Title: S-R2F2U-Net: a single-stage model for teeth segmentation

Authors: Mrinal Kanti Dhar; Mou Deb

Addresses: Electrical and Electronic Engineering, Leading University, Sylhet, Bangladesh ' Electrical Engineering, South Dakota School of Mines and Technology, SD, USA

Abstract: Precision tooth segmentation is crucial in the oral sector because it provides location information for orthodontic therapy, clinical diagnosis, and surgical treatments. In this paper, we investigate residual, recurrent, and attention networks to segment teeth from panoramic dental images. Based on our findings, we suggest three models: single recurrent R2U-Net (S-R2U-Net), single recurrent filter double R2U-Net (S-R2F2U-Net), and single recurrent attention enabled filter double (S-R2F2-Attn-U-Net). Particularly, S-R2F2U-Net, as emphasised in the title, outperforms state-of-the-art models in terms of accuracy and dice score. A hybrid loss function combining cross-entropy loss and dice loss is used in training. In addition, it reduces around 45% of model parameters compared to the original R2U-Net. Models are trained and evaluated on the UFBA-UESC dataset that contains 1,500 extra-oral panoramic X-ray images and divided into ten categories based on the structural variations. S-R2F2U-Net achieves 97.31% accuracy and 93.26% dice score. Codes are available on https://github.com/mrinal054/ teethSeg_sr2f2u-net.

Keywords: tooth segmentation; semantic segmentation; deep learning; recurrent module; attention module.

DOI: 10.1504/IJBET.2024.141569

International Journal of Biomedical Engineering and Technology, 2024 Vol.46 No.2, pp.81 - 100

Received: 03 Aug 2023
Accepted: 29 Jan 2024
Published online: 24 Sep 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article