Clinical tooth segmentation based on local enhancement

Abstract

The tooth arrangements of human beings are challenging to accurately observe when relying on dentists’ naked eyes, especially for dental caries in children, which is difficult to detect. Cone-beam computer tomography (CBCT) is used as an auxiliary method to measure patients’ teeth, including children. However, subjective and irreproducible manual measurements are required during this process, which wastes much time and energy for the dentists. Therefore, a fast and accurate tooth segmentation algorithm that can replace repeated calculations and annotations in manual segmentation has tremendous clinical significance. This study proposes a local contextual enhancement model for clinical dental CBCT images. The local enhancement model, which is more suitable for dental CBCT images, is proposed based on the analysis of the existing contextual models. Then, the local enhancement model is fused into an encoder–decoder framework for dental CBCT images. At last, extensive experiments are conducted to validate our method.

Semi or fully automatic tooth segmentation in CBCT images: a review

“… Yang et al (2021) amalgamate the 2D U-Net model with an enhanced ACM, addressing the challenges that DL methods may face in automatic topology changes. Compared to other 2D U-Net-based models ( Rao et al, 2020 ; Lee et al, 2020 ; Wu et al, 2022 ; Tao & Wang, 2022 ), these two approaches can achieve higher Dice/IoU scores with a smaller dataset (10 in Yang et al (2021) and 24 in Li et al (2020) ).…”

“…The expanded modules were introduced by Wu et al (2022) and Tao & Wang (2022) into the 2D U-Net. Wu et al (2022) integrate a local feature enhancement module (LE) into the decoder network to fully leverage accurate semantic and location context information across the input image. Tao & Wang (2022) introduce an attention module into the 2D U-Net network to amplify the importance of critical information.…”

“… Tao & Wang (2022) introduce an attention module into the 2D U-Net network to amplify the importance of critical information. These models ( Wu et al, 2022 ; Tao & Wang, 2022 ) optimize the 2D U-Net-based model, and their IoU scores are approximately 0.84.…”

Deep learning-based tooth segmentation methods in medical imaging: A review

“…The experiment achieved a DSC of 94.8% on 97 3D CBCT dental images. Wu et al 81 proposed a local enhancement module for dental CBCT images based on U-Net and DeepLabV3+. 82 This local enhancement module references the ASPP module of DeepLabV3+ and focuses more on enhancing local features.…”