White Spot and Fluorotic Lesions Detection On Dental Images Using Deep Learning

سال انتشار: 1400
نوع سند: مقاله کنفرانسی
زبان: انگلیسی
مشاهده: 244

متن کامل این مقاله منتشر نشده است و فقط به صورت چکیده یا چکیده مبسوط در پایگاه موجود می باشد.
توضیح: معمولا کلیه مقالاتی که کمتر از ۵ صفحه باشند در پایگاه سیویلیکا اصل مقاله (فول تکست) محسوب نمی شوند و فقط کاربران عضو بدون کسر اعتبار می توانند فایل آنها را دریافت نمایند.

این مقاله در بخشهای موضوعی زیر دسته بندی شده است:

استخراج به نرم افزارهای پژوهشی:

لینک ثابت به این مقاله:

شناسه ملی سند علمی:

UTCONF05_178

تاریخ نمایه سازی: 13 تیر 1400

چکیده مقاله:

Dentists usually face dental tissue anomalies which are difficult to distinguish from each other.These anomalies require fundamentally different treatment. In clinical applications, diagnosing the correct types of anomalies at a high classification rate is of importance. Deep learning has been employed in numerous applications such as medical image segmentation and analysis. In this paper, we utilize a commonly used deep neural structures named convolutional neural network (CNN) to classify dental photographs in order to detect white spot and fluorotic lesions. We captured ۷۱۰ photos from ۱۰۲ patients who were referred to dentistry clinics. ۴۰۰۰ cropped dental images were collected and labeled by three independent dentists. All the labels were reviewed by an independent expert. We assume four classes including: ۱) normal, ۲) white spot, ۳) fluorosis and ۴) other to classify images. Histogram equalization is used to minimize light reflection which is the main reason for misclassification. Three pre-trained classification models including AlexNet, GoogleNet and SqueezeNet are employed to classify images. We performed ten-fold cross validation to evaluate our deep learning models. The average classification performances were ۸۷.۴%, ۸۳.۶% and ۹۲.۵% for AlexNet, GoogleNet and SqueezeNet. Deep learning-based methods showed significant results to detect the most common lesions in dentistry. Considering our small dataset, results were comparable and our proposed methods can be used in future generations of decision support systems to help dentists.

نویسندگان

Rosana Farjaminejad

Faculty of Dentistry, Tehran Medical Science, Islamic Azad University, Tehran, Iran

Morteza Zangeneh Soroush

Assistant Professor, Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran Department of Biomedical Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran