面向牙齿X-RAY图像分割的牙齿模型

Research on Tooth X-Ray Image Segmentation Based on U-Net

医学图像分割技术为临床诊断和治疗提供关键依据,但传统的图像分割技术易将图像过度分割且分割效果不够精细。为解决这些问题,本文研究采用U-Net网络分割牙齿X-Ray图像,通过合理选择池化操作、激活函数和周期数目,解决牙齿X-Ray图像中存在的牙齿与周围组织的对比度低、边界模糊、牙齿与背景分布不均、牙齿与组织粘连等问题。实验结果表明,U-Net网络具备有效的牙齿X-Ray图像的分割性能。

Medical image segmentation technology provides key basis for clinical diagnosis and treatment,but traditional image segmentation techniques are prone to over segmentation of images and the segmentation effect is not precise enough.To address these issues,this article investigates the use of U-Net network for segmenting dental X-Ray images.By selecting pooling operations,activation functions,and the number of epochs reasonably,problems such as low contrast between teeth and surrounding tissues,blurred boundaries,uneven distribution of teeth and background,and adhesion between teeth and tissues in dental X-ray images can be solved.The experimental results indicate that the U-Net network has effective segmentation performance for dental X-ray images.