Enhancing Precision in Gastric Cancer Detection: A Synergistic Approach with Convolutional Neural Networks and Metrological Integration

Gastric cancer (GC) is one of the major causes of cancerrelated mortality. It requires major attention in terms of early detection and proper diagnosis. Endoscopy and analysis of endoscopic images is the most common detection method of GC in the early stage. However, specialised professionals are required for endoscopy. This becomes a bottleneck for patients in remote areas and is a major cause of late diagnosis. Hence, there is a need for an efficient, cost-effective user-friendly tool that can identify GC with optimal accuracy and less effort. Recently, Artificial intelligence (AI) has emerged as a promising tool that can enhance precision and accuracy in diagnosis, treatment and patient care. In this paper, a deep learning-based strategy has been presented as a possibility of identifying GC at early stage. For this purpose, clinical parameters, like, complete blood count (CBC) and endoscopic images were correlated to delineate cancerous and non-cancerous ulceration in GI tract. For this preliminary study, GC related kaggle-based open-source data bank has been used to develop various pilot algorithms. A PyCharm based interface has been developed to identify the changes in CBC parameters. If these parameters are beyond a pre-defined cut-off, then the corresponding endoscopic images are analysed. A couple of Convolutional Neural Networks (CNNs) based pre-trained models, like, VGG16 and NasNetMobile have been developed on these images. A comparative analysis has been performed to determine the accuracy, precision, recall, F1 score, sensitivity, specificity and run-time. The algorithms developed provide proof of evidences that deep learning networks can be used to identify GC accurately and sensitively and further can distinguish GC from non-malignant tissues. Thus, there is a potential role of deep learning algorithms in GC diagnosis, segmentation as well as classifications. Integrating CNNs into gastric cancer detection leverages advanced image analysis. Metrology ensures the accuracy of measurements used in the diagnostic process, enhancing the reliability of CNN-based models. This synergy aims to improve the precision of gastric cancer detection by combining sophisticated deep learning tools with precise metrological data, contributing to more accurate and reproducible diagnostic outcomes in the field of medical imaging.