Abstract:Teeth segmentation is a crucial and fundamental player for doctors in diagnosis and treatment planning in dentistry. Due to the blurred interdental boundaries, variations in noise, and the complexities arising from the orientation and overlapping of dental structures within oral images, the segmentation process becomes extremely challenging and time-consuming. Nowadays, computational tools have been introduced as promising strategies for automating teeth segmentation. As one of them, this paper presents a novel architecture called LCAT-Net, designed to address these challenges and improve teeth segmentation in panoramic X-rays. The proposed architecture incorporates several components to address the above challenges. Firstly, it leverages the main components of the Half-UNet for lightweight feature extraction, starting from ghost modules, unified channel numbers, and full-scale feature fusion. Secondly, to give our model the ability to focus on critical regions for improved differentiation in complex dental structures, a convolutional block attention module (CBAM) is integrated into the network. Thirdly, the architecture incorporates a novel multi-scale context fusion (MCF) module, our proposed MCF module extracts multi-scale spatial information through a spatial context fusion (SCF) block, followed by a CBAM block that learns to balance channel-wise features. The network uses a Dense skip connection module (DSM) to reduce the semantic gap. Experiments on three dental panoramic X-ray image datasets of Children, Adults, and Combined (Children and Adults) consisting of 193, 1776, and 1969 X-rays show that our model outperformed the SOTA models in teeth segmentation, with a high mean Dice-scores of 0.9235, 0.9444, 0.9405, respectively. While requiring significantly fewer parameters and floating-point operations (FLOPs) than existing methods.

Intergrated Segmentation and Detection Models for Dentex Challenge 2023

DENTEX: An Abnormal Tooth Detection with Dental Enumeration and Diagnosis Benchmark for Panoramic X-rays

A Sequential Framework for Detection and Classification of Abnormal Teeth in Panoramic X-rays

SSAD: Self-supervised Auxiliary Detection Framework for Panoramic X-ray based Dental Disease Diagnosis

DETDet: Dual Ensemble Teeth Detection

Instance Segmentation and Teeth Classification in Panoramic X-rays

Clinically Applicable System For 3D Teeth Segmentation in Intraoral Scans using Deep Learning

A Deep Learning Framework with Pruning RoI Proposal for Dental Caries Detection in Panoramic X-ray Images.

Robust Automated Teeth Identification from Dental Radiographs Using Deep Learning.

A dual-labeled dataset and fusion model for automatic teeth segmentation, numbering, and state assessment on panoramic radiographs

3DTeethSeg'22: 3D Teeth Scan Segmentation and Labeling Challenge

Automatic Detection of Tooth-Gingiva Trim Lines on Dental Surfaces

YOLOrtho -- A Unified Framework for Teeth Enumeration and Dental Disease Detection

Collaborative deep learning model for tooth segmentation and identification using panoramic radiographs

STSN-Net: Simultaneous Tooth Segmentation and Numbering Method in Crowded Environments with Deep Learning

LCAT-Net: Lightweight Context-Aware Deep Learning Approach for Teeth Segmentation in Panoramic X-rays

STS MICCAI 2023 Challenge: Grand challenge on 2D and 3D semi-supervised tooth segmentation

Automatic Tooth Segmentation and 3D Reconstruction from Panoramic and Lateral Radiographs

Tooth Defect Segmentation in 3D Mesh Scans Using Deep Learning

Deep learning-based efficient diagnosis of periapical diseases with dental X-rays

Deep Learning-Based Detection of Impacted Teeth on Panoramic Radiographs