(Poster) Deep Learning Models for Histopathological Classification of Gastric Epithelial Tumors
Leonard Florian Nuta, Loretta Ichim, Dan Popescu · 2025
Gastric cancer is the fifth most diagnosed and the fourth leading cause of cancer death worldwide. Aside from deep learning having advanced in the tasks of histopathological image analysis, the limited amount of training data has been a major problem. We propose a method based on transfer learning using convolutional neural networks for binary classification of stomach tissue patches. The approach was developed on the publicly available GasHisSDB dataset. The CNNs were selected based on the related works studied. The experimental results highlight a test accuracy of 99.0% and an F1-score of 99.1% on the GasHisSDB dataset. This performance is an improvement in the purpose of successfully classifying histopathological images with gastric epithelial tumors.