AI Case Study
Researchers at Cornell University are using deep learning to develop dental restorations with better accuracy than human eye
Researchers at Cornell University are working to improve the accuracy of digitally designed dental crown restorations beyond that of human eye. They are using generative adversarial network to predict the customized crown-filled depth scan from the crown-missing depth scan and opposing depth scan and incorporates additional space constraints and statistical compatibility with the designs. The results so far have achieved better accuracy and functionality than human technicians.
Healthcare Equipment And Supplies
"Computer vision has advanced significantly that many discriminative approaches such as object recognition are now widely used in real applications. We present another exciting development that utilizes generative models for the mass customization of medical products such as dental crowns. In the dental industry, it takes a technician years of training to design synthetic crowns that restore the function and integrity of missing teeth. Each crown must be customized to individual patients, and it requires human expertise in a time-consuming and labor-intensive process, even with computer-assisted design software. We develop a fully automatic approach that learns not only from human designs of dental crowns, but also from natural spatial profiles between opposing teeth. The latter is hard to account for by technicians but important for proper biting and chewing functions. Built upon a Generative Adversarial Network architecture (GAN), our deep learning model predicts the customized crown-filled depth scan from the crown-missing depth scan and opposing depth scan. We propose to incorporate additional space constraints and statistical compatibility into learning. Our automatic designs exceed human technicians' standards for good morphology and functionality, and our algorithm is being tested for production use."
The generated crowns not only reach similar morphology quality
as human experts’ designs but support better functionality enabled by learning through statistical features.
"Network Architecture: We follow closely the architecture design of pix2pix . For generator G, we use the U-Net  architecture, which contains an encoder decoder architecture, with symmetric skip connections. It has been shown to produce strong results when there is a spatial correspondence between input and output pairs. The encoder architecture is: C64-C128-C256-C512-C512-C512-C512-C512, where C# denotes a convolution layer followed by the number of filters. The decoder architecture is: CD512-CD512-CD512-C512-C256-C128-C64, where CD# denotes a deconvolution layer followed by the number of filters. After the last layer in the decoder, a convolution is applied to map to the number of output channels, followed by a Tanh function. BatchNorm is applied after every convolutional layer except for the first C64 layer in the encoder. All ReLUs in the encoder are leaky with slope 0.2, while ReLUs in the decoder are regular. For discriminator D, the architecture is: C64-C128-C256-C512. After the last layer, a convolution is applied to map to a 1 dimensional output, followed by a Sigmoid function. BatchNorm is applied after every convolution layer except for the first C64 layer. "
R And D
"Dental Computer aided design (CAD) has advanced significantly in the recent years but still needs human intervention. Researchers at Cornell University are building a model to improve the accuracy of the design."
"Dataset contains 1500 training, 1570 validation, and
243 hard testing cases each of which has 2D scan images of the prepared jaw, opposing jaw, and the gap distances between two jaws from the original intra-oral 3D scan model."