Terence A. Riauka

Thresholding in PET images of static and moving targets

Continued therapeutic gain in the treatment of non-small-cell lung cancer (NSCLC) will depend upon our ability to escalate the dose to the primary tumour while minimizing normal tissue toxicity. Both these objectives are facilitated by the accurate definition of a target volume that is as small as possible. To this end, both tumour immobilizations via deep inspiratory breath-hold, along with positron emission tomography (PET), have emerged as two promising approaches. Though PET is an excellent means of defining the general location of a tumour focus, its ability to define exactly the geometric extent of such a focus strongly depends upon selection of an appropriate image threshold. However, in clinical practice, the image threshold is typically not chosen according to consistent, well-established criteria. This study explores the relationship between image threshold and the resultant PET-defined volume using a series of F-18 radiotracer-filled hollow spheres of known internal volumes, both static and under oscillatory motion. The effects of both image threshold and tumour motion on the resultant PET image are examined. Imaging data are further collected from a series of simulated gated PET acquisitions in order to test the feasibility of a patient-controlled gating mechanism during deep inspiratory breath-hold. This study illustrates quantitatively considerable variability in resultant PET-defined tumour volumes depending upon numerous factors, including image threshold, size of the lesion, the presence of tumour motion and the scanning protocol. In this regard, when using PET in treatment planning for NSCLC, the radiation oncologist must select the image threshold very carefully to avoid either under-dosing the tumour or overdosing normal tissues.

Photon propagation and detection in single‐photon emission computed tomography—An analytical approach

An analytical theory of photon propagation and detection in single-photon emission computed tomography (SPECT) for collimated detectors is developed from first principles. The total photon detection kernel is expressed as a sum of terms due to the primary and the Compton scattered photons. The primary as well as contributions due to every order of Compton scattering are calculated separately. The model accounts for the three-dimensional depth dependence of the collimator holes as well as for nonhomogeneous attenuation. No specific assumptions about the boundary or the homogeneity of the attenuating medium are made. The energy response of the detector is also modeled by the theory. Analytical expressions are obtained for various contributions to the photon detection kernel, and the multidimensional integrals involved are calculated using standard numerical integration methods. Theoretically calculated projections and scatter fractions for the primary and the first through second scattering orders are compared with our own experimental results for a small cylindrical primary radiation source immersed at various positions in a uniform cylindrical phantom. Also, theoretically calculated scatter fractions for a small spherical (pointlike) source in a uniform elliptic phantom are compared with experimental and Monte Carlo simulation results taken from the recent literature. The results from the analytical method are essentially exact and are free from the inaccuracies inherent in the numerical simulation methods used to deal with the photon propagation and detection problem in SPECT so far. The method developed here is unique in the sense that it provides accurate theoretical predictions of results averaged over an infinite number of simulations or experiments. We believe that our theory enhances an intuitive understanding of the complex image formation process in SPECT and is an important step toward solving the inverse problem, that of reconstructing the primary radiation source distribution from the measured gamma camera projections.

Threshold modification for tumour imaging in non-small-cell lung cancer using positron emission tomography.

AimPositron emission tomography (PET) has been used increasingly in the staging and radiotherapy treatment planning of non-small-cell lung cancer (NSCLC). This study investigates the factors that affect the resultant size of a given image on PET. MethodsPET was used to assess the geometric characteristics of a series of radioisotope-filled, stationary spheres of known volume, surrounded by positron-emitting radioactive tracer of variable activity. The resultant PET-derived spherical volumes were then referenced to the known spherical volumes in order to illustrate quantitatively the potential influence of image threshold, tumour size and background concentration. This influence was further illustrated by clinical examples. ResultsConsidering the diameter of the spheres used in this study (10–48 mm), higher image thresholds were required for accurate rendering of the smallest spherical volumes. This inverse relationship was most consistently illustrated at the lowest background intensity ratios. ConclusionPET-derived volumes of NSCLC must be interpreted with caution. The data presented in this study may be used to guide the selection of appropriate image thresholds for potential clinical application.

Assessment of a commercially available automatic deformable registration system.

In recent years, a number of approaches have been applied to the problem of deformable registration validation. However, the challenge of assessing a commercial deformable registration system – in particular, an automatic registration system in which the deformable transformation is not readily accessible – has not been addressed. Using a collection of novel and established methods, we have developed a comprehensive, four‐component protocol for the validation of automatic deformable image registration systems over a range of IGRT applications. The protocol, which was applied to the Reveal‐MVS system, initially consists of a phantom study for determination of the systems general tendencies, while relative comparison of different registration settings is achieved through postregistration similarity measure evaluation. Synthetic transformations and contour‐based metrics are used for absolute verification of the systems intra‐modality and inter‐modality capabilities, respectively. Results suggest that the commercial system is more apt to account for global deformations than local variations when performing deformable image registration. Although the protocol was used to assess the capabilities of the Reveal‐MVS system, it can readily be applied to other commercial systems. The protocol is by no means static or definitive, and can be further expanded to investigate other potential deformable registration applications. PACS numbers: 87.19.xj, 87.56.Da, 87.57.nj

Properties of noise in positron emission tomography images reconstructed with filtered-backprojection and row-action maximum likelihood algorithm.

Noise levels observed in positron emission tomography (PET) images complicate their geometric interpretation. Post-processing techniques aimed at noise reduction may be employed to overcome this problem. The detailed characteristics of the noise affecting PET images are, however, often not well known. Typically, it is assumed that overall the noise may be characterized as Gaussian. Other PET-imaging-related studies have been specifically aimed at the reduction of noise represented by a Poisson or mixed Poisson + Gaussian model. The effectiveness of any approach to noise reduction greatly depends on a proper quantification of the characteristics of the noise present. This work examines the statistical properties of noise in PET images acquired with a GEMINI PET/CT scanner. Noise measurements have been performed with a cylindrical phantom injected with 11C and well mixed to provide a uniform activity distribution. Images were acquired using standard clinical protocols and reconstructed with filtered-backprojection (FBP) and row-action maximum likelihood algorithm (RAMLA). Statistical properties of the acquired data were evaluated and compared to five noise models (Poisson, normal, negative binomial, log-normal, and gamma). Histograms of the experimental data were used to calculate cumulative distribution functions and produce maximum likelihood estimates for the parameters of the model distributions. Results obtained confirm the poor representation of both RAMLA- and FBP-reconstructed PET data by the Poisson distribution. We demonstrate that the noise in RAMLA-reconstructed PET images is very well characterized by gamma distribution followed closely by normal distribution, while FBP produces comparable conformity with both normal and gamma statistics.

A conjugate of gemcitabine with bisphosphonate (Gem/BP) shows potential as a targeted bone-specific therapeutic agent in an animal model of human breast cancer bone metastases.

Bone metastases in advanced breast cancer patients remains a significant treatment challenge. Bisphosphonates are now a routine first line treatment for prevention and treatment of skeletal damage caused by malignancies and, moreover, have shown an ability to transport therapeutic drugs to the bone. Here, we describe the effect of a conjugate between the potent anticancer drug gemcitabine and a bisphosphonate molecule (Gem/BP) in an animal model of breast cancer metastases. We have previously demonstrated the targeting of this compound to bone in normal mice using an analog labeled with the radionuclide 99mTc. Using a bone metastasis model in nude mice produced by intracardiac injection of the human breast cancer cell line MDA-MB-231BO, we examined the effect of Gem/BP and gemcitabine in reducing the frequency and severity of osteolytic bone lesions. High-resolution radiographs and microPET images showed that Gem/ BP reduced the number and size of bone metastases relative to the gemcitabine-treated and the untreated control groups. Histological examination of the humeri and femurs of the control and gemcitabine groups revealed large metastatic cancer lesions in the outer and inner cortices and the medullary cavities. In contrast, Gem/BP-treated mice showed occasional small wedge-shaped metastases under the periosteum of the outer cortex and very occasionally in the medulla. These findings suggest that Gem/BP should be further evaluated for use in the treatment of bone metastases in breast cancer.

Experimental and numerical investigation of the 3D SPECT photon detection kernel for non-uniform attenuating media

Experimental tests for non-uniform attenuating media are performed to validate theoretical expressions for the photon detection kernel, obtained from a recently proposed analytical theory of photon propagation and detection for SPECT. The theoretical multi-dimensional integral expressions for the photon detection kernel, which are computed numerically, describe the probability that a photon emitted from a given source voxel will trigger detection of a photon at a particular projection pixel. The experiments were performed using a cylindrical water-filled phantom with large cylindrical air-filled inserts to simulate inhomogeneity of the medium. A point-like, a short thin cylindrical and a large cylindrical radiation source of 99Tcm were placed at various positions within the phantom. The values numerically calculated from the theoretical kernel expression are in very good agreement with the experimentally measured data. The significance of Compton-scattered photons in planar image formation is discussed and highlighted by these results. Using both experimental measurements and the calculated values obtained from the theory, the kernels size is investigated. This is done by determining the square N x N pixel neighbourhood of the gamma camera that must be connected to a particular radiation source voxel to account for a specific fraction of all counts recorded at all camera pixels. It is shown that the kernels size is primarily dependent upon the source position and the properties of the attenuating medium through Compton scattering events, with 3D depth-dependent collimator resolution playing an important but secondary role, at least for imaging situations involving parallel hole collimation. By considering small point-like sources within a non-uniform elliptical phantom, approximating the human thorax, it is demonstrated that on average a 12 cm x 12 cm area of the camera plane is required to collect 85% of the total count recorded. This is a significantly larger connectivity than the 3 cm x 3 cm area required if scattering contributions are ignored and only the 3D depth-dependent collimator resolution is considered.

Monte Carlo investigation of single cell beta dosimetry for intraperitoneal radionuclide therapy

Single event spectra for five beta-emitting radionuclides (Lu-177, Cu-67, Re-186, Re-188, Y-90) were calculated for single cells from two source geometries. The first was a surface-bound isotropically emitting point source and the second was a bath of free radioactivity in which the cell was submerged. Together these represent a targeted intraperitoneal radionuclide therapy. Monoenergetic single event spectra were calculated over an energy range of 11 keV to 2500 keV using the EGSnrc Monte Carlo system. Radionuclide single event spectra were constructed by weighting monoenergetic single event spectra according to radionuclide spectra appropriate for each source geometry. In the case of surface-bound radioactivity, these were radionuclide beta decay spectra. For the free radioactivity, a continuous slowing down approximation spectrum was used that was calculated based on the radionuclide decay spectra. The frequency mean specific energy per event increased as the energy of the beta emitter decreased. This is because, at these energies, the stopping power of the electrons decreases with increasing energy. The free radioactivity produced a higher frequency mean specific energy per event than the corresponding surface-bound value. This was primarily due to the longer mean path length through the target for this geometry. This information differentiates the radionuclides in terms of the physical process of energy deposition and could be of use in the radionuclide selection procedure for this type of therapy.

Planar imaging quantification using 3D attenuation correction data and Monte Carlo simulated buildup factors

A new method to correct for attenuation and the buildup of scatter in planar imaging quantification is presented. The method is based on the combined use of 3D density information provided by computed tomography to correct for attenuation and the application of Monte Carlo simulated buildup factors to correct for buildup in the projection pixels. CT and nuclear medicine images were obtained for a purpose-built nonhomogeneous phantom that models the human anatomy in the thoracic and abdominal regions. The CT transverse slices of the phantom were converted to a set of consecutive density maps. An algorithm was developed that projects the 3D information contained in the set of density maps to create opposing pairs of accurate 2D correction maps that were subsequently applied to planar images acquired from a dual-head gamma camera. A comparison of results obtained by the new method and the geometric mean approach based on published techniques is presented for some of the source arrangements used. Excellent results were obtained for various source-phantom configurations used to evaluate the method. Activity quantification of a line source at most locations in the nonhomogeneous phantom produced errors of less than 2%. Additionally, knowledge of the actual source depth is not required for accurate activity quantification. Quantification of volume sources placed in foam, Perspex and aluminium produced errors of less than 7% for the abdominal and thoracic configurations of the phantom.

Approximate 3D iterative reconstruction for SPECT.

Compared with slice-by-slice approaches for SPECT reconstruction, three-dimensional iterative methods provide a more accurate physical model and an improved SPECT image. Clinical application of these methods, however, is limited primarily to their computational demands. This paper investigates the methods for approximate 3D iterative reconstruction that greatly reduce this demand by excluding from the reconstruction the smaller magnitude elements of the system matrix. A new method is described which is designed to control the resulting bias in the SPECT image for a given reduction in computation. The approximate methods were compared to fully 3D iterative reconstruction in terms of SPECT image bias and visual quality. All methods were incorporated into the ML-EM algorithm and applied to data from 3D mathematical and experimental brain phantoms. The SPECT images reconstructed by the approximate methods exhibited a positive bias throughout the image that was in general smaller with the new method (in the rage of 2%-6%). The bias was smallest in locally hot regions and largest in locally cold regions. The high quality brain phantom images demonstrated the capability of the new method in realistic imaging contexts. The time per iteration for an entire 3D brain phantom on a modern workstation using the approximate 3D method was 7.0 s.