An evaluation of lightweight deep learning techniques in medical imaging for high precision COVID-19 diagnostics

Lightweight deep CNN-based models for early detection of COVID-19 patients from chest X-ray images

“… Reference Method Data Size Accuracy (%) Trainable Parameters COVID-19 Normal (Anunay Gupta et al, 2021 ) Inception-V3 361 365 97.00 24,000,000 ( George et al, 2023 ) GrayVIC model 2,250 2,250 98.06 2,684,650 ( Hussain et al, 2021 ) CoroDet 500 800 99.10 2,873,609 ( Jyoti et al, 2023 ) ResNet50 2409 2866 95.67 23,591,810 (A. I. Khan et al, 2020 ) CoroNet 284 310 99.00 33,969,964 ( Malik et al, 2022 ) Vgg16 4630 1583 93.00 14,715,714 ( Nayak et al, 2021 ) ResNet-34 775 775 98.33 21,800,000 ( Nayak et al, 2023 ) LW-CORONet 2358 8066 96.25 680,000 ( Sahin, 2022 ) CNN_Model 3,626 10,198 96.71 667,458 ( Ukwandu et al, 2022 ) MobileNet-V2 1,200 1,341 99.60 3,538,984 Proposed Model Lightweight CNN Model 3,616 10,192 98.55 ...…”

“… Reference Method Data Size Accuracy (%) Trainable Parameters COVID-19 Normal Pneumonia ( Apostolopoulos & Mpesiana, 2020 ) VGG19 224 504 714 93.48 143,667,240 (Anunay Gupta et al, 2021 ) Inception-V3 361 365 362 97.00 24,000,000 ( George et al, 2023 ) GrayVIC model 2,250 2,250 2,250 97.41 2,684,650 ( Hussain et al, 2021 ) CoroDet 500 800 800 94.20 2,874,635 ( Heidari et al, 2020 ) VGG16 445 2,880 5,179 94.50 138,000,000 (A. I. Khan et al, 2020 ) CoroNet 284 310 657 95.00 33,969,964 ( Nayak et al, 2023 ) LW-CORONet 2358 8066 5575 95.67 680,000 ( Ukwandu et al, 2022 ) MobileNet-V2 1,200 1,341 1,345 94.50 3,538,984 (L. Wang et al, 2020 ) COVID-Net 358 8,066 5,538 93.30 11,750,000 ( Zebin & Rezvy, 2021 ) EfficientNetB0 ...…”

“…( Ukwandu et al, 2022 ) proposed three lightweight CNN models through fine-tuning MobileNetV2 architecture to detect COVID-19 infection using chest X-ray images. The first model was designed for 3-class categorization (COVID-19, Normal, Viral Pneumonia infection).…”