Enhanced Detection of Gastric Cancer in Endoscopic Images Using Advanced Deep Convolutional Neural Networks with Custom Architectures

Gastric cancer, also referred to as stomach cancer, is a serious health issue, particularly due to the need for accurate interpretation and classification of endoscopic imaging data. This paper presents an innovative deep transfer learning approach, known as the customized deep convolutional neural network model, aimed at enhancing the interpretation and classification of endoscopic images related to stomach cancer. A significant challenge in medical image processing lies in the segmentation of gastric intestinal tumors, which is crucial for enabling automated detection, as these tumors can differ significantly in size and shape. This diagnostic method shows gastric cancer deep learning architecture. The private datasets are been taken for gastric cancer detection and segmentation process. Deep learning model using the Pycharm python automated detection application that been uses for the segmentation purpose and its works best in custom network that is named as GIST-Net (Gastrointestinal stromal tumors) network with U-Net and Res-Net combines to segment the accurate region of tumor. The next part of medical image processing is to classifying the tumors with DenseNet, EfficientNet, inceptionv3, and MobileNet. The image gets masked, and then predict the binary nodes by eliminating the mucous membrane of internal organs and features are gets extracted, and iterating to find the tumors in the image. Furthermore, the extraction prediction computes parameters based on the regions and pixels shared between the ground truth and the predicted segmentation. It classifies medical images by efficiently training the dataset in a short time. Deep neural networks are then employed to detect, localize, and segment objects in computer vision for clinical image analysis. Additionally, a Python-based graphical user interface (GUI) program is utilized to preprocess endoscopic images, identify tumors using deep CNN architecture in mixed networks, and segment stomach cancer.The identify the segment and classification is been customized by the deep learning neural networks and analysis the result in accuracy higher accuracy range in the gastro intestinal tumor identification methods and it approach the image-based pixel segmentation and classifying the ranges with new customized algorithms.Early finding analysis of tumor its improves treatment and survival.