A system for simultaneous near-infrared reflectance and transillumination imaging of occlusal carious lesions

Multispectral SWIR images of the pulp-chamber of posterior teeth in vitro

Intraoral imaging of teeth with SWIR light provides increased contrast of dental caries and restorative materials compared to visible inspection and digital radiography. The objective of this study was to investigate the SWIR optical properties of the dental pulp-chamber floor, walls and canal orifices. We imaged in vitro extracted human posterior teeth at 1300-nm and 1500-1700-nm in reflectance and transillumination and compared the tissues properties with visible light images and quantitative light fluorescence. Transillumination of posterior teeth at both 1300-nm and 1500-1700-nm yielded significantly higher contrast between the pulp-chamber floor and walls than all other methods tested.

Caries Detection with Near-Infrared Transillumination Using Deep Learning

Dental caries is the most prevalent chronic condition worldwide. Early detection can significantly improve treatment outcomes and reduce the need for invasive procedures. Recently, near-infrared transillumination (TI) imaging has been shown to be effective for the detection of early stage lesions. In this work, we present a deep learning model for the automated detection and localization of dental lesions in TI images. Our method is based on a convolutional neural network (CNN) trained on a semantic segmentation task. We use various strategies to mitigate issues related to training data scarcity, class imbalance, and overfitting. With only 185 training samples, our model achieved an overall mean intersection-over-union (IOU) score of 72.7% on a 5-class segmentation task and specifically an IOU score of 49.5% and 49.0% for proximal and occlusal carious lesions, respectively. In addition, we constructed a simplified task, in which regions of interest were evaluated for the binary presence or absence of carious lesions. For this task, our model achieved an area under the receiver operating characteristic curve of 83.6% and 85.6% for occlusal and proximal lesions, respectively. Our work demonstrates that a deep learning approach for the analysis of dental images holds promise for increasing the speed and accuracy of caries detection, supporting the diagnoses of dental practitioners, and improving patient outcomes.

Multispectral near-infrared reflectance and transillumination imaging of occlusal carious lesions: Variation in lesion contrast with lesion depth

In vivo and in vitro studies have demonstrated that near-infrared (NIR) light at λ=1300-1700-nm can be used to acquire high contrast images of enamel demineralization without interference of stains. The objective of this study was to determine if a relationship exists between the NIR image contrast of occlusal lesions and the depth of the lesion. Extracted teeth with varying amounts of natural occlusal decay were measured using a multispectral-multimodal NIR imaging system which captures λ=1300-nm occlusal transillumination, and λ=1500-1700-nm cross-polarized reflectance images. Image analysis software was used to calculate the lesion contrast detected in both images from matched positions of each imaging modality. Samples were serially sectioned across the lesion with a precision saw, and polarized light microscopy was used to measure the respective lesion depth relative to the dentinoenamel junction. Lesion contrast measured from NIR cross-polarized reflectance images positively correlated (p<0.05) with increasing lesion depth and a statistically significant difference between inner enamel and dentin lesions was observed. The lateral width of pit and fissures lesions measured in both NIR cross-polarized reflectance and NIR transillumination positively correlated with lesion depth.