Publications by authors named "Katsusuke Yamashita"
This study aims to evaluate the potential enhancement in implant classification performance achieved by incorporating artificially generated images of commercially available products into a deep learning process of dental implant classification using panoramic X-ray images. To supplement an existing dataset of 7,946 in vivo dental implant images, a three-dimensional scanner was employed to create implant surface models. Subsequently, implant surface models were used to generate two-dimensional X-ray images, which were compiled along with original images to create a comprehensive dataset.
View Article and Find Full Text PDFArticle Synopsis
- The study evaluated how accurately deep learning can determine the relationship between the inferior alveolar canal and mandibular third molars using a dataset of 1279 digital radiographs.
- The researchers specifically focused on two types of analysis: contact (presence or absence of contact between the molar and canal) and continuity (bone continuity as seen on CT scans).
- Results showed that the ResNet50v2 model using sharpness-aware minimization (SAM) performed well in contact analysis (accuracy 0.860, AUC 0.890), but the models struggled with continuity analysis, showing limited effectiveness.
Article Synopsis
- * The researchers compared the performance of CNN model VGG16 using both SAM and stochastic gradient descent (SGD) optimizers, with and without a learning rate scheduler over 300 epochs.
- * Results showed that SAM, particularly with the learning rate scheduler, achieved the highest accuracy (11.2), AUC (0.9328), and reduced overfitting, suggesting its potential for enhancing oral cytological diagnosis.
Attention mechanism, which is a means of determining which part of the forced data is emphasized, has attracted attention in various fields of deep learning in recent years. The purpose of this study was to evaluate the performance of the attention branch network (ABN) for implant classification using convolutional neural networks (CNNs). The data consisted of 10191 dental implant images from 13 implant brands that cropped the site, including dental implants as pretreatment, from digital panoramic radiographs of patients who underwent surgery at Kagawa Prefectural Central Hospital between 2005 and 2021.
View Article and Find Full Text PDFArticle Synopsis
- - This study assesses the effectiveness of convolutional neural networks (CNNs) in classifying mandibular third molars using panoramic X-rays according to Pell and Gregory and Winter’s classifications.
- - Researchers analyzed 1,330 images of third molars and found that single-task learning outperformed multi-task learning in diagnostic accuracy, indicated by significant statistical results (p < 0.05).
- - The findings suggest that applying these specific classifications through deep learning models can improve the accuracy of identifying the positioning of mandibular third molars, representing a pioneering effort in this area.
Article Synopsis
- * Researchers analyzed 9,767 dental implant images from 12 different brands and stages, using five deep convolutional neural network (CNN) models to evaluate classification performance.
- * Results showed that deeper networks with more parameters improved classification accuracy, and multi-task learning enhanced brand classification and significantly boosted metrics for treatment phase classification.
Article Synopsis
- The study focused on using panoramic X-ray images to assess the accuracy of various dental implant brands through deep convolutional neural networks (CNNs) using transfer learning techniques.
- A total of 8,859 implant images from 11 different implant systems were analyzed, sourced from patients who received dental implants at a hospital in Japan between 2005 and 2019.
- Among five evaluated CNN models, the finely tuned VGG16 model achieved the best classification performance for dental implants, followed by the finely tuned VGG19, indicating their effectiveness in distinguishing between the different implant systems.