Evaluation of multi-task learning in deep learning-based positioning classification of mandibular third molars.

Pell and Gregory, and Winter's classifications are frequently implemented to classify the mandibular third molars and are crucial for safe tooth extraction. This study aimed to evaluate the classification accuracy of convolutional neural network (CNN) deep learning models using cropped panoramic radiographs based on these classifications. We compared the diagnostic accuracy of single-task and multi-task learning after labeling 1330 images of mandibular third molars from digital radiographs taken at the Department of Oral and Maxillofacial Surgery at a general hospital (2014-2021). The mandibular third molar classifications were analyzed using a VGG 16 model of a CNN. We statistically evaluated performance metrics [accuracy, precision, recall, F1 score, and area under the curve (AUC)] for each prediction. We found that single-task learning was superior to multi-task learning (all p < 0.05) for all metrics, with large effect sizes and low p-values. Recall and F1 scores for position classification showed medium effect sizes in single and multi-task learning. To our knowledge, this is the first deep learning study to examine single-task and multi-task learning for the classification of mandibular third molars. Our results demonstrated the efficacy of implementing Pell and Gregory, and Winter's classifications for specific respective tasks.

Endoscopic-assisted resection of pleomorphic adenoma in the accessory parotid gland.

Background : An accessory parotid gland (APG) is a common anatomical structure that occurs in 10%-56% of individuals. Pleomorphic adenomas are the most common benign tumors of the APG, and their ideal treatment is surgical excision, although there is a risk for aesthetic disorders and facial nerve damage due to the site of origin. Moreover, despite being benign, these tumors are known to recur. Therefore, it is necessary to achieve both reliable excision and avoidance of facial nerve damage. Case presentation : We report a case of a 49-year-old Japanese man with a mass in his left cheek. The lesion was diagnosed as a benign salivary gland tumor derived from the APG by computed tomography imaging, magnetic resonance imaging and fine needle aspiration cytology. We resected the tumor using modified high submandibular incision under the endoscopic-assisted field of view. Discussion and Conclusions : The tumor was less invasive and reliably resected using an endoscope. In surgical treatment, the endoscopic-assisted technique is very useful to achieve complete tumor resection and prevent relapse while avoiding serious complications due to surgical procedures. J. Med. Invest. 68 : 376-380, August, 2021.

Multi-Task Deep Learning Model for Classification of Dental Implant Brand and Treatment Stage Using Dental Panoramic Radiograph Images.

It is necessary to accurately identify dental implant brands and the stage of treatment to ensure efficient care. Thus, the purpose of this study was to use multi-task deep learning to investigate a classifier that categorizes implant brands and treatment stages from dental panoramic radiographic images. For objective labeling, 9767 dental implant images of 12 implant brands and treatment stages were obtained from the digital panoramic radiographs of patients who underwent procedures at Kagawa Prefectural Central Hospital, Japan, between 2005 and 2020. Five deep convolutional neural network (CNN) models (ResNet18, 34, 50, 101 and 152) were evaluated. The accuracy, precision, recall, specificity, F1 score, and area under the curve score were calculated for each CNN. We also compared the multi-task and single-task accuracies of brand classification and implant treatment stage classification. Our analysis revealed that the larger the number of parameters and the deeper the network, the better the performance for both classifications. Multi-tasking significantly improved brand classification on all performance indicators, except recall, and significantly improved all metrics in treatment phase classification. Using CNNs conferred high validity in the classification of dental implant brands and treatment stages. Furthermore, multi-task learning facilitated analysis accuracy.

Oriented immobilization of basic fibroblast growth factor: Bioengineered surface design for the expansion of human mesenchymal stromal cells.

E. coli expressed recombinant basic fibroblast growth factor (bFGF) with histidine-tag (bFGF-His) was immobilized onto the surface of a glass plate modified with a Ni(II)-chelated alkanethiol monolayer. The immobilization is expected to take place through the coordination between Ni(II) and His-tag. The bFGF-immobilized surface was exposed to citrate buffer solution to refold in situ the surface-immobilized bFGF. The secondary structure of immobilized bFGF-His was analyzed by solid-phase circular dichroism (CD) spectroscopy. Immortalized human mesenchymal stromal cells (hMSCs) were cultured on the bFGF-His-immobilized surface to examine their proliferation. CD spectroscopy revealed that the immobilized bFGF initially exhibited secondary structure rich in α-helix and that the spectrum was gradually transformed to exhibit the formation of ß-strands upon exposure to citrate buffer solution, approaching to the spectrum of native bFGF. The rate of hMSC proliferation was 1.2-fold higher on the bFGF-immobilized surface treated with in situ citrate buffer, compared to the polystyrene surface. The immobilized bFGF-His treated in situ with citrate buffer solution seemed to be biologically active because its secondary structure approached its native state. This was well demonstrated by the cell culture experiments. From these results we conclude that immobilization of bFGF on the culture substrate serves to enhance proliferation of hMSCs.