Table 7.

Comparative analysis of state-of-the-art algorithms in terms of different performance metrics with our proposed approach.

Previous Work	DL Model used for FL	Task	Dataset	Performance Measures
		Covid-19, Pneumonia, Lung opacity and Normal X-Ray Image classification
[82]	Lightweight Communication efficient CNN
			CXR Dataset	Accuracy= 92.96%
		Classify Pneumonia and Normal using CXR images		AUC = 0.9185 and Accuracy = 0.8029
[83]	CNN		CXR Dataset
		Classification of Covid-19, Pneumonia, Lung opacity and Normal X-Ray Images
[84]	CNN and Image Augmentation
			CXR Dataset	Accuracy = 89.43%
		Classify MRI images into Yes -Tumors and Non-Tumor		Precision 0.91, recall 0.96 and F1 Score 0.93 for No Tumor, While Precision 0.91, recall 0.82 and F1 Score 0.86 for Yes Tumor
[85]	Vgg16, Inception V2, DenseNet 121		MRI Dataset
[55]	FedFocus: Integrated CNN and FL	Covid-19 detection using CXR images	CXR Dataset	Accuracy = 92%
				Accuracy = 97%, Sensitivity 98.11% and Specificity 95.89% for FL-ResNet50 while Accuracy = 94.40%, Sensitivity 96.15% and Specificity 92.66 for FL-VGG-16
[86]	VGG-16, ResNet50	Screening Covid-19 using CXR	CXR Dataset
[87]	MobileNet v2, ResNeXt and ResNet18	Detection of Covid-19 using Pneumonia CXR images		ResNet18 performed best with Sensitivity = 91.26%
			CXR Dataset
		Classifying into COVID-19, TB, PNA, and healthy individuals using CXR images as well as Glioma, Meningioma, Pituitary and no tumor using MRI images		Accuracy = 99.24%, Error Rate = 0.0076, Cohen’s Kapp = 0.9967, and Average F1 Score = 0.995 for CXR Dataset. For MRI dataset, Accuracy = 99%, Error Rate = 0.01, Cohen’s Kapp = 0.9698 and Average F1 Score = 0.995
Proposed Work	FDEIoL		CXR and MRI Datasets