An Open-Source Application for Monitoring Cardiorespiratory Motion with Bi-Planar Fluoroscopic Imaging
Abstract
Purpose
To develop an open-source application capable of retrospectively monitoring the position of radiopaque markers from bi-planar fluoroscopy images. To enable investigation of the cardiorespiratory motion of implanted cardiac leads from bi-planar fluoroscopy from the VERO4DRT linac for cardiac radioablation patients.
Methods
An application was developed in MATLAB with modules for (i) loading and filtering 2D kV cine image projections, (ii) semi-automatically tracking the position of radiopaque markers using normalized cross correlation image template matching, (iii) triangulating points in 3D from bi-planar views, (iv) separating respiratory- and cardiac-induced motions using a low- and high-pass filter, respectively, and (v) saving all data for further analysis. The performance of the application was tested by simulating combined cardiorespiratory motions using an in-house cardiac motion phantom coupled with the BrainLAB ExacTrac respiratory motion phantom and imaged with 5 Hz bi-planar fluoroscopy on the VERO4DRT linac. Four experimental scenarios were performed with varying imaging angles, sinusoidal cardiac and respiratory motion amplitudes, and periods. The analyzed respiratory and cardiac motions from the monitoring application were fit to sinusoids in MATLAB to compare with the expected values.
Results
The developed open-source application is hosted on GitHub. The measured amplitude of respiratory and cardiac motion was within 0.1 mm and 0.2 mm, respectively, of that expected. Initial results suggested that the predicted period of respiratory motion deviated up to 8% from that expected. Further investigation using a digital counter revealed that the respiratory motion phantom moved more slowly than programmed, resulting in agreement between the measured and expected period within 10 ms. The measured frequency of cardiac motion was within 3 bpm of expected.
Conclusion
The developed application allows for monitoring the position of radiopaque objects from bi-planar fluoroscopy images. Future work extending the application to similar use-cases such as motion monitoring using CBCT projections, is warranted.