Lung Cancer Detection Using Hybrid Convolution neural Network and Recurrent neural Network | ||||
| Mansoura Journal for Computer and Information Sciences | ||||
| Volume 19, Issue 1, December 2024, Page 63-84 PDF (1.53 MB) | ||||
| Document Type: Original Research Articles. | ||||
| DOI: 10.21608/mjcis.2024.312462.1006 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
Omnia Alaa Mosaad  ; Muhammad H. Zayyan; Mohammed Alrahmawy; Samir Elmougy
| ||||
| Department of Computer Science, Faculty of Computers and Information Sciences, Mansoura University, Mansoura, Egypt | ||||
| Abstract | ||||
| Humans are susceptible to the common and serious disease known as cancer. Lung Cancer (LC) is considered these days as the most common form of cancer in many nations. In this paper, we developed a five-stage method for detecting lung cancer in CT images, which includes preprocessing the image with a Wiener filter, segmenting the image using global thresholding, feature extraction, feature selection, and classification. Statistical and morphological data are combined to create a gray-level-co-occurrence matrix (GLCM), which is used to extract textural features during the feature extraction step. To extract deep features, hybrid Convolution Neural Networks (CNN) and Recurrent Neural Networks (RNN) are also employed. The Slime Mould Algorithm (SMA) is then used to choose the best features using a wrapper method fitness function that considers the criterion's correctness. The classification techniques are then used. Using 100 samples of lung CT images as a sub-dataset, the suggested method is assessed. The experiment results show that SMA is the best feature selection algorithm among other used algorithms, in which it reaches a 95\% accuracy rate, based on Lung Image Database Consortium Image Collection (LIDC-IDRI). This dataset includes 1018 images of malignant and healthy tissue. Moreover, Residual Neural Network (ResNet 18) is shown to be the best classification technique among other used techniques, reaching 98.5\% accuracy, 98.5\% sensitivity, and 99.5\% specificity. | ||||
| Keywords | ||||
| Artificial intelligence; lung cancer detection; deep learning; image processing; and image classification | ||||
|
Statistics Article View: 214 PDF Download: 202 |
||||
