Detection of Pulmonary Lesions Using Convolutional Neural Networks in X-Ray Images and Capsule Neural Networks

In this study, a novel approach based on Capsule Neural Networks is presented for detecting pulmonary lesions in medical images. Timely and accurate detection of pulmonary lesions is of critical importance in reducing mortality among patients with pulmonary diseases, particularly during the early stages of the disease. Capsule Neural Networks are considered a suitable tool for this purpose due to their high capability in identifying complex features and spatial relationships within image data. In this study, CT and X-ray images of patients with pulmonary lesions and healthy individuals were collected and, following preprocessing procedures including normalization and data augmentation, were used as inputs to the Capsule Neural Network model. The proposed model extracts important features and identifies patterns associated with pulmonary lesions. Model performance was evaluated using metrics such as accuracy, sensitivity, and the F۱-score. The obtained results demonstrated that the Capsule Neural Network achieved good performance and high accuracy in detecting pulmonary lesions. This study highlights the effectiveness of Capsule Neural Networks in the diagnosis of pulmonary diseases and their potential application in medical systems for the automated detection of lesions.

کلیدواژه ها:

Pulmonary Lesion Detection ، Convolutional Neural Networks ، x-Ray images ، Capsule Neural Networks