Philippe P. Bruyant

Feasibility of stereo-infrared tracking to monitor patient motion during cardiac SPECT imaging

Patient motion during cardiac SPECT imaging can cause diagnostic imaging artifacts. We investigated the feasibility of monitoring patient motion using the Polaris motion-tracking system. This system uses passive infrared reflection from small spheres to provide real-time position data with vendor stated 0.35 mm accuracy and 0.2 mm repeatability. In our configuration, the Polaris system views through the SPECT gantry toward the patients head. List-mode event data were temporally synchronized with motion-tracking data utilizing a modified LabVIEW virtual instrument that we have employed in previous optical motion-tracking investigations. Calibration of SPECT to Polaris coordinates was achieved by determining the transformation matrix necessary to align the position of four reflecting spheres as seen by Polaris, with the location of Tc-99m activity placed inside the sphere mounts as determined in SPECT reconstructions. We have successfully tracked targets placed on volunteers in simulated imaging positions on the table of our SPECT system. We obtained excellent correlation (R/sup 2/>0.998) between the change in location of the targets as measured by our SPECT system and the Polaris. We have also obtained excellent agreement between the recordings of the respiratory motion of four targets attached to an elastic band wrapped around the abdomen of volunteers and from a pneumatic bellows. We used the axial motion of point sources as determined by the Polaris to correct the motion in SPECT image acquisitions yielding virtually identical point source full-width at half-maximum and full-width at tenth-maximum values, and profiled maximum heart wall counts of cardiac phantom images, compared to the reconstructions with no motion.

Correction of the respiratory motion of the heart by tracking of the center of mass of thresholded projections: a simulation study using the dynamic MCAT phantom

During normal breathing, heart motion is about 15 mm along the body axis in humans. We propose a method to track and to correct this motion after a list-mode acquisition which involves the recording of a signal proportional to respiratory volume. We use the dynamic MCAT (DMCAT) chest phantom to simulate 24 temporal frames regularly spaced during the respiratory cycle, for 60 projection angles over 360/spl deg/. A 15-mm respiratory translation motion is simulated for the heart, liver and spleen. Thresholding of projections is used to reduce the influence of static activity on calculation of the axial center-of-mass (aCOM). Variation in the impact of attenuation as a function of projections and noise in the low-count projections rebinned from list-mode acquisitions is seen to limit ones ability to track respiratory motion using the aCOM. By including the recording of a signal proportional to the relative respiratory volume with the list-mode acquisition counts from different respiratory cycles can be combined to produce projections with common respiratory volumes. We have determined that the aCOMs determined from summing these common-volume based projections over the anterior to left-anterior oblique projection angles can be used to track respiratory motion as a function of the volume signal. Using this information on the variation of the aCOM as a function of the volume signal, the entire list-mode acquisition can then be rebinned into a projection set which is corrected for respiratory motion. After motion tracking, the mean absolute difference between the true motion curve and the aCOM curve is 0.10 cm for noisy studies. After correction no heart motion is visible on a cine display of projections. The polar map of myocardial MIBI uptake after motion correction is closer to that obtained when no respiratory motion is present than without correction.

A robust visual tracking system for patient motion detection in SPECT: hardware solutions

Our overall research goal is to devise a robust method of tracking and compensating patient motion by combining an emission data based approach with a visual tracking system (VTS) that provides an independent estimate of motion. Herein, we present the latest hardware configuration of the VTS, a test of the accuracy of motion tracking by it, and our solution for synchronization between the SPECT and the optical acquisitions. The current version of the VTS includes stereo imaging with sets of optical network cameras with attached light sources, a SPECT/VTS calibration phantom, a black stretchable garment with reflective spheres to track chest motion, and a computer to control the cameras. The computer also stores the JPEG files generated by the optical cameras with synchronization to the list-mode acquisition of events on our SPECT system. Five Axis PTZ 2130 network cameras (Axis Communications AB, Lund, Sweden) were used to track motion of spheres with a highly retroreflective coating using stereo methods. The calibration phantom is comprised of seven reflective spheres designed such that radioactivity can be added to the tip of the mounts holding the spheres. This phantom is used to determine the transformation to be applied to convert the motion detected by the VTS into the SPECT coordinates system. The ability of the VTS to track motion was assessed by comparing its results to those of the Polaris infrared tracking system (Northern Digital Inc., Waterloo, ON, Canada). The difference in the motions assessed by the two systems was generally less than 1 mm. Synchronization was assessed in two ways. First, optical cameras were aimed at a digital clock and the elapsed time estimated by the cameras was compared to the actual time shown by the clock in the images. Second, synchronization was also assessed by moving a radioactive and reflective sphere three times during concurrent VTS and SPECT acquisitions and comparing the time at which motion occurred in the optical and SPECT images. The results show that optical and SPECT images stay synchronized within a 150-ms range. The 100-Mbit network load is less than 10%, and the computers CPU load is between 15% and 25%; thus, the VTS can be improved by adding more cameras or by increasing the image size and/or resolution while keeping an acquisition rate of 30 images per second per camera.

Optimization of iterative reconstructions of /sup 99m/Tc cardiac SPECT studies using numerical observers

In this paper, we investigate the use of a numerical observer to optimize ordered-subset expectation maximization (OSEM) reconstructions for the detection of coronary artery disease (CAD). The parameters optimized were the iteration number and the full-width at half-maximum of three-dimensional Gaussian postfiltering. The numerical observer employed in the optimization was the channelized Hotelling observer (CHO). The CHO had been used previously to rank tumor detection accuracy for different reconstruction strategies in Ga-67 images, showing good agreement with the rankings of human observers. The intent of this paper was to determine if this CHO could also be employed for the detection of CAD. Results indicate that when grayscale (quantized) images are used, the CHO optimization results correlate well with human observers. On the other hand, when the CHO was used with floating-point images, it provided very good detection performance even when the images were excessively filtered. This result was at odds with the human-observer performance which showed a decrease in detection accuracy with highly smoothed images. This reflects the need to better model the detection task of the human observers who usually view and rank grayscale images and by appropriately modeling the image noise that quantization introduces, we show that the CHO can better match human-observer detection performance.

Numerical observer study of MAP-OSEM regularization methods with anatomical priors for lesion detection in /sup 67/Ga images

With the increasing number of multimodality systems, registered anatomical data is becoming available for use as patient-specific attenuation maps and anatomical priors in iterative reconstruction. The goal of this work is to investigate whether the introduction of anatomical information in MAP-EM reconstruction can improve lesion detection in /sup 67/Ga images of the chest region. Specifically we investigated three hypotheses. The first hypothesis was that use of information on organ boundaries in MAP-EM will increase the accuracy of tumor detection. The second hypothesis was that use of lesion boundaries in addition to organ boundaries will increase the detection accuracy of lesions when the lesion actually has an elevated concentration of activity compared to background. The third hypothesis was that use of the organ boundary prior in MAP-EM results in improved lesion detection accuracy for lesions with an elevated activity concentration even when the lesion boundary prior was also used for sites where there was not an elevated activity concentration. These hypotheses were investigated using Monte Carlo simulated projections of the mathematical cardiac-torso (MCAT) phantom. The organ and lesion boundaries for use as anatomical priors were obtained from segmentation of the original MCAT activity slices. The priors were used to determine voxel inclusion in the neighborhood of a voxel for a quadratic smoothing prior employed in De Pierros MAP-OSEM reconstruction algorithm. Two contrasts for lesion/background were investigated, 12/1, and 22.5/1. Three values for /spl beta/, the parameter controlling the weighting of the prior, were also investigated. A numerical observer was used to determine the average lesion detection accuracy for multiple sites throughout the mediastinum. The area under the ROC curve (AUC) for the numerical observer was used as the metric for detection accuracy. No evidence was observed in support of the first hypothesis. That is, use of the anatomical prior including just organ boundaries did not tend to increase the AUC over that of not using priors. Evidence was found in support of the second hypothesis however. That is, especially for the lower contrast lesions, the AUC increased when the lesion boundary was included in the prior. Finally, evidence was also observed in support of the third hypothesis. That is, again especially at the lower contrast, the use of the anatomical prior increased detection accuracy for lesions with increased uptake even when the prior was also used for matching sites with no elevation in uptake. Thus, use of the anatomical prior for lesion boundary when there was no elevated lesion uptake did not tend to increase the false-positive rate as one might have guessed would happen.

Estimation of the Rigid-Body Motion From Three-Dimensional Images Using a Generalized Center-of-Mass Points Approach

We present an analytical method for the estimation of rigid-body motion in sets of three-dimensional (3-D) SPECT and PET slices. This method utilizes mathematically defined generalized center-of-mass points in images, requiring no segmentation. It can be applied to compensation of the rigid-body motion in both SPECT and PET, once a series of 3-D tomographic images are available. We generalized the formula for the center-of-mass to obtain a family of points comoving with the objects rigid-body motion. From the family of possible points we chose the best three points which resulted in the minimum root-mean-square difference between images as the generalized center-of-mass points for use in estimating motion. The estimated motion was used to sum the sets of tomographic images, or incorporated in the iterative reconstruction to correct for motion during reconstruction of the combined projection data. For comparison, the principle-axes method was also applied to estimate the rigid-body motion from the same tomographic images. To evaluate our method for different noise levels, we performed simulations with the MCAT phantom. We observed that though noise degraded the motion-detection accuracy, our method helped in reducing the motion artifact both visually and quantitatively. We also acquired four sets of the emission and transmission data of the Data Spectrum Anthropomorphic Phantom positioned at four different locations and/or orientations. From these we generated a composite acquisition simulating periodic phantom movements during acquisition. The simulated motion was calculated from the generalized center-of-mass points calculated from the tomographic images reconstructed from individual acquisitions. We determined that motion-compensation greatly reduced the motion artifact. Finally, in a simulation with the gated MCAT phantom, an exaggerated rigid-body motion was applied to the end-systolic frame. The motion was estimated from the end-diastolic and end-systolic images, and used to sum them into a summed image without obvious artifact. Compared to the principle-axes method, in two of the three comparisons with anthropomorphic phantom data our method estimated the motion in closer agreement to the Polaris system than the principal-axes method, while the principle-axes method gave a more accurate estimation of motion in most cases for the MCAT simulations. As an image-driven approach, our method assumes angularly complete data sets for each state of motion. We expect this method to be applied in correction of respiratory motion in respiratory gated SPECT, and respiratory or other rigid-body motion in PET

Human-observer LROC study of lesion detection in Ga-67 SPECT images reconstructed using MAP with anatomical priors

We compare the image quality of SPECT reconstruction with and without an anatomical prior. Area under the localization-response operating characteristic (LROC) curve is our figure of merit. Simulated Ga-67 citrate images, a SPECT lymph-nodule imaging agent, were generated using the MCAT digital phantom. Reconstructed images were read by human observers. Several reconstruction strategies are compared, including rescaled block iterative (RBI) and maximum-a-posteriori (MAP) with various priors. We find that MAP reconstruction using prior knowledge of organ and lesion boundaries significantly improves lesion-detection performance (p < 0.05). Pseudo-lesion boundaries, regions without increased uptake which are incorrectly treated as prior knowledge of lesion boundaries, do not decrease performance.

Assessing a system to detect patient motion in SPECT imaging using stereo optical cameras

Patient motion, which causes artifacts in reconstructed images, can be a serious problem in SPECT imaging. If patient motion can be detected and quantified, the reconstruction algorithm can compensate for the motion. Most previous approaches to detecting patient motion have relied on only the acquired projection data, using, for example, consistency checks or motion-tracking to detect motion. Our approach is based on optical tracking of the patient using a pair of web cameras to acquire stereo images. The stereo images are analyzed by a visual tracking system (VTS) that detects changes in the stereo images over time to track locations on the patient surface. Patient surface motion can then be used to infer motion within the patient body, which will be used to correct for patient motion. The system consists of a three-headed SPECT system and two web cameras connected to a PC.

An Assessment of a Low-Cost Visual Tracking System (VTS) to Detect and Compensate for Patient Motion During SPECT

Patient motion is inevitable in SPECT and PET due to the lengthy period of time patients are imaged and patient motion can degrade diagnostic accuracy. The goal of our studies is to perfect a methodology for tracking and correcting patient motion when it occurs. In this paper we report on enhancements to the calibration, camera stability, accuracy of motion tracking, and temporal synchronization of a low-cost visual tracking system (VTS) we are developing. The purpose of the VTS is to track the motion of retro-reflective markers on stretchy bands wrapped about the chest and abdomen of patients. We have improved the accuracy of 3D spatial calibration by using a MATLAB optical camera calibration package with a planar calibration pattern. This allowed us to determine the intrinsic and extrinsic parameters for stereo-imaging with our CCD cameras. Locations in the VTS coordinate system are transformed to the SPECT coordinate system by a VTS/SPECT mapping using a phantom of 7 retro-reflective spheres each filled with a drop of Tc99m. We switched from pan, tilt and zoom (PTZ) network cameras to fixed network cameras to reduce the amount of camera drift. The improved stability was verified by tracking the positions of fixed retro-reflective markers on a wall. The ability of our VTS to track movement, on average, with sub-millimeter and sub-degree accuracy was established with the 7-sphere phantom for 1 cm vertical and axial steps as well as for an arbitrary rotation and translation. The difference in the time of optical image acquisition as decoded from the image headers relative to synchronization signals sent to the SPECT system was used to establish temporal synchrony between optical and list-mode SPECT acquisition. Two experiments showed better than 100 ms agreement between VTS and SPECT observed motion for three axial translations. We were able to track 3 reflective markers on an anthropomorphic phantom with a precision that allowed us to correct motion such that no loss in visual quality was noted in motion corrected slices relative to motion free slices.

Motion correction for cardiac SPECT using a RBI-ML partial-reconstruction approach

In single photon computed tomography, patient motion can significantly affect image quality. Methods to correct for patient motion rely on available information about the motion or stable algorithms have to be developed to detect and describe object motion. In this work we introduce a reconstruction method that corrects for rigid body motion and we investigate a method to exploit the motion information that might be hidden in the projection data and possibly modify motion descriptions that were retrieved by external motion tracking devices. The search method was based on reconstructing only parts of the projection angles using an RBI-ML partial-reconstruction approach (PRA). It tests a motion description with only one forward projection per detector head instead of a full reconstruction using all projections. A figure of merit based on the per angle likelihood function in projection space was introduced. With simulated MCAT data we could show that the PRA was able to identify the optimum 3D rigid body motion correction with an error of 1 pixel in each cartesian direction. Future work will address 3D translations, multiple motions, detector resolution compensation and will also concentrate on the improvement of the figure of merit.