Improvement of Vibration and End-to-End Latency for Optical Motion Detection System in Magnetic Resonance Imaging
This dissertation presents improvements to both software and hardware for a marker-less optical motion detection system used in magnetic resonance imaging. Multiple devices have been developed to eliminate artifacts in MRI caused by patient movement, as motion is a common cause of artifact. Most developed devices require attaching pieces to a subject’s body; however, our optical motion detection system is attachment-free. Previously, we successfully reduced imaging artifacts retrospectively using carotid artery MR data. We used gathered motion data to determine beginning and ending times of the movement, then discarded the corresponding k-space lines within the movement duration to eliminate artifacts. Now we would like to push this process from retrospective correction to real time. However, the camera and laser mounting vibration, as well as the system time delay variation, generate inaccuracies which make it difficult to find the exact k-space lines to be removed. To make the artifact-eliminating process more efficient, we reduced the overall end-to-end latency from 110ms 21ms to 100ms 12ms by coding the motion processing program in Python with OpenCV library. We also designed a new laser and camera holder using 3D printing with PLA for better stability. The noise and drift in both x and y directions are decreased with the new holder compared to the previous setup. The same result for decrease in vibration was seen during phantom scanning with T1-weighted (TR = 100ms) gradient echo (GRE) and with T1-weighted turbo spin-echo (TSE) quadruple inversion-recovery (QIR) (TR = 800ms) sequence. Furthermore, the 3D printed holder decreases the overall assembly time required to set up the system. Overall, the new design of the system reduces the event-to-display latency, the latency variation, the hardware vibration, and the assembly time. The new design improves the old design and provides a path to pursue prospective optical motion correction.
- Bioengineering