<![CDATA[Chest or thoracic X-ray examination is the most commonly used supporting examination in the diagnosis of lung diseases. In addition to being quick, X-rays are more economical than CT scans or laboratory blood tests. Before determining the disease that appears from the lung image in the chest X-ray image, the doctor first determines the boundaries of the lung area. Not every chest X-Ray image has a normal lung image, some display abnormal images that look white haze or morphological changes due to lung disease processes. As one of the CNN architectures that can be used in segmenting the lungs is UNET which is an encode-decoder architecture. This research tries to train a deep learning model with U-Net CNN architecture. From the results of our experiments, the proposed model can show the ability to recognize lung boundaries even though there are abnormal or foggy lung images. The performance of the model is calculated by measuring the pixel accuracy value and the overlap value with Jaccard Index (IoU), the values of both are 98.25% and 94.54% respectively.]]>
Copyrights © 2023
