Division of Biology and Medicine
BioMed Core Facilities

XROMM Service

Overview

The W.M. Keck Foundation XROMM Facility biplanar room at Brown University consists of two Varian model G-1086 x-ray tubes, two EMD Technologies model EPS 45-80 pulsed x-ray generators, two Dunlee model TH9447QXH590 image intensifiers (16" diameter), and two Phantom v10 high-speed digital video cameras. The x-ray tubes are suspended from the ceiling by tube cranes and the IIs are mounted on mobile gantries. The components are set up such that the two x-ray beams intersect each other close to the IIs. A treadmill, trackway or other animal support/enclosure is placed such that the research animal performs the behavior of interest (running, jumping, flying) in the volume where the x-ray beams intersect.

The Brown University system can deliver pulsed x-ray generation at up to 100 Hz, and can record in continuous x-ray generation mode at up to 1000 fps. With the Phantom v10 cameras, the pixel resolution is 1800x1800. Overall resolution of the imaging chain is about 2 line pairs/mm. Radio-opaque beads can be tracked to within ±0.1 mm in 3D space. The x-ray system was designed and integrated by Marty Kulis of Imaging Systems and Service, Painesville OH (mkulis@issi-na.com; 440-724-8002).

Bone morphology data come from a 3D computer model of the bone surf3D Modelaces from CT, laser scanning, or MRI. Each bone is an object that can be manipulated individually in computer animation space. These models are specific to each individual study subject (human or animal).

 

Rotating 3D Models
3D models of pig skull and lower jaw from a CT scan. XROMM requires independent models of each bone for re-animation. Models created in Amira, cleaned in Geomagic, and animated in Autodesk Maya. Animation by E. Brainerd.

Bone motion data come from twXRAY Movieo high-speed, biplanar x-ray movies. In some studies it is possible to surgically implant radiopaque beads into the bones for marker-based XROMM analysis. In other studies, markerless XROMM analysis must be used. In markerless XROMM, the bone models are aligned to the x-ray images, either manually or using an autoregistration algorithm (e.g. You et al., 2001).


X-ray video setup
Biplanar x-ray video data collection for a study of mastication in minipigs. Two C-arm fluoroscopes retrofitted with high-speed video cameras collect x-ray video from two perspectives. The resulting images can be seen on the computer monitors in the background. Video by E. Brainerd.

Lateral view movie
Minipig mastication in lateral x-ray projection. Bones appear dark and air appears white in this x-ray positive movie. Video was recorded at 250 frames per second and is played back here at approximately 1/4 real speed. Video by K. Metzger.

Ventro-dorsal view movie
Minipig mastication in ventro-dorsal x-ray projection. Bones appear dark and air appears white in this x-ray positive movie. Video was recorded at 250 frames per second and is played back here at approximately 1/4 real speed. Video by K. Metzger.

ReanimationThe data on bone shape and bone movement are combined into an XROMM animation. Rigid body kinematics from the x-ray movies are applied to the bone models to re-animate the actual movement that was performed by the individual subject at the time of recording.

 

Lateral view movie
Precise re-animation of the CT bone models to match the movement recorded in the x-ray movies. In this study, small metal markers were used to determine the correct pose and position of the bones in 3D space (to within ±0.1 mm). Maya animation by D. Baier.

Ventro-dorsal view movie
Precise re-animation of the CT bone models to match the movement recorded in the x-ray movies (ventro-dorsal view). Note that the whole bone does not have to be in view to reconstruct the position of every point on a rigid bone. Maya animation by D. Baier.

Operation