are crucial for managing the pandemic.

Purpose of the Study:

• To investigate the efficacy of quantum machine learning (QML) and classical machine learning (CML) for analyzing COVID-19 CT images.
• To develop and compare QML and CML models for accurate COVID-19 classification.
• To address data limitations by generating synthetic CT images using Conditional Generative Adversarial Networks (CGAN).

Main Methods:

• Phase I: Generation of synthetic CT images using CGAN to augment the dataset for improved model training and testing.
• Phase II: Development and application of both CML and QML models for classifying COVID-19 positive and healthy lung CT images.
• Utilized two distinct datasets, POF Hospital and UCSD-AI4H, for model evaluation.

Main Results:

• The proposed models demonstrated high performance across both datasets.
• On the POF Hospital dataset, the models achieved precision, accuracy, recall, and F1-scores of 0.94.
• On the UCSD-AI4H dataset, the models achieved precision (0.96), accuracy (0.96), recall (0.95), and F1-scores (0.96).

Conclusions:

• The developed QML and CML approaches show significant promise for COVID-19 detection in CT scans.
• The proposed method surpasses the performance of recent state-of-the-art techniques in this domain.
• The study highlights the potential of integrating quantum computing with machine learning for medical image analysis in infectious diseases.