Objective To develop an imaging segmentation algorithm based on adaptive thresholding to be used for thyroid volume estimation in radionuclide imaging.Methods First, image preprocessing was performed on images collected by thyroid radionuclide imaging, which contained smoothness, enhancement, and grey level transformation. Second, adaptive thresholding algorithm and morphological operations were adapted to extract the rough thyroid area. Finally, the maximum height and the area of each lobe can be achieved by setting reference points so that the thyroid volume could be calculated. Results The thyroid volume obtained through radionuclide imaging was used as the reference standard. Assessment indexes such as deviation, precision, relative differences, and degree of association were selected to compare different methods used to estimate thyroid volumes. The results of the comparison indicate that the approach of thyroid volume estimation proposed in this research was not only highly correlated with the results obtained through ultrasonography (R2=0.99), the result also has the best deviation (0.9), the lowest precision (±2.14 mL), and relative differences (3.2%±4.36%).Conclusion The approach to thyroid volume estimation proposed in this research is precise, simple and convenient, and can avoid dependence on ultrasound, which can help physicians determine the individualized dosage regimen for each patient in the treatment of thyroid disease.%目的：开发一种基于自适应阈值的图像分割算法，用于核素显像中甲状腺体积的测定。方法首先对甲状腺核素显像的图像进行平滑、增强、灰度变换等预处理；然后用自适应阈值算法和形态学处理分割出甲状腺区域；最后在甲状腺区域上设置参考点，得出最大高度和每叶甲状腺面积，进而计算出甲状腺体积。结果以超声测得的甲状腺体积作为参考标准，选取偏差、精度、相对误差、相关度等评价指标对不同的甲状腺体积测定方法进行比较。不同方法的对比分析结果显示，基于本研究方法所测定的甲状腺体积不仅与超声测定结果高度相关（R2=0.99），且精度（±2.14 mL）最高，测量偏差（0.9）、相对误差（3.2%±4.36%）最低。结论本研究开发的甲状腺体积测定方法精确、简便，有助于核医学科医生为患者制定个体化剂量方案。
展开▼
