Lungs Disease Dataset (4 types)

Any damage that affects the normal functioning of the lungs is termed as a lung disease,which can prove fatal if not detected early. To address this challenge, two innovative techniques proposedfor the lung disease classification, supporting medical professionals to diagnose and provides preventivemeasures at an early stage. The proposed Model 1 integrates a custom MobileNetV2L2 architecture, thatbuilds upon the MobileNetV2 framework through fine-tuning and customization. This model incorporates aridge regularizer within its dense layer to enhance its performance. The Proposed Model 2, built on CNN asits foundational block, is fine-tuned with ELU as the activation function, replacing ReLU, and incorporatesthe L2 regularization technique. The proposed research utilizes two publicly available datasets: DS1(DataSet1), which is the Lung Disease 5-class dataset, and DS2(Data Set2), which is the Lung Disease 4-classdataset and are collected from Kaggle. The results from the proposed Model 1 provides better performancethan state-of-the-art techniques like EfficientNet B0, InceptionV3, ResNet, and InceptionResNetV2. Itachieved a training accuracy of 99.53%, validation accuracy of 100%, and test accuracy of 95.51%. Theproposed Model 2 provides outsatnding performance, with a training accuracy of 96.79%, validationaccuracy of 91.56%, and testing accuracy reaching 99.26% The proposed research serves as a valuabletool for doctors, providing a secondary opinion in the diagnostic process.