基于Hessian矩阵及梯度熵的疑似肺结节检测算法

Suspected pulmonary nodule detectien algorithm based on Hessian matrix and grads entropy

提出一种基于Hessian矩阵的多尺度增强滤波和梯度熵选择的疑似肺结节检测算法。首先,构造基于Hessian矩阵的多尺度圆形增强滤波器对肺部计算机断层扫描(CT)图像中的圆形区域进行增强,使得类似圆形的疑似结节区域得以增强,其他干扰区域得以抑制。然后,计算增强后的圆形肺结节区域(孤立圆、线交叉形成的圆)在原始灰度图像中相应区域的梯度熵,采用切比雪夫不等式确定孤立圆梯度熵分布区间,进行疑似结节区域的选择。实验结果表明,该算法能较好地检测肺部CT图像中疑似肺结节区域,为后续的肺结节检测以及诊断打下基础。

A suspected pulmonary nodule detection algorithm is presented based on multi-scale enhancement filtering of Hessian matrix and selecting of grads entropy. Firstly, the multi-scale circular filter based on Hessian matrix is constructed to enhance the circular area of the pulmonary computed tomography （CT） images, which enhances the circular suspected pulmonary nodule and suppresses other interfering areas. Secondly, the grads entropy of the area of the original gray image corresponding to the area of the enhanced circle （the solitary circle and the circle formed by the intersection of the lines） is calculated, and the range of the grads entropy of the solitary circle confirmed by Chebyshev inequality is used to select the suspected pulmonary nodules. Experiment results indicate that the algorithm detects the suspected pulmonary nodule accurately, which is a basis for further pulmonary nodule detection and diagnose.