基于小波变换和Canny算子的膈肌运动规律自动检测方法研究

Automatic detection algorithm of the diaphragm motion based on Canny edge detection and wavelettransform

目的 探索基于小波变换和Canny算子的膈肌运动规律自动检测方法.方法 采用小波变换对实时采集的患者自由呼吸状态下的胸部透视序列图像进行增强,对增强后的图像使用Canny算子实现膈肌边缘提取,通过Matlab编程跟踪患者图像中膈肌的位置变化,建立患者膈肌的呼吸运动曲线.结果 平静自由呼吸状态下,自动检测膈肌运动规律曲线与手工测量的曲线在幅度和周期上比较一致.患者在一次XVI Motion-ViewTM透视过程中约包含6～7个呼吸周期.膈肌头脚方向运动幅度不完全一致,大小为6.7 ～8.0 mm,平均为7.4 mm.同一患者不同分次间呼吸周期差异不明显,但在患者情绪激动或者由于自身原因如咳嗽等引起的剧烈呼吸运动时,则会产生明显差异.结论 与手工测量方法相比,小波变换和Canny算子的膈肌检测方法能自动有效检测患者呼吸运动变化,分析时间短、精度高.

Objective To develop a new automatic detection algorithm of the diaphragm motion based on Canny edge detection and wavelet transform. Methods On-line fluoroscopic images under free breathing were enhanced by using the wavelet transform. After the wavelet transform, edge detection was carried out for the enhanced image. Canny edge detection algorithm was used to achieve the diaphragm edge. Programs were written in Matlab to track the position of the diaphragm. The diaphragm movement curves were derived to evaluate the characteristics of patients respiratory motion. Results Under calm free breathing ,the amplitude and period of diaphragm motion acquired by means of the wavelet transform and Canny edge detection were in good agreement with manual measurement. There were six to seven respiratory cycles in a XVI Motion- ViewTM. The magnitude of diaphragm movement was not exactly the same in the eranio-caudal （CC） direction. The magnitude was from 6. 7 mm to 8. 0 mm with an average of 7. 4 mm. The movements of the respiratory motion cycles had little variations in amplitude and period for the same patient between fractions except emotional excitement or cough. Conclusions The automatic diaphragm detection methods developed in this paper are precise, and can effectively reflect the characteristics of the respiratory motion. The method can save much time and improve the measure precision greatly compared with the manual measurement.