Segmentation of the Lungs on X-Ray Thorax Images with the U-Net CNN Architecture

Lungs are one of the most important parts human body. They very susceptible to various disorders and diseases. For this reason, it is necessary detect or diagnose lungs. In study, we present a method for lung segmentation using CNN U-Net architecture. The initial stage was pre-processed did 1-1 correspondence equalize amount training data testing resized image so all images have same size. process continued with CLAHE (Contrast Limited Adaptive Histogram Equalization), after that, carried out according method. This study used dataset from Kaggle website. results architecture in get an average accuracy 91.68%, sensitivity 92.80%, specificity 89.15%, precision 95.07, F1-Score 93. 92%. Based on performance evaluation results, concluded that proposed great valid lungs X-Ray Thorax images.