From Dental Hard Tissue to Digital Simulation: Analysis of Non-Ionizing Caries Diagnostics Using Digital Twins

• Applicants: Dr. Katrin Heck; Professor Dr. Alwin Kienle; Dr. Elias Walter
• Subject Area: Dentistry, Oral Surgery
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 494559609

The present project aims to analyze, objectify, and optimize optical caries diagnostics in the shortwave infrared (SWIR) range using transillumination and reflection with digital twins. Digital twins, virtual representations of real systems, enable the investigation of the depiction of healthy and carious dental hard tissues, considering various influencing factors (e.g., wavelength, type and extent of caries) and allow for three-dimensional simulations. Through the simulation of light propagation within teeth in clinically relevant scenarios, the underlying mechanisms of caries representation in images can be better understood and result in a more valid and reliable caries detection. To achieve this, the optical properties of dental hard tissues such as dentin and enamel are determined to expand the fundamental knowledge of light propagation in dental tissue and develop simulation-based models. Automated algorithms and neural networks are developed in parallel to precisely and rapidly segment tooth structures and caries lesions in microtomographic images (gammaCT) which result in high resolution digital 3D models of teeth. These AI models aim to shorten analysis time and standardize the use of gammaCT as a high-resolution procedure in caries research. They also serve as reference for the objective analysis of transillumination and reflection images and the development of digital twins. To objectively evaluate SWIR images, a standardized analysis of image properties (radiomics) is conducted using statistical comparisons and hierarchical cluster analysis, aiming to assign caries sizes and locations to specific image patterns. This analysis is complemented by morphological investigations of the tooth surfaces to quantify the influence of surface morphology on caries depiction and elucidate the underlying mechanisms. This could help explain the high rate of false-positive findings in reflection images. For in-silico analyses, a Monte Carlo code is developed to simulate light propagation within a complex tooth model under various conditions. This allows systematic investigations to be carried out, analyzing the impact of individual tooth and model components. Finally, based on these measurements and analyses, an optical digital twin will be developed that realistically replicates light propagation in teeth and the depiction of carious lesions. Through modifications of tooth form, structure, and caries characteristics, clinical SWIR images can be simulated. These digital replica allow for the identification of factors that influence the accuracy of caries diagnostics, contributing to the improvement of diagnostic precision by providing well-founded insights into the depiction of carious structures through non-invasive optical methods.

• DFG Programme: Research Grants