Search

A technique for producing accurate models of vascular segments for use in experiments that assess vessel geometry and flow has been developed and evaluated. The models are compatible with x-ray, ultrasound, and magnetic resonance (MR) imaging systems. In this paper, a model of the human carotid artery bifurcation, is evaluated that has been built using this technique. The phantom consists of a thin-walled polyester-resin replica of the bifurcation through which a blood-mimicking fluid may be circulated. The phantom is surrounded by an agar tissue-mimicking material and a series of fiducial markers. The blood- and tissue-mimicking materials have x-ray, ultrasound, and MR properties similar to blood and tissue; fiducial markers provide a means of aligning images acquired by different modalities. The root-mean-square difference between the inner wall geometry of the constructed model and the desired dimensions was 0.33 mm. Static images were successfully acquired using x-ray, ultrasound, and MR imaging systems, and are free of significant artifacts. Flow images acquired with ultrasound and MR agree qualitatively with each other, and with previously published flow patterns. Volume-flow measurements obtained with ultrasound and MR were within 4.4% of the actual values.

In this article, the authors present a method for quickly and easily constructing test phantoms for PET and SPECT quality assurance. As a demonstration, they constructed a complex prototype test phantom, showing the strengths of the construction method. Images taken using a PET/CT and a SPECT scanner are presented, along with a qualitative evaluation of PET/CT using the test phantom. The construction technique provides a quick, easy, and cost effective means of constructing a phantom for use in nuclear medicine imaging.

Previously, our team used Monte Carlo simulation to demonstrate that a gamma camera could potentially be used as an online image guidance device to visualize seeds during permanent breast seed implant procedures. This could allow for intraoperative correction if seeds have been misplaced. The objective of this study is to describe an experimental evaluation of an online gamma-camera imaging of permanent seed implantation (OGIPSI) prototype. The OGIPSI device is intended to be able to detect a seed misplacement of 5 mm or more within an imaging time of 2 min or less. The device was constructed by fitting a custom built brass collimator (16 mm height, 0.65 mm hole pitch, 0.15 mm septal thickness) on a 64 pixel linear array CZT detector (eValuator-2000, eV Products, Saxonburg, PA). Two-dimensional projection images of seed distributions were acquired by the use of a digitally controlled translation stage. Spatial resolution and noise characteristics of the detector were measured. The ability and time needed for the OGIPSI device to image the seeds and to detect cold spots was tested using an anthropomorphic breast phantom. Mimicking a real treatment plan, a total of 52 103Pd seeds of 65.8 MBq each were placed on three different layers at appropriate depths within the phantom. The seeds were reliably detected within 30 s with a median error in localization of 1 mm. In conclusion, an OGIPSI device can potentially be used for image guidance of permanent brachytherapy applications in the breast and, possibly, other sites.

A new technique for producing anthropomorphic radiologic phantoms is described. Information from a patient radiograph is used to design the phantom. Optical densities from a digitized radiograph are converted to thicknesses of a phantom material, with corrections for geometry, x‐ray spectrum, and scatter. Numerically controlled machining techniques are used to transcribe the anatomic details into the phantom material. High spatial frequencies are reproduced separately from low spatial frequencies using photochemical techniques in order to replicate finer details in the phantom. The result is a phantom consisting of a number of overlays. Radiographs of a breast phantom produced using the technique are shown, energy equivalence to attenuation of breast tissue is assessed, and methods of incorporating fine detail and quantitative features in the phantom are suggested.

Introduction: The performance of tumor segmentation methods can only be evaluated by comparison with observers' contours when the true pathologic extent is unknown. All observers' contours contain bias and it is unclear how to best combine segmentations to estimate “truth”. Objective: To develop a method that optimally combines observers' contours to estimate a “true” reference for evaluating tumor segmentation techniques. Methods: A probabilistic “truth” was estimated from multiple observers' segmentations via maximum‐likelihood analysis using the simulated‐annealing (SA) and Expectation‐Maximum optimization algorithms. A qualitative ranking of the performance levels of observers' segmentations, was introduced to steer the method when the relative qualities of input contours are known. The SA‐based method was evaluated first using digital phantoms which were “true” tumors and simulated observers' contours by shifting, shrinking and expanding the “true” tumors. It, secondly, was evaluated using clinical data. The tumor volumes of 12 head and neck cancer (HNC) patients were contoured by three radiation oncologists using CT alone and subsequently, using PET/CT. Results: The SA‐based method exactly determined the “truth” in digital phantom studies. In the clinical study of HNC patients the mean and the range of sensitivity of the SA method were 0.90 and 0.70–0.98, respectively, when compared to a pre‐defined “probabilistic” reference. Conclusions: This work suggests that the SA method can accurately estimate a “truth” to validate image segmentation methods. It also provides a logical means of combining segmentations from different sources, which could potentially improve accuracy of radiation therapy or surgical target localization.

Partial volume correction (PVC) is often needed to correct for limited spatial resolution in quantitative Positron Emission Tomography (PET) and Single Photon Emission Computed Tomography (SPECT) studies. In conventional region-based PVC methods, spill over between regions segmented from coregistered computed tomography (CT) or magnetic resonance (MR) images is accounted for by calculating regional spread functions (RSFs) in a geometric transfer matrix (GTM) framework. This paper describes a new analytically derived symmetric GTM (sGTM) method that considers spill over between RSFs rather than between regions. The sGTM is mathematically equivalent to Labbe's method, however it is region-based rather than voxel-based and it avoids handling large matrices. The sGTM method was validated using an MR-based 3D digital brain phantom and a physical phantom containing spheres 5 mm to 30 mm in diameter. The sGTM method was compared to the GTM method in terms of accuracy, precision, noise propagation, and robustness, i.e. effects of mis-registration or point spread function (PSF) estimation errors. The results showed that the sGTM method has accuracy similar to that of the GTM method, and within 5% of the true value. However, the sGTM method showed better precision and noise propagation than the GTM method, especially for spheres smaller than 13 mm. Moreover, the sGTM method was more robust than the GTM method when misregistration or errors in estimates of PSF occurred. In conclusion, the sGTM method was analytically derived and validated and shown to exhibit better noise characteristics and robustness compared to the GTM method.

Automated methods using CT‐image‐based texture features have shown promise for segmentation of lungtumours. Combining CT with PET features for use in segmentation has been shown to improve segmentation accuracy compared to using either modality separately and has potential for use in accurate internal target volume definition in lungcancer. One issue of particular importance in lungtumor segmentation is the effect of motion on the measures extracted from PET and CTimages. This aspect is under investigation using maximum intensity projections (MIPs) in addition to temporally gated CT, PET and un‐gated PET data sets. Preliminary results from 13 patients (9 diagnosed with lungcancer; 4 diagnosed with cancers outside the lung area, representing healthy lungtissue) show that with gated CT, PET and CT homogeneity, PET Entropy, and CT Coarseness are some of the strongest discriminators for use in classification with statistical distance measures of up to 2.0 between normal and abnormal tissue. The texture features from MIPs of 4 of the patients show that they present somewhat less discriminating feature values, as to be expected. Their usefulness within the context of segmentation remains to be investigated.

Introduction: Previous attempts to segment tumours for radiation therapy targeting based on FDG‐PET image thresholds have had little success. However, if the texture information available in PET and CTimages is used, more accurate and reliable differentiation of abnormal and normal tissues may be possible. Objective: To develop an automated image segmentation method for head and neck cancer (HNC) using texture analysis of co‐registered FDG‐PET/CT images. Methods: CO‐registered Multi‐modality Pattern Analysis Segmentation System (COMPASS) was developed using a region‐of‐interest‐based Decision Tree K‐Nearest‐Neighbors (DTKNN) classifier. 14 PET and 13 CT texture features such as coarseness, busyness and Left/right symmetrical ratio were calculated for each voxel from corresponding PET and CTimages within a window centered on the voxel. Then the voxel was classified as “tumor” or “non‐tumor” using the DTKNN classifier. PET/CT images of 10 patients with HNC who had their primary tumors and positive nodes manually segmented by three radiation oncologists were used for evaluation. Results: The sensitivity per patient was 83%±19% when “true positive voxels” were defined as those voxels identified by at least two physicians as tumor. The specificity was 95%±2% when “true negative” voxels were all soft tissue voxels not identified by any of three physicians as tumor. Results of COMPASS were significantly better than those of three previously published PET threshold‐based methods. Conclusions: This work suggests that an automated segmentation method based on texture classification of FDG‐PET/CT images has the potential to provide accurate delineation of HNC.

Purpose: To develop a robust technique that can combine biased segmentations from observers or multiple imaging modalities to estimate a “true” reference for evaluating tumor segmentation techniques. Method and Materials: A probabilistic “truth” was estimated from a group of observers' segmentations via the analysis of a maximum likelihood (ML) using simulated annealing (SA) and Expectation Maximum (EM) optimization algorithms. A loose condition, qualitative ranking of performance level of segmentations, was introduced to steer the SA based method to find an optimal solution when the relative qualities of input contours are known. The SA based method was compared with a previously published “truth” estimation method “STAPLE”, firstly, using digital phantoms, then, using clinical data ‐ the gross tumor volumes (GTVs) of 12 head and neck cancer patients contoured by three experienced radiation oncologists using CT alone and subsequently, using PET /CT. Results: The SA based method accurately determined the “truth” in digital phantom studies. In comparison, STAPLE was not correctly estimate the “truth”. In the study of the 12 head and neck patients, with a defined probabilistic “true” reference, the mean and range of sensitivity of the SA method were 0.90(0.70–0.98) compared to STAPLE 0.75 (0.45–0.94); the mean and range of volume ratio between the SA method and the reference were 0.964 (0.790–1.048), and between STAPLE and the reference were 0.762 (0.464–1.005). The reference volumes produced by the SA method were reliable and reasonable in all cases. Conclusions: This work suggests that the SA method can accurately estimate a “truth” to validate segmentation methods which could potentially improve accuracy of radiation therapy or surgical target localization. The technique provides a reliable and convenient tool that could be used to optimally combine localization information from different sources to delineate clinical targets.

Purpose: To automatically segment the radiation target for treatment of head and neck cancer (HNC) from FDG‐PET/CT images using a textural classifier and to compare the automated results with contours defined by expert observers. Method and Materials: 27 image features, including textural features from Spatial Gray‐Level Dependence Matrices and Neighborhood Gray‐Tone‐Difference Matrices, as well as statistical and structural features were calculated for 476 head and neck regions of interest (ROIs) in PET/CT images of 20 patients with HNC and 20 patients with lungcancer. A voxel based automated segmentation method using a Decision Tree (DT) based K nearest neighbors (KNN) classifier was developed based on the features in these ROIs. PET/CT images of another 10 head and neck patients who had all primary tumors and positive nodes manually segmented by three radiation oncologists were used to evaluate the method. Features were calculated for each voxel from corresponding PET and CTimages within a window centered on the voxel. All voxels of head and neck soft tissues from the below the eye to the apex of the lung were automatically segmented. Results: The specificity was 95% ± 2% when all “true negative” voxels were considered to be all soft tissue voxels excluding the ROIs considered abnormal by one or more of three radiation oncologists. Sensitivity was 84% ± 19% when “true positive” was considered the intersection of at least two physicians' abnormal ROIs and sensitivity was 90% ± 16% when all “true positive” was the intersection of the abnormal ROIs of all three physicians. Conclusion: This work suggests that an automated segmentation method based on texture classification of FDG‐PET/CT images has potential to provide accurate delineation of HNC. This could potentially lead to reduction in inter‐observer variability in target delineation and improved accuracy of treatment delivery.

A new technique for producing anthropomorphic radiologic phantoms is described. Information from a patient radiograph is used to design the phantom. Optical densities from a digitized radiograph are converted to thicknesses of a phantom material, with corrections for geometry, x-ray spectrum, and scatter. Numerically controlled machining techniques are used to transcribe the anatomic details into the phantom material. High spatial frequencies are reproduced separately from low spatial frequencies using photochemical techniques in order to replicate finer details in the phantom. The result is a phantom consisting of a number of overlays. Radiographs of a breast phantom produced using the technique are shown, energy equivalence to attenuation of breast tissue is assessed, and methods of incorporating fine detail and quantitative features in the phantom are suggested.

Objective: To assess whether a proposed SPECT device could address the requirements of resolving distributions of permanent breast brachytherapy 103 Pd seeds following implantation; while maintaining an acceptable imaging time to allow for correction of misplaced seeds. Method:Monte Carlo simulations of a cadmiumzinc telluride crystal‐based gamma camera were used to assess whether the detection of 22 keV photons emitted from the 103 Pd seeds was feasible. A hexagonal parallel hole collimator, hole length 38 mm, diameter 1.2 mm with 0.2 mm septa was modeled. The design of the gamma camera device was evaluated on two phantom models. The first model consisted of a simple representation of the clinical problem by simulating the breast as 8 cm diameter sphere of breast tissue containing a central, 1cm cubic distribution of 8 seeds. The second simulation presented a more accurate depiction of the clinical problem, where the breast model was based on the pre‐implant CT scan of a typical breast brachytherapy patient and the activity was simulated from the patient's corresponding treatment plan. Results: The spherical phantom yielded promising results after 24 s of imaging time, where the maximum error between the center of mass of the seeds in the reconstructed image and the simulated seed location was 1.02 mm. The results from the clinically accurate simulation revealed that individual seeds could not be identified from the reconstructed images after 2 minutes of imaging. However, the strands of seeds, arranged in each needle were localized to a maximum error of 1.9 mm. Conclusion: The online gamma‐camera approach to imaging the 103 Pd seeds is feasible for simple seed distributions. Additional improvements to the collimator design and the gamma camera orbit are required before the gamma camera device will be able to distinguish each seed in an implanted seed distribution.

Objective: To generate realistic computer‐simulated PET/CT tumourimages with characteristics typical of the tumour (shape, pixel intensity and movement over time) and of the imaging system (acquisition parameters and resolution). Methods: An initial tumour volume was segmented from the CT scan of a lungcancer patient via region growing. To improve the axial resolution and mimic the contour curvature and pixel variations in the transaxial plane, fractal analysis and interpolation was used. Breathing motion with a period and extents typical of lungcancer patients was then applied to the fractaltumour. To produce PET and CTimages, the moving tumour was sampled according to acquisition characteristics typical of the two modalities. For CT, a slice thickness of 3.0mm, acquisition time of 0.7s and resolution of 2mm was assumed. For PET, it was assumed that tumour motion was captured within one bed position over an acquisition time of 180s, and the imager resolution was 6mm. Results: Following fractalinterpolation, the tumour exhibited similar curvature and gray‐level gradation in all three imaging planes. Application of motion and blurring produced CTimages different in appearance from the stationary fractaltumour volume and the total volume traced by the moving object. Motion and blurring produced PETimages similar to the volume traced by the moving object, graded by the time spent in each location within the volume and blurred due to the degraded resolution of the scanner.Conclusions: Realistic computer‐simulated PET/CT tumourimages can be generated using fractal analysis and motion modeling.

Objective: To assess whether a proposed SPECT device could resolve implanted breast brachytherapy seeds within an acceptable imaging time for correction of misplaced seeds. Method:Monte Carlo simulations of a cadmiumzinc telluride crystal‐based gamma camera were used to assess whether the detection of 22 keV photons emitted from 103 Pd seeds was feasible. A 5×5 cm detector, fitted with a hexagonal parallel hole collimator (hole length 38 mm, diameter 1.2 mm, septa 0.2 mm) was modeled. Projections were taken every 7.5 degrees, with a radius of rotation of 10 cm, and images reconstructed using an OSEM algorithm. The phantom modeled consisted of an 8 cm diameter sphere of breast tissue containing a central, 1 cm cubic distribution of 8 seeds, which were each 5 mm long and 0.8 mm wide titanium tubes with an inner radius of 0.32 mm. Results: An acquisition duration of 24 seconds yielded images with a FWHM of 6.0 mm and a scatter fraction of 8.2%. The error between the center of mass of the reconstructed image and the physical seed location was 0.81 ± 0.16 mm, when the phantom was imaged for 24 seconds. Conclusion: The online gamma‐camera approach to imaging the 103 Pd seeds is feasible for a simple seed distribution. The high contrast between the seeds and the non‐radioactive background allow a practical acquisition time of a minute or less. Additional simulations will be required to assess the system design for more realistic seed distributions in a geometry more closely modeling a patient.

Purpose: To determine whether quantitatively segmented PETimages could be used to identify the volume containing a tumor and its total motion. If possible, PET could provide individualized internal target volumes (ITV) in lungcancer.Method and Materials: A physiological phantom containing background level of Na‐22 was used. Two spheres filled with 0.5 mCi/ml of Na‐22 were used to simulate tumors; each was oscillated within one lung of the phantom with 4 preset motion extents in S/I, A/P, and M/L directions. PET and CTimaging were performed on an integrated PET/CT scanner. A CT‐based GTV was generated using a threshold of −850 HU. A population‐based margin of 15 mm, reflecting both motion and set‐up uncertainties, was added to generate a CT‐based PTV. A PET‐based ITV was defined using a threshold of three standard deviations above normal lung background. A set‐up margin of 7.5 mm was added to PET‐based ITVs to create PTVs. Image‐based PTVs were compared to ideal PTVs. Clinical validation of this methodology was performed on 7 patients with parenchymal lung lesions with the addition of digital fluoroscopy. 18‐FDG was used for patient PET scanning. Results: For the phantom study, PET‐based PTVs were closer to the ideal PTV than those based on CT. While the PET‐based PTVs were approximately half the size of the CT‐based PTVs, in no case would the PET‐based PTVs have resulted in geographical miss. For majority of the patients, PET accurately predicted or slightly over‐predicted the tumor motion extents compared to fluoroscopy; differences were within 2 voxels. Conclusion: Based on the phantom study and initial clinical validation, we have found that quantitatively segmented PETimages can provide an accurate individualized ITV that correlates with a tumor and its motion. Conflict of Interest: Research was supported by NCI Canada with funds from Ontario Cancer Society.