Improving performance of deep learning models using 3.5D U-Net via majority voting for tooth segmentation on cone beam computed tomography

Abstract

Deep learning allows automatic segmentation of teeth on cone beam computed tomography (CBCT). However, the segmentation performance of deep learning varies among different training strategies. Our aim was to propose a 3.5D U-Net to improve the performance of the U-Net in segmenting teeth on CBCT. This study retrospectively enrolled 24 patients who received CBCT. Five U-Nets, including 2Da U-Net, 2Dc U-Net, 2Ds U-Net, 2.5Da U-Net, 3D U-Net, were trained to segment the teeth. Four additional U-Nets, including 2.5Dv U-Net, 3.5Dv5 U-Net, 3.5Dv4 U-Net, and 3.5Dv3 U-Net, were obtained using majority voting. Mathematical morphology operations including erosion and dilation (E&D) were applied to remove diminutive noise speckles. Segmentation performance was evaluated by fourfold cross validation using Dice similarity coefficient (DSC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV). Kruskal–Wallis test with post hoc analysis using Bonferroni correction was used for group comparison. P < 0.05 was considered statistically significant. Performance of U-Nets significantly varies among different training strategies for teeth segmentation on CBCT (P < 0.05). The 3.5Dv5 U-Net and 2.5Dv U-Net showed DSC and PPV significantly higher than any of five originally trained U-Nets (all P < 0.05). E&D significantly improved the DSC, accuracy, specificity, and PPV (all P < 0.005). The 3.5Dv5 U-Net achieved highest DSC and accuracy among all U-Nets. The segmentation performance of the U-Net can be improved by majority voting and E&D. Overall speaking, the 3.5Dv5 U-Net achieved the best segmentation performance among all U-Nets.

CBCT-based synthetic CT generated using CycleGAN with HU correction for adaptive radiotherapy of nasopharyngeal carcinoma

“…First, due to the computational power limitation in hardwares, the current approach was only able to implement the 2D CycleGAN. Even results can be reasonably expected if a three-dimensional model could be adopted 32 – 34 . Secondly, due to the rigid registration algorithm used for pre-processing, there is a certain difference between RCT images and CBCT images.…”

CBCT-based synthetic CT generation using a hybrid approach for adaptive radiotherapy of nasopharyngeal carcinoma

“…The information between layers of the image was ignored. It is expected to achieve better results if three-dimensional model was adopted [31][32][33] . Secondly, due to the rigid registration algorithm used for pre-processing, there is a certain RCT images and CBCT images.…”

Automatic Segmentation of Teeth, Crown–Bridge Restorations, Dental Implants, Restorative Fillings, Dental Caries, Residual Roots, and Root Canal Fillings on Orthopantomographs: Convenience and Pitfalls

“…In 1959, Arthur Samuel defined machine learning as “the ability of machines to learn results for which they were not specifically programmed” [ 9 ]. He also developed a checkers game that can work in a computer environment, learn from its own mistakes, and thus improve itself [ 1 , 10 , 11 , 12 ].…”

“…Common applications of AI in oral diagnosis and dentomaxillofacial radiology are as follows: Oral cancer prognosis and assessment of oral cancer risk [ 45 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 ]; Determination of temporomandibular joint disorder progression and temporomandibular internal derangements [ 27 , 30 , 34 , 38 , 63 ]; Interpretation of conventional 2D imaging [ 31 , 64 , 65 , 66 , 67 , 68 ]; Interpretation of cone beam computed tomography and other 3D imaging methods [ 1 , 10 , 12 , 17 , 18 , 19 , 21 , 23 , 27 , 69 , 70 , 71 ]. …”

“…Interpretation of cone beam computed tomography and other 3D imaging methods [ 1 , 10 , 12 , 17 , 18 , 19 , 21 , 23 , 27 , 69 , 70 , 71 ].…”