在手持二维超声探头扫描重建三维超声图像之前,需要进行超声探头标定.为了在标定图像数量少的情况下,快速有效地标定超声探头,提出一种基于光学定位的超声探头标定方法.设计一种由两个互相垂直的平面组成的立体标定模板.在标定过程中,手持超声探头扫描标定模板,由光学定位设备获取摄像机与模板的变换矩阵、超声探头与摄像机的变换矩阵.每次扫描得到超声声束与模板三正交轴的3个交点,共构造6个标定方程.当采集两幅以上标定图像时,即可非线性优化求解得到超声图像与超声探头的变换矩阵,未知量包括6个空间变量和两个超声图像比例因子.仿真实验表明,标定误差随图像数量增加而减小,当图像数量大于30幅时,标定平均误差小于1 mm;真实实验表明,该方法在采用20幅以上标定图像时,标定平均误差小于2 mm.该方法平均标定精度满足超声影像导航手术的需求,并为进一步的三维超声重建奠定基础.
Ultrasound probe calibration is a necessary step in the process of 3D ultrasound imaging by a freehand 2D ultrasound probe. In order to calibrate ultrasound probe quickly and effectively with small number of calibration images, a flexible calibration method based on optical tracker was presented for determining the unknown transformation matrix. First, a calibration pattern containing two mutually perpendicular planes .was developed. In the process of calibration, the calibration pattern was scanned by a freehand 2D ultrasound probe. Transformation matrice were calculated from camera to calibration pattern and from ultrasound probe to ultrasound image. There were 3 crosspoints between ultrasound beam and three orthogonal axes of the pattern in each trial, which setup 6 calibration equations. The unknown transformation matrix and 2 scaling factors were solved by nonlinear fitting, when at least 2 calibration images were captured. Experiments with synthetic data show that calibration error decreases with increased number of images. The error was 1 mm when the number of images was 30. Experiments with real data show that our method has high accuracy with only 20 calibration images, meeting the demand of ultrasound surgical navigation and lays the groundwork for 3D ultrasound imaging reconstruction.