Improving Early Detection and Classification of Lung Diseases With Innovative MobileNetV2 Framework

Abstract

Any condition that damages or impedes the normal operation of the lungs is classified as a lung disease, and failure to identify and address it early can potentially lead to false outcomes. To address this challenge, two innovative techniques are proposed for lung disease classification, supporting medical professionals in diagnosing and providing preventive measures at an early stage. The Proposed Model 1 integrates a custom MobileNetV2L2 architecture, that builds upon the MobileNetV2 framework through fine-tuning and customization. This model enhances performance by incorporating a ridge or L2 Regularizer within its dense layer. The Proposed Model 2, custom CNN2 built on CNN as its foundational block, is fine-tuned with ELU as the activation function, replacing ReLU, and incorporates the ridge or L2 regularization technique. The proposed research utilizes two publicly available datasets: DS1(Data Set1), which is the Lung Disease 5-class dataset, and DS2(Data Set2), which is the Lung Disease 4-class dataset and is collected from Kaggle. In comparison to cutting-edge methods like EfficientNet B0, InceptionV3, ResNet, and InceptionResNetV2, the Proposed Model 1’s outcomes perform better. It attained 100% validation accuracy, 99.53% training accuracy, and 95.51% test accuracy. Proposed Model 2 achieved testing accuracy of 99.26%, validation accuracy of 91.56%, and training accuracy of 96.79%, the suggested Proposed Model 2 performs quite well. As a supplementary opinion during the diagnostic process, the proposed research is a useful tool for pulmonologists.

Keywords

• Convolutional neural networks
• contrast limited histogram equalization technique
• DenseNet
• InceptionV3
• InceptionResNetV2
• MobileNetV2