DCNET: A Novel Implementation of Gastric Cancer Detection System through Deep Learning Convolution Networks

• Published in: 2022 International Conference on Advanced Computing Technologies and Applications (ICACTA) pp. 1 - 5
• Main Authors: Sharanyaa, S., Vijayalakshmi, S., Therasa, M., Kumaran, U., Deepika, R.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 04-03-2022
• Subjects: Computational modeling; Computer architecture; Deep learning; Feature extraction; Gastric cancer; Image color analysis; Image prrocessing; Machine learning; Pathology; Pathology images; Predictive models
• DOI: 10.1109/ICACTA54488.2022.9752960

To evaluate the Early Gastric Cancer (EGC) detection in humans one of the most common diseases that act as neoplastic disease and second-largest fatal tumor-based disease. Medical imaging techniques and screening equipment such as endoscopy, Computer tomography scanning help the medical industry to detect gastric cancer. The system focuses on implementing a robust prediction scheme that uses image processing techniques to detect the early stage of cancer through lightweight techniques. The test image from the pathology database named BioGPS is preprocessed initially to remove the noisy part of the pixels. The extraction of color features is done using the color threshold algorithm by tuning the image color bands separately. From the R, G, B band the extracted unique feature pixels are mapped in the feature vectors. The cancer part is highlighted by the combination of the R band that associates more with Red pixel points. These formulated pixel vectors are unique and more precise. This is further fetched to the deep Color-Net model (Deep CNET) that compares the training vector with the test vector to find the maximum correlation. The higher the match score the classified results determine the presence of gastric cancer and highlight the spread area from the given test pathology data. Further the system performance is measured using accuracy, precision, recall and F1-Score.