Two-Stream Regression Network for Dental Implant Position Prediction

TL;DR

This paper introduces a two-stream deep learning framework combining an implant region detector and a multi-scale patch embedding regression network to accurately predict dental implant positions, addressing challenges of similar textures and variable spacing.

Contribution

The novel TSIPR framework integrates a rich annotation extension, multi-scale feature extraction, and a global-local transformer-convolution encoder for improved implant position prediction.

Findings

Achieves superior accuracy over existing methods.

Effectively handles variable tooth spacing and similar textures.

Demonstrates robustness through extensive cross-validation.

Abstract

In implant prosthesis treatment, the design of the surgical guide heavily relies on the manual location of the implant position, which is subjective and prone to doctor's experiences. When deep learning based methods has started to be applied to address this problem, the space between teeth are various and some of them might present similar texture characteristic with the actual implant region. Both problems make a big challenge for the implant position prediction. In this paper, we develop a two-stream implant position regression framework (TSIPR), which consists of an implant region detector (IRD) and a multi-scale patch embedding regression network (MSPENet), to address this issue. For the training of IRD, we extend the original annotation to provide additional supervisory information, which contains much more rich characteristic and do not introduce extra labeling costs. A multi-scale patch embedding module is designed for the MSPENet to adaptively extract features from the images with various tooth spacing. The global-local feature interaction block is designed to build the encoder of MSPENet, which combines the transformer and convolution for enriched feature representation. During inference, the RoI mask extracted from the IRD is used to refine the prediction results of the MSPENet. Extensive experiments on a dental implant dataset through five-fold cross-validation demonstrated that the proposed TSIPR achieves superior performance than existing methods.