Abstract

Acute appendicitis remains a significant cause of acute abdominal pain requiring emergency intervention. Timely and accurate differentiation between uncomplicated and complicated (perforated) acute appendicitis is critical for optimizing patient management and minimizing morbidity and mortality rates. Conventional diagnostic techniques, such as clinical assessment, biochemical analysis, and imaging procedures, sometimes encounter difficulties in differentiating uncomplicated from complicated appendicitis due to overlapping clinical characteristics and interobserver variability. Recent advancements in deep learning (DL) architectures have transformed medical diagnostics, offering new opportunities for more precise disease classification. Convolutional neural networks (CNNs) and other DL-based models have demonstrated significant potential in analyzing radiological images, improving diagnostic accuracy, and reducing false negatives. These models can extract subtle imaging features that may not be easily identifiable by human evaluation, thus enhancing early detection and guiding timely surgical intervention. This narrative review explores the role of DL in differentiating uncomplicated and complicated appendicitis, assessing current methodologies, their performance metrics, limitations, and clinical implications. The findings highlight the potential for DL to revolutionize appendicitis diagnostics, ultimately contributing to improved patient outcomes and streamlined clinical workflows.

Key words: Acute appendicitis, perforated acute appendicitis, deep learning, classification