Automatic detection of the mental foramen for estimating mandibular cortical width in dental panoramic radiographs: the seventh survey of the Tromsø Study (Tromsø7) in 2015–2016 ...

• Main Authors: Edvardsen, Isak Paasche, Teterina, Anna, Johansen, Thomas, Myhre, Jonas Nordhaug, Godtliebsen, Fred, Bolstad, Napat Limchaichana
• Format: Article in Journal/Newspaper
• Language: unknown
• Published: SAGE Journals 2022
• Subjects: 111708 Health and Community Services; FOS: Health sciences; Cardiology; 110599 Dentistry not elsewhere classified; FOS: Clinical medicine; 110305 Emergency Medicine; 110306 Endocrinology; 110308 Geriatrics and Gerontology; 110904 Neurology and Neuromuscular Diseases; 111099 Nursing not elsewhere classified; 111199 Nutrition and Dietetics not elsewhere classified; 111299 Oncology and Carcinogenesis not elsewhere classified; 110314 Orthopaedics; 110604 Sports Medicine; 110315 Otorhinolaryngology; 111702 Aged Health Care; 111403 Paediatrics; 111599 Pharmacology and Pharmaceutical Sciences not elsewhere classified; 170199 Psychology not elsewhere classified; FOS: Psychology; 110319 Psychiatry incl. Psychotherapy; 111799 Public Health and Health Services not elsewhere classified; 110203 Respiratory Diseases; 160807 Sociological Methodology and Research Methods; FOS: Sociology; 110323 Surgery; Tromsø
• Online Access: https://dx.doi.org/10.25384/sage.c.6314472; https://sage.figshare.com/collections/Automatic_detection_of_the_mental_foramen_for_estimating_mandibular_cortical_width_in_dental_panoramic_radiographs_the_seventh_survey_of_the_Troms_Study_Troms_7_in_2015_2016/6314472

ObjectiveTo apply deep learning to a data set of dental panoramic radiographs to detect the mental foramen for automatic assessment of the mandibular cortical width.MethodsData from the seventh survey of the Tromsø Study (Tromsø7) were used. The data set contained 5197 randomly chosen dental panoramic radiographs. Four pretrained object detectors were tested. We randomly chose 80% of the data for training and 20% for testing. Models were trained using GeForce RTX 2080 Ti with 11 GB GPU memory (NVIDIA Corporation, Santa Clara, CA, USA). Python programming language version 3.7 was used for analysis.ResultsThe EfficientDet-D0 model showed the highest average precision of 0.30. When the threshold to regard a prediction as correct (intersection over union) was set to 0.5, the average precision was 0.79. The RetinaNet model achieved the lowest average precision of 0.23, and the precision was 0.64 when the intersection over union was set to 0.5. The procedure to estimate mandibular cortical width showed acceptable ...