A new approach makes it easier for AI to spot pulp cavities: Study

A group of researchers from China proposed a novel data pipeline based on micro-CT data for training the U-Net network to realize the automatic and accurate segmentation of pulp cavity and tooth on cone-beam computed tomography (CBCT) images.

The researchers collected CBCT data and micro-CT data of thirty teeth. CBCT data were processed and transformed into a small field of view and high-resolution CBCT images of each tooth. Twenty-five sets were randomly assigned to the training set and the remaining five sets to the test set. We used two data pipelines for U-Net network training: one manually labelled by an endodontic specialist as the control group, and one processed from the micro-CT data as the experimental group. The 3D models constructed using micro-CT data in the test set were taken as the ground truth. Dice similarity coefficient (DSC), precision rate (PR), recall rate (RR), average symmetric surface distance (ASSD), Hausdorff distance (HD) and morphological analysis were utilized for performance evaluation.

The results of the study are as follows:

· The segmentation accuracy of the experimental group measured by DSC, PR, RR, ASSD, and HD were 96.20±0.58%, 97.31±0.38%, 95.11±0.97%, 0.09±0.01mm, and 1.54±0.51mm in tooth and 86.75±2.42%, 84.45±7.77%, 89.94±4.56%, 0.08±0.02mm, 1.99±0.67mm in the pulp cavity, respectively, which were better than the control group.

· Morphological analysis suggested the segmentation results of the experimental group were better than those of the control group.

Thus, the researchers concluded that this study proposed an automatic and accurate approach for tooth and pulp cavity segmentation on CBCT images, which can be applied in researches and clinical tasks.

Reference:

Micro-Computed Tomography Guided Artificial Intelligence for Pulp Cavity and Tooth Segmentation on Cone-beam Computed Tomography by Lin X et. al published in the Journal of Endodontics.

DOI: https://doi.org/10.1016/j.joen.2021.09.001