Radiation and Cancer Physics

SS 37 - Physics 11 - Online Imaging and Motion Management

266 - Characterization of Markerless Motion Tracking using Fast-Switching Dual Energy Fluoroscopy

Wednesday, October 24
11:20 AM - 11:30 AM
Location: Room 302

Characterization of Markerless Motion Tracking using Fast-Switching Dual Energy Fluoroscopy
J. C. Roeske1, H. Mostafavi2, M. Haytmyradov1, A. Wang2, M. Surucu1, L. Zhu2, R. Patel1, and M. M. Harkenrider1; 1Department of Radiation Oncology, Stritch School of Medicine, Loyola University Chicago, Maywood, IL, 2Varian Medical Systems, Palo Alto, CA

Purpose/Objective(s): Template-based matching algorithms are currently being considered for markerless motion tracking (MMT) of lung tumors using single energy (SE) fluoroscopy. A limitation of these algorithms is that tracking may not be accurate when there is significant overlap of tumor/bone on planar images. In these cases, dual energy (DE) imaging may be used to create a soft tissue only image, potentially improving the accuracy of MMT. The goal of this study is to evaluate the accuracy of MMT using a fast-kV-switching DE imaging system.

Materials/Methods: A fast-kV-switching real-time DE imaging system, emulating a clinical on-board imager, was implemented on a bench top that included an x-ray tube, amorphous silicon (a-Si) flat panel detector, and x-ray generator with custom firmware and software. A programmable motion phantom, consisting of a torso with embedded ribs and spine along with a simulated tumor in a lung-type cavity, was used. SE and DE images were obtained at fixed angles as well as over 360 degrees of rotation allowing the tumor projection to overlap with varying amounts of bone. Soft tissue images obtained from the DE fluoroscopic sequences were created using a frame-by-frame weighted logarithmic subtraction. Separately, MMT requires a template that was derived from the contoured simulated tumor obtained from a CT scan of the phantom. A template-based matching algorithm was then used to track tumor motion throughout the DE and SE fluoroscopy sequences. This algorithm shifts the template across the image and calculates the normalized cross correlation (NCC) between the template and image, resulting in a match score surface. 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). Previous studies showed higher PSR values are correlated with tracking accuracy.

Results: For fixed gantry imaging, MMT was able to track 296/380 (78%) of SE images. Tracking failure was due to tumor/bone overlap. The removal of bone through DE imaging increased the success rate to 379/380 (99%) (p < 0.0001). The average PSR was 4.28 +/- 1.04 and 6.25 +/-1/41 for SE and DE, respectively (p < 0.0001). For the rotational acquisition, the number of images that the algorithm was able to track was 414/450 (92%) vs. 419/450 (93%) for SE and DE, respectively (p=0.61). However, the PSR values were 5.06 +/- 1.13 and 6.30 +/- 1.46 for SE and DE, respectively (p < 0.0001). Tracking accuracy was determined by comparing the ability of MMT to reproduce the programmed waveform. The root mean square error (RMSE) of tracked vs. the programmed waveform was 0.24 +/- 0.11 mm for SE vs. 0.15 +/- 0.04 mm for DE (p = 0.03).

Conclusion: A fast-kV-switching real-time DE prototype has been implemented and utilized to provide DE imaging for MMT. The principle advantage of this approach is the ability to accurately track tumors where MMT fails on SE fluoroscopy due to overlapping bony anatomy.

Author Disclosure: J.C. Roeske: Research Grant; Varian Medical Systesm. Speaker's Bureau; Varian Medical Systesm. Travel Expenses; Varian Medical Systesm. H. Mostafavi: Stock; Varian Medical Systems. M. Haytmyradov: None. A. Wang: Stock; Varian Medical Systems. M. Surucu: Secretary; AAPM Midwest Chapter. L. Zhu: Stock; Varian Medical Systems. R. Patel: None. M.M. Harkenrider: Radiation oncology program director and Trustee; Chicago Radiological Society.

John Roeske, PhD

Loyola University Medical Center

Loyola Univ Med Center: Professor and Director of Physics: Employee

Varian Medical Systesm: Advisory Board, Research Grants, Speaker's Bureau, Travel Expenses

John C. Roeske, PhD is a Professor and Chief of the Medical Physics Division in the Department of Radiation Oncology at Loyola University Medical Center and the Cardinal Bernadin Cancer Center. He earned his PhD at the University of Chicago where his thesis work focussed on the dosimetry of radioimmunotherapy. An NIH-funded researcher, Dr. Roeske's current intestests include dual energy imaging, markerless motion tracking and image-guided radiotherapy.


Send Email for John Roeske


266 - Characterization of Markerless Motion Tracking using Fast-Switching Dual Energy Fluoroscopy


Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Characterization of Markerless Motion Tracking using Fast-Switching Dual Energy Fluoroscopy