An Effective Hybrid Deep Learning Technique for Covid-19 Detection Using InceptionV3 and optimized Squeeze Net

Abstract Covid-19 is a virus that affects both animals and people and is quickly expanding. This deadly illness affects people's daily lives, health, and economy. The most significant imaging methods for detecting COVID-19 are a computed tomography (CT) scan and chest X-ray. For this epidemic, all researchers are striving for efficient solutions and quick treatment approaches. Fast and accurate automated detection approaches have been introduced to eliminate the requirement for medical professionals. This work proposes an efficient SqueezeNet for Covid classification to solve this challenge. Inception V3 is employed to extract some of the most important features. SqueezeNet is optimized by Deer Hunt Optimization algorithm (DHO) to improve the performance. From chest x-ray images, a multi-classification DL approach for diagnosing COVID-19, pneumonia, normal, and tuberculosis is proposed. The proposed categorization strategy is implemented in Python and compared to other approaches. In terms of f1-score, sensitivity, specificity, and accuracy experiments have shown that the proposed method outperforms other recent and state-of-the-art methodologies.