الفهرس 
TitleOnly 14 pages are availabe for public view
 |  | from | 105 |  |  |
| | | from | 105 | | |
Abstract
Mesenteric lymphadenopathy can be caused by almost any cancer. Lymphoma, which may cause lymphadenopathy practically everywhere in the body, is the most frequent malignancy linked to mesenteric lymphadenopathy. Although mesenteric lymphadenopathy is not common, it can commonly lead to lymphadenopathy in the chest, retroperitoneum, or superficial lymph node chains. Abdominal sonography and CT imaging are the most frequently used screening methods for detecting abdominal lymph node enlargement. The size, localization, and infiltration pattern must be known for proper classification. The size cannot determine whether lymph nodes are benign or malignant of the lesion alone. The aim of the present study was to use the python software package for the precise identification of lymph node lesions compared to that of the healthy controls.
The study included 150 CT images of pathologically proven Abdominal Lymphadenopathy patients and Healthy Controls from worldwide databases. Images were preprocessed before splitting into training and testing datasets using the Convolution Neural Network (CNN) classification algorithm. The CNN with optimized kernel and C parameter was the machine learning algorithm used in this investigation. For these data, the optimal training parameter combination of the CNN model was C = 200 and linear kernel.
The training, validation, and testing accuracies were 1.00, 0.99, and 0.99, respectively, with a mean squared error of 0.16 and a score of 0.02 seconds. The output vector of CT images consists of two classes: Abdominal Lymphadenopathy and Healthy Control, and precision, Recall, F1-Score, and Specificity results. The precision and specificity were as high as 99% and 100% for Abdominal Lymphadenopathy patients and Healthy Controls, respectively. Thus, the CNN is a powerful tool for assisting radiologists in classifying and diagnosing CT images for Abdominal Lymphadenopathy with great precision.