| Abstrakt: |
A range of infectious bacteria and non-infectious factors can both induce and lead to pneumonia, which is an illness of the pulmonary parenchyma. Several age categories are vulnerable, but children under the age of five tend to be especially vulnerable. Chest x-rays, the most frequent radiological test, are a highly significant modality for which numerous uses have been studied. Additionally, x-ray equipment has an upside due to their reduced radiation exposure over imaging technologies like tomography and their potential for accessibility from remote locations. Unfortunately, it might be challenging for physicians to identify pediatric pneumonia because x-ray scans are not always clear or because of human traits like weariness and inattentiveness. In this research, the authors provide a framework for brightness preserving and contrast boosting to strengthen the precision of existing deep learning models. Different cutting-edge deep learning models including VGG16, GoogLeNet, AlexNet, DenseNet-121, InceptionV3, ResNet-18, ResNet-50, ResNet-101, and SqueezeNet are tested both with and without the improvement suggested in this research. The results obtained clearly show the improvement in classification accuracy attained after implementing the proposed image enhancement. The experiments exhibit a maximum of 10.57% improvement in the overall accuracy attained without enhancing the image with the proposed framework. The highest accuracy recorded is 91.05% with RestNet-101. [ABSTRACT FROM AUTHOR] |
