A Hybrid Deep Learning CNN Model for Enhanced COVID-19 Detection from Computed Tomography (CT) Scan Images

Early detection of COVID-19 plays a vital role in enabling timely treatment and curbing the spread of the virus. This study introduces a novel hybrid deep learning model tailored to identify COVID-19 infections from chest CT scan images, aiming to support healthcare professionals facing overwhelming diagnostic demands. Our approach integrates the strengths of three pre-trained convolutional neural networks namely VGG16, DenseNet121, and MobileNetV2, each known for their robust feature extraction capabilities. These models independently extract deep features from input CT images, capturing both low-level and high-level representations essential for accurate classification. To address potential redundancy and reduce the computational burden, Principal Component Analysis (PCA) is employed for dimensionality reduction. The refined feature vectors from all three models are then concatenated to form a comprehensive feature representation, which is subsequently passed to a Support Vector Classifier (SVC) for final classification. Our hybrid architecture enables the model to leverage the complementary strengths of each CNN while maintaining efficiency. We evaluated our proposed model on a dataset consisting of 2,108 training images and 373 test images, comprising both COVID-positive and non-COVID samples. Comparative analysis with individual CNN models showed that our hybrid model achieved superior performance, reaching an accuracy of 98.93%. It also outperformed standalone models in precision, recall, F1-score, and ROC-AUC, highlighting its potential as a highly reliable and efficient diagnostic aid.