Cognitive Learning Models for Predicting Apical Microleakage Using Pre- and Post-Operative CBCT Data

Abstract

Apical microleakage remains a major factor affecting the long-term success of endodontic treatment, often leading to treatment failure if undetected. Conventional diagnostic methods rely heavily on clinical expertise and qualitative image interpretation, which can be subjective and limited in sensitivity. This study proposes the application of cognitive learning models to predict apical microleakage by leveraging pre- and post-operative cone-beam computed tomography (CBCT) data. The framework integrates imaging-derived features from both treatment stages to capture structural and morphological changes associated with microleakage development. Advanced learning techniques are employed to model complex spatial patterns and variations within CBCT scans, enabling improved predictive accuracy compared to traditional assessment approaches. Experimental results demonstrate that the proposed cognitive learning-based approach enhances early detection and risk prediction of apical microleakage, supporting more informed clinical decision-making. The findings highlight the potential of intelligent imaging analysis to augment endodontic diagnostics and contribute to improved treatment outcomes.