Using Deep Learning Technique, Multi-Task Identification of Abnormal Positioned Tooth in Maxillary Sinus Through a Method that Combines 2D CNN and Transformer Technology

Read full text Summarise with VIRAT
Authors: R. Vimala, D. M. D. Preethi
Cited by: 0
References: 22

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The detection and treatment of abnormally positioned tooth within the maxillary sinus present significant clinical challenges, resulting in complications such as Odontogenic factors are reported amounting to approximately 10-40% of maxillary sinusitis cases.These challenges are intensified by limited access to the sinus cavity, close proximity to vital anatomical structures, and the potential for postoperative issues such as sinus membrane damage.Therefore, accurate detection and localization are essential for planning safe and effective surgical intervention.This research proposes a multi-task model built using deep learning techniques for identifying and localizing abnormally positioned tooth within the maxillary sinus, utilizing a 2D Hybrid CNN-Transformer architecture enhanced with Dynamic Spatial Attention (DSA).Convolutional layers are employed for extracting local spatial features, while Transformer modules learn global contextual information.The integration of spatial attention mechanisms further improves detection precision and anatomical awareness.Trained on a dataset of 300 CBCT scans, the framework achieved 94% classification accuracy, 87.5% precision, 93.3% recall, and 88.7% mean Average Precision (mAP).Comparative benchmarking against ResNet50-YOLOv5 and U-Net-CNN demonstrates improved detection robustness and enhanced generalization capability.The integration of attention-guided feature refinement and optimized post-processing establishes a computationally efficient and anatomically consistent solution for automated maxillary sinus pathology assessment, supporting AI-assisted diagnosis and surgical planning in maxillofacial radiology.

Fields of Study & Subjects

Dental Radiography and Imaging Advanced Neural Network Applications Medical Imaging and Analysis

Publication details

Journal / source
Traitement du signal
Volume
43
Issue
1
Publication date
28 February 2026
Language
EN