Room: Room 205
Purpose: To evaluate the effect of reference CT slice thickness on the accuracy of markerless motion tracking (MMT) for single energy (SE) and dual energy (DE) fluoroscopy.
Methods: A fast-kV-switching DE imaging system was implemented on a bench top. SE and DE images were acquired on a static phantom with tumors ranging from 0.5-2.5cm, as well as with a programmable motion phantom. Bone-suppressed images, obtained from the DE fluoroscopic sequences, were created using logarithmic subtraction. Separately, a template-based matching algorithm was used to track tumor location on both sequences. Templates were generated using CT scans of each phantom. The CT reconstruction slice thickness was varied from 0.75-3.00mm. A match score surface was calculated by shifting the template across the image and calculating the normalized cross correlation (NCC). The offset at which the NCC has a maximum value represents the potential target position. The strength of this peak relative to NCC values away from the peak, called side lobe values, is quantified by the peak-to-side lobe ratio (PSR).
Results: The maximum slice thickness required to achieve a tracking success rate of > 90% was evaluated vs. tumor dimension for the static phantom. As expected smaller tumors required a finer slice thickness. However, the maximum slice thickness was generally larger for DE vs. SE. For SE imaging the tumor size/slice thickness ratio was nearly a constant of 10, whereas for DE imaging the ratio was ~7. For the motion phantom, DE images had a larger PSR vs. SE imaging, indicating a higher quality template match. For this target a slice thickness of â‰¤1.5 mm was needed to achieve a stable PSR value.
Conclusion: The maximum CT slice thickness for template tracking is a function of the tumor size. For most smaller tumors a CT slice thickness of â‰¤1.5-2.0mm is required for MMT.
Funding Support, Disclosures, and Conflict of Interest: Supported by NIH R01CA207483.