An Enhanced Feed-Forward Neural Network for Gastrointestinal Stromal Tumor Detection in CT Images

Abstract

Gastrointestinal stromal tumors (GISTs) are rare soft-tissue neoplasms that can be life-threatening if not detected at an early stage. Accurate and automated analysis of computed tomography (CT) images plays a crucial role in supporting timely diagnosis and treatment planning. This study proposes an enhanced feed-forward artificial neural network (IFFANN)–based classification framework for the automatic detection of GISTs from CT images. A curated dataset comprising GIST-positive and non-GIST CT scans was assembled from multiple sources and annotated by expert radiologists. The proposed network integrates optimized feature learning through convolutional layers followed by a feed-forward classification stage to improve discrimination between pathological and normal cases. The model was trained and evaluated using cross-validation to ensure robustness and reduce overfitting. Experimental results demonstrate that the proposed IFFANN achieves high accuracy, sensitivity, specificity, and F1-score in distinguishing GIST cases from non-GIST samples. The findings indicate that the proposed approach can effectively support computer-aided diagnosis systems and assist radiologists in improving diagnostic accuracy and efficiency. Further validation of larger and more diverse datasets will strengthen the generalizability of the proposed method.