Image Voxel Size Research Articles

Automatic diameter and orientation distribution determination of fibrous materials in micro X-ray CT imaging data.

Fibrous nanomaterials such as electrospun materials have many uses ranging from tissue engineering to biosensors. High-resolution imaging is an important component in the characterization of these materials. Important parameters required to predict and study the properties of fibre rich materials include diameter and orientation distribution as well as fibre spacing. The orientations and the relative dimensions of the fibres can be measured via specially designed imaging software. Difficulties in this measurement process can arise if fibres are distributed in close proximity to each other in relation to the resolution of the imaging modality. For example, if some automation is required in the measurement process and, particularly, if the automated processes are not designed for situations where the fibres are in close proximity to each other. This work is therefore concerned with the development of automated measurement techniques to provide estimates of the diameters of fibres and also the orientation distribution. The software automatically detects special points in the fibrous materials where fibres can be considered to have some delineation from surrounding fibres. These sparse points are considered to be points at which estimates of the fibres' properties can be quantified. Aligned and randomly distributed electrospun poly(caprolactone) nanofibres were prepared. Imaging of these materials was performed with an X-ray Computer Tomography system with an image voxel size of 0.15 × 0.15 × 0.15μm3 . Scanning Electron Microscopy images were also obtained. Fibre diameters estimated using images from both modalities using the developed techniques were found to be in agreement. Orientation distribution was summarized with multiscale entropy and found to be consistent with visual observation across different scales. LAY DESCRIPTION: Fibres are present in many types of materials which can include, for some materials, very small fibres e.g. a few nanometres in diameter. Very small fibres are present in collagen and elastin which are common tissues of many organs in many types of living things. The sizes of these very small fibres and how they are arranged are important information that can help in the understanding of the overall properties of these materials. Materials with very small fibres can also be synthesized using specialised techniques. The properties of these synthesized fibrous materials are also important to help in understanding how the materials will perform in various different applications. Applications are many and can range from tissue engineering to drug delivery. Some properties of these materials can be shown, visually, with the aid of 3D imaging techniques such as X-ray Computer Tomography (XCT) or in 2D, with Scanning Electron Microscopy (SEM) but at a higher magnification. The work described here is centred around the development of computer algorithms to automatically determine material properties from 3D XCT images. Tests are performed with material samples, where the fibres are aligned (in semi-parallel fashion) and another where the fibres are randomly oriented (criss-crossing). The tests show that the developed algorithms are able to successfully and relatively accurately determine the diameters of the fibres. The tests also show that it is possible to quantify the relative randomness of the orientations of the fibres.

The Effect of Voxel Size on the Reliability and Reproducibility of Alveolar Bone Crest Identification on Cone Beam CT Scans

Objective: the aim of this study is to compare the reliability and reproducibility of landmarks representing the labial alveolar bone crest of maxillary and mandibular teeth using two different CBCT machines with different image resolutions and voxel size and accordingly different radiation dose. Material & Methods: 340 landmarks representing the alveolar bone crest opposite each tooth from the central incisor up to the second molar on sixteen CBCT data sets were identified. The landmarks were located by the principle operator, and relocated by the same principle operator to evaluate the intra-observer reproducibility, then located by the second observer to evaluate the inter-observer reliability error. For testing the intra-observer reproducibility and inter-observer reliability of the landmark identification within each scanner data, the Mean Absolute error (MAE) and the Dahlberg error (DE) and Intra-class correlation coefficient tests with 95% confidence intervals were used. Independent samples t-test was used to compare the intra-observer and inter-observer error between the two scanners. Results: The Mean Absolute Error (MAE), the Dahlberg Error (DE) and the Intra-Class Coefficient (ICC) tests showed excellent intra-observer reproducibility values in the x, y and z-axis for the Galileos scanner, with the least reproducibility in the z-axis. The same concordance was found for the Planmeca CBCT scanner, with excellent reliability in the x, y and z-axis, and the least reproducibility in the z-axis. For the inter-observer reliability testing, the same pattern was found for each CBCT scanner. Very good inter-observer reliability was found in the x, and y dimensions, while the least was found in the vertical z-dimension. Conclusion: The Galileos and the Planmeca CBCT scanners produced comparable reliability and reproducibility for identification of alveolar bone crest landmarks. The vertical axes showed the least accurate coordinates for landmarks identification.

FDG-PET Imaging and Assessment of Tumor Dose Response (SF2) Distribution

To derive tumor cell survival fraction after 2 Gy (SF2) distribution from multiple FDG-PET images obtained during the early weeks of radiotherapy, and evaluate the predictability of SF2 distribution on tumor local failure and recurrent location. FDG-PET/CT images with PET image voxel size of 4 x 4 x 4 mm3 obtained at the pre, weekly and 3/6 months post chemo-radiation treatment for each of 18 HN cancer patients were used for the study. Four of the 18 patients had biopsy proven local failure (22.2%) with the median follow-up time 22.2 (7.5 ∼ 43.3) months. All weekly and post-treatment PET/CT images of each patient were registered voxel by voxel to the pre-treatment planning PET/CT image. Tumor dose response marker, SF2(v), for each tumor voxel v was derived using the linear regression on the logarithm of the first 4 weekly tumor metabolic ratios defined as lnTMRi(v) = ln(SUVi(v)/SUV0(v)), i = 1 to 4. Sensitivity and specificity of SF2-Volume parameters on the prediction of local tumor control and failure were calculated. The relationship between a locally recurrent tumor volume and radio-resistant tumor volume was evaluated. The locally recurrent tumor volume was defined as the iso-SUV volume formed using the %SUVmax on the post-treatment PET image, while the radio-resistant tumor volume was created using the tumor voxels with SF2 > 0.9. The inter- and intra-tumor heterogeneities of SF2 distribution were large with the individual mean from 0.56 to 0.89, and the individual STD from 0.08 to 0.16. All failures had either large mean SF2 or large STD. Sensitivity and specificity, as well as the accuracy, of using V(SF2 > x) to predict treatment local failure were listed in Table 1. Tumor sub-volume with SF2 > 0.9 shows highly predictability on tumor local recurrence after the standard chemo-radiotherapy. In addition, Tumor local recurrent volume (defined using the 70, 80, or 90% of the Post-SUVmax) always contains the radio-resistant tumor voxels (Table 1). Tumor voxel dose response SF2 can be derived using multiple FDG-PET imaging obtained during the early weeks of radiation treatment. Distribution of SF2 has significant predictability power on tumor local failure, and the recurrent location. It can be used as the objective function to target radio-resistant tumor cells.Abstract 1062; Table 1Tumor Volume(SF2>x)V(SF2>0.9)V(SF2>1.0)V(SF2>1.1)Sensitivity0.751.01.0Specificity1.01.01.0Accuracy0.941.01.0AUC0.961.01.0P-value0.0030.0020.002Optimal Cutoff (cc)4.620.960.14Locally Recurrent Volume (LRV)(CC)70% Post-SUVmax 1.09 ∼ 2.8280% Post-SUVmax 0.58 ∼ 1.1590% Post-SUVmax 0.2 ∼ 0.45Radio-Resistant Volume in LRV(CC)0.7 ∼ 1.340.38 ∼ 0.70.19 ∼ 0.38 Open table in a new tab

P.99 - Reliability of DTI-based muscle-volumetry as compared to conventional T1-based manual segmentation

The purpose of this study is to compare two methods of muscle segmentation to assess muscle volumes in human calf muscles. In this prospective cross-sectional study, we evaluated lower leg muscles of 18 healthy volunteers (9 females, mean age 27.6 ± 3.8 years) using muscle diffusion tensor imaging (DTI) and tractography. MRI was performed using a 3T MRI (Achieva 3T X, Philips medical Systems) and a 16 channel Philips Torso XL coil. The lower legs were separated into two FOV to improve shimming. For every FOV a sequence of T1w, T2w and diffusion weighted imaging were acquired. Post procession was done with Mathematica 11 and ExploreDTI (running under Matlab). 3D fiber tractography and DTI analysis were performed for each whole muscle structure. Conventional volumetry based on T1w images was manually performed using 3D Slicer software. Correlation analysis between manual segmented volumes and DTI-based tractvolumes were performed as well as Cronbachs alpha analysis to assess reliability of outcome values. The tract-based volumetry showed a significantly positive correlation (p < 0.05) with conventional T1w based volumetry in every lower leg muscle (for both left and right legs). Internal consistency of both methods was found to be significant (Cronbachs alpha 0.656 – 0.908). Absolut volumes of DTI based volumetry show overall higher muscle volumes than the conventional T1w based volumetry. We could show that muscle DTI information can not only be used to assess microstructural muscle properties and DTI parameters, but can also be used to segment calf muscles and to estimate individual muscle volumes. This could be important for longitudinal evaluations or group comparisons of different muscle diseases with DTI. Muscle volume can provide additional disease specific information besides the diffusion specific information and can be acquired in the same sequence. The differences in absolute muscle volumes are most likely due to partial volume effects caused by different voxel size of T1w images and DTI-images.

Effect of image segmentation & voxel size on micro-CT computed effective transport & elastic properties

Digital rock technology is rapidly evolving interdisciplinary field with many promises, including fast turnaround times for core analysis, repeatable analysis, and multiphysics simulation. We study the impact of the image segmentation threshold and image voxel size on image-computed permeability, elastic moduli, and electrical conductivity. Improved quantitative understanding of such effects is critically important when comparing laboratory-measured rock properties with those computed on digital images of rocks. We analyze properties of over 500 binary microstructures (each of size 10243 voxels) segmented using different segmentation algorithms. We find that uncertainty in computed rock properties, induced due to the choice of segmentation threshold, increases with coarsening of image voxels. Segmentation of the same rock, acquired with different voxels sizes, lead to small variations in porosity but induces relatively large variation in flow and electrical rock properties. This uncertainty is larger for rocks of lower porosity. We find that coarsening of image voxels also leads to rounding and smoothing of pore throats and a sharp decline in specific surface area which results in an increase in computed permeability. We propose simple models to remove the bias in rock properties when image-derived porosity is either overestimated or underestimated due to the choice of the segmentation threshold. This approach is particularly useful when comparing laboratory-measured rock properties to those derived from digital rocks since often the image calculated porosity and the laboratory-measured porosity are not the same. Even after such corrections, the impact of finite image voxel size would still need to be compensated.

Multi-institutional MicroCT image comparison of image-guided small animal irradiators

To recommend imaging protocols and establish tolerance levels for microCT image quality assurance (QA) performed on conformal image-guided small animal irradiators. A fully automated QA software SAPA (small animal phantom analyzer) for image analysis of the commercial Shelley micro-CT MCTP 610 phantom was developed, in which quantitative analyses of CT number linearity, signal-to-noise ratio (SNR), uniformity and noise, geometric accuracy, spatial resolution by means of modulation transfer function (MTF), and CT contrast were performed. Phantom microCT scans from eleven institutions acquired with four image-guided small animal irradiator units (including the commercial PXi X-RAD SmART and Xstrahl SARRP systems) with varying parameters used for routine small animal imaging were analyzed. Multi-institutional data sets were compared using SAPA, based on which tolerance levels for each QA test were established and imaging protocols for QA were recommended. By analyzing microCT data from 11 institutions, we established image QA tolerance levels for all image quality tests. CT number linearity set to R2 > 0.990 was acceptable in microCT data acquired at all but three institutions. Acceptable SNR > 36 and noise levels <55 HU were obtained at five of the eleven institutions, where failing scans were acquired with current-exposure time of less than 120 mAs. Acceptable spatial resolution (>1.5 lp mm−1 for MTF = 0.2) was obtained at all but four institutions due to their large image voxel size used (>0.275 mm). Ten of the eleven institutions passed the set QA tolerance for geometric accuracy (<1.5%) and nine of the eleven institutions passed the QA tolerance for contrast (>2000 HU for 30 mgI ml−1). We recommend performing imaging QA with 70 kVp, 1.5 mA, 120 s imaging time, 0.20 mm voxel size, and a frame rate of 5 fps for the PXi X-RAD SmART. For the Xstrahl SARRP, we recommend using 60 kVp, 1.0 mA, 240 s imaging time, 0.20 mm voxel size, and 6 fps. These imaging protocols should result in high quality images that pass the set tolerance levels on all systems. Average SAPA computation time for complete QA analysis for a 0.20 mm voxel, 400 slice Shelley phantom microCT data set was less than 20 s. We present image quality assurance recommendations for image-guided small animal radiotherapy systems that can aid researchers in maintaining high image quality, allowing for spatially precise conformal dose delivery to small animals.

Validation of biomechanical deformable image registration in the abdomen, thorax, and pelvis in a commercial radiotherapy treatment planning system

The accuracy of deformable image registration tools can vary widely between imaging modalities and specific implementations of the same algorithms. A biomechanical model-based algorithm initially developed in-house at an academic institution was translated into a commercial radiotherapy treatment planning system and validated for multiple imaging modalities and anatomic sites. Biomechanical deformable registration (Morfeus) is a geometry-driven algorithm based on the finite element method. Boundary conditions are derived from the model-based segmentation of controlling structures in each image which establishes a point-to-point surface correspondence. For each controlling structure, material properties and fixed or sliding interfaces are assigned. The displacements of internal volumes for controlling structures and other structures implicitly deformed are solved with finite element analysis. Registration was performed for 74 patients with images (mean vector resolution) of thoracic and abdominal 4DCT (2.8mm) and MR (5.3mm), liver CT-MR (4.5mm), and prostate MR (2.6mm). Accuracy was quantified between deformed and actual target images using distance-to-agreement (DTA) for structure surfaces and the target registration error (TRE) for internal point landmarks. The results of the commercial implementation were as follows. The mean DTA was ≤1.0mm for controlling structures and 1.0-3.5mm for implicitly deformed structures on average. TRE ranged from 2.0mm on prostate MR to 5.1mm on lung MR on average, within 0.1mm or lower than the image voxel sizes. Accuracy was not overly sensitive to changes in the material properties or variability in structure segmentations, as changing these inputs affected DTA and TRE by ≤0.8mm. Maximum DTA >5mm occurred for 88% of the structures evaluated although these were within the inherent segmentation uncertainty for 82% of structures. Differences in accuracy between the commercial and in-house research implementations were ≤0.5mm for mean DTA and ≤0.7mm for mean TRE. Accuracy of biomechanical deformable registration evaluated on a large cohort of images in the thorax, abdomen and prostate was similar to the image voxel resolution on average across multiple modalities. Validation of this treatment planning system implementation supports biomechanical deformable registration as a versatile clinical tool to enable accurate target delineation at planning and treatment adaptation.

Minimum lesion detectability as a measure of PET system performance

BackgroundA phantom in combination with an imaging protocol was developed to measure the limit of small lesion detection on different PET systems. Seven small spheres with inner diameters ranging from 3.95 up to 15.43 mm were imaged in a Jaszczak ECT Phantom, in air, in a cold background, and with sphere to background contrast ratios of 15:1 down to 1.88:1. The imaging times varied from 1 to 16 min. The imaging protocol was performed on the Gemini TF and Vereos by Philips, the mCT and HRRT by Siemens, and the Discovery 710 by General Electric. For each scanning condition, the images were reconstructed with image voxel sizes of 1 to 4 mm cubic voxels. The reconstruction method used for each system was the one recommended by the manufacture to achieve best small image lesion detection results. A human observer study was performed to determine the smallest observable sphere for each scanning condition.ResultsAll systems were able to image the smallest sphere of 3.95 mm inner diameter at the 15 to 1 signal to background ratio when imaged for 16 min. For a typical whole body per bed position scan time of 2 to 4 min, the smallest imaged sphere varied between 4.95 and 6.23 mm at the 15:1 contrast ratio and 12.43 and 15.43 mm at a contrast ratio of 1.88:1. In general, all systems were consistent with the Rose criteria when determining lesion detectability.ConclusionsBesides demonstrating that the current state of the art clinical PET/CT systems have the same lesion detection ability, the study demonstrates how sensitive scan time can be to detecting small lesions which have a relatively small contrast uptake in the range of just 2:1. This should help guide imaging protocols to use longer scan times over regions of the subject in which small lesions are suspect.

Conditional reconstruction: An alternative strategy in digital rock physics

Digital rock physics (DRP) is a newly developed method based on imaging and digitizing of 3D pore and mineral structure of actual rock and numerically computing rock physical properties, such as permeability, elastic moduli, and formation factor. Modern high-resolution microcomputed tomography scanners are used for imaging, but these devices are not widely available, and 3D imaging is also costly and it is a time-consuming procedure. However, recent improvements of 3D reconstruction algorithms such as crosscorrelation-based simulation and, on the other side, the concept of rock physical trends have provided some new avenues in DRP. We have developed a modified work flow using higher order statistical methods. First, a high-resolution 2D image is divided into smaller subimages. Then, different stochastic subsamples are generated based on the provided 2D subimages. Eventually, various rock physical parameters are calculated. Using several subsamples allows extracting rock physical trends and better capturing the heterogeneity and variability. We implemented our work flow on two DRP benchmark data (Berea sandstone and Grosmont carbonate) and a thin-section image from the Grosmont carbonate formation. Results of realization models, pore network modeling, and autocorrelation functions for the real and reconstructed subsamples reveal the validity of the reconstructed models. Furthermore, the agreement between static and dynamic methods indicates that subsamples are representative volume elements. Average values of the subsamples’ properties follow the reference trends of the rock sample. Permeability trends pass the actual results of the benchmark samples; however, elastic moduli trends find higher values. The latter can be due to image resolution and voxel size, which are generated by imaging tools and reconstruction algorithms. According to the obtained results, this strategy can be introduced as a valid and accurate method where an alternative method for standard DRP is needed.

Development of a protocol to quantify local bone adaptation over space and time: Quantification of reproducibility

In vivo micro-computed tomography (µCT) scanning of small rodents is a powerful method for longitudinal monitoring of bone adaptation. However, the life-time bone growth in small rodents makes it a challenge to quantify local bone adaptation. Therefore, the aim of this study was to develop a protocol, which can take into account large bone growth, to quantify local bone adaptations over space and time. The entire right tibiae of eight 14-week-old C57BL/6J female mice were consecutively scanned four times in an in vivo µCT scanner using a nominal isotropic image voxel size of 10.4µm. The repeated scan image datasets were aligned to the corresponding baseline (first) scan image dataset using rigid registration. 80% of tibia length (starting from the endpoint of the proximal growth plate) was selected as the volume of interest and partitioned into 40 regions along the tibial long axis (10 divisions) and in the cross-section (4 sectors). The bone mineral content (BMC) was used to quantify bone adaptation and was calculated in each region. All local BMCs have precision errors (PE%CV) of less than 3.5% (24 out of 40 regions have PE%CV of less than 2%), least significant changes (LSCs) of less than 3.8%, and 38 out of 40 regions have intraclass correlation coefficients (ICCs) of over 0.8. The proposed protocol allows to quantify local bone adaptations over an entire tibia in longitudinal studies, with a high reproducibility, an essential requirement to reduce the number of animals to achieve the necessary statistical power.

Image Integration to Guide Catheter Ablation in Scar‐Related Ventricular Tachycardia

Although multi-detector computed tomography (MDCT) and cardiac magnetic resonance (CMR) can assess the structural substrate of ventricular tachycardia (VT) in ischemic cardiomyopathy (ICM), non-ICM (NICM), and arrhythmogenic right ventricular cardiomyopathy (ARVC), the usefulness of systematic image integration during VT ablation remains undetermined. A total of 116 consecutive patients (67 ICM; 30 NICM; 19 ARVC) underwent VT ablation with image integration (MDCT 91%; CMR 30%; both 22%). Substrate was defined as wall thinning on MDCT and late gadolinium-enhancement on CMR in ICM/NICM, and as myocardial hypo-attenuation on MDCT in ARVC. This substrate was compared to mapping and ablation results with the endpoint of complete elimination of local abnormal ventricular activity (LAVA), and the impact of image integration on procedural management was analyzed. Imaging-derived substrate identified 89% of critical VT isthmuses and 85% of LAVA, and was more efficient in identifying LAVA in ICM and ARVC than in NICM (90% and 90% vs. 72%, P < 0.0001), and when defined from CMR than MDCT (ICM: 92% vs. 88%, P = 0.026, NICM: 88% vs. 72%, P < 0.001). Image integration motivated additional mapping and epicardial access in 57% and 33% of patients. Coronary and phrenic nerve integration modified epicardial ablation strategy in 43% of patients. The impact of image integration on procedural management was higher in ARVC/NICM than in ICM (P < 0.01), and higher in case of epicardial approach (P < 0.0001). Image integration is feasible in large series of patients, provides information on VT substrate, and impacts procedural management, particularly in ARVC/NICM, and in case of epicardial approach.

Online virtual isocenter based radiation field targeting for high performance small animal microirradiation

Advances in precision microirradiators for small animal radiation oncology studies have provided the framework for novel translational radiobiological studies. Such systems target radiation fields at the scale required for small animal investigations, typically through a combination of on-board computed tomography image guidance and fixed, interchangeable collimators. Robust targeting accuracy of these radiation fields remains challenging, particularly at the millimetre scale field sizes achievable by the majority of microirradiators. Consistent and reproducible targeting accuracy is further hindered as collimators are removed and inserted during a typical experimental workflow. This investigation quantified this targeting uncertainty and developed an online method based on a virtual treatment isocenter to actively ensure high performance targeting accuracy for all radiation field sizes. The results indicated that the two-dimensional field placement uncertainty was as high as 1.16 mm at isocenter, with simulations suggesting this error could be reduced to 0.20 mm using the online correction method. End-to-end targeting analysis of a ball bearing target on radiochromic film sections showed an improved targeting accuracy with the three-dimensional vector targeting error across six different collimators reduced from mm (mean ± SD) to mm for an isotropic imaging voxel size of 0.1 mm.

Evaluation of in-vivo measurement errors associated with micro-computed tomography scans by means of the bone surface distance approach

In vivo micro-computed tomography (µCT) scanning is an important tool for longitudinal monitoring of the bone adaptation process in animal models. However, the errors associated with the usage of in vivo µCT measurements for the evaluation of bone adaptations remain unclear. The aim of this study was to evaluate the measurement errors using the bone surface distance approach. The right tibiae of eight 14-week-old C57BL/6 J female mice were consecutively scanned four times in an in vivo µCT scanner using a nominal isotropic image voxel size (10.4 µm) and the tibiae were repositioned between each scan. The repeated scan image datasets were aligned to the corresponding baseline (first) scan image dataset using rigid registration and a region of interest was selected in the proximal tibia metaphysis for analysis. The bone surface distances between the repeated and the baseline scan datasets were evaluated. It was found that the average (±standard deviation) median and 95th percentile bone surface distances were 3.10 ± 0.76 µm and 9.58 ± 1.70 µm, respectively. This study indicated that there were inevitable errors associated with the in vivo µCT measurements of bone microarchitecture and these errors should be taken into account for a better interpretation of bone adaptations measured with in vivo µCT.

Intrinsic diffusion sensitivity of the balanced steady-state free precession (bSSFP) imaging sequence.

The purpose of this work was to analyze the intrinsic diffusion sensitivity of the balanced steady-state free precession (bSSFP) imaging sequence, meaning the observation of diffusion-induced attenuation of the bSSFP steady-state signal due to the imaging gradients. Although these diffusion effects are usually neglected for most clinical gradient systems, such strong gradient systems are employed for high resolution imaging of small animals or MR Microscopy. The impact on the bSSFP signal of the imaging gradients characterized by their b-values was analyzed with simulations and experiments at a 7T animal scanner using a gradient system with maximum gradient amplitude of approx. 700 mT/m. It was found that the readout gradients have a stronger impact on the attenuation than the phase encoding gradients. Also, as the PE gradients are varying with each repetition interval, the diffusion effects induce strong modulations of the bSSFP signal over the sequence repetition cycles depending on the phase encoding gradient table. It is shown that a signal gain can be obtained through a change of flip angle as a new optimal flip angle maximizing the signal can be defined. The dependency of the diffusion effects on relaxation times and b-values were explored with simulations. The attenuation increases with T2. In conclusion, diffusion attenuation of the bSSFP signal becomes significant for high resolution imaging voxel size (roughly < 100 μm) of long T2 substances.

Improving the detection of small lesions using a state-of-the-art time-of-flight PET/CT system and small-voxel reconstructions.

A major disadvantage of (18)F-FDG PET involves poor detection of small lesions and lesions with low metabolism, caused by limited spatial resolution and relatively large image voxel size. As spatial resolution and sensitivity are better in new PET systems, it is expected that small-lesion detection could be improved using smaller voxels. The aim of this study was to test this hypothesis using a state-of-the-art time-of-flight PET/CT device. (18)F-FDG PET scans of 2 image-quality phantoms (sphere sizes, 4-37 mm) and 39 consecutive patients with lung cancer were analyzed on a time-of-flight PET/CT system. Images were iteratively reconstructed with standard 4 × 4 × 4 mm voxels and smaller 2 × 2 × 2 mm voxels. For the phantom study, we determined contrast-recovery coefficients and signal-to-noise ratios (SNRs). For the patient study, (18)F-FDG PET-positive lesions in the chest and upper abdomen with a volume less than 3.0 mL (diameter, <18 mm) were included. Lesion mean and maximum standardized uptake values (SUVmean and SUVmax, respectively) were determined in both image sets. SNRs were determined by comparing SUVmax and SUVmean with background noise levels. A subanalysis was performed for lesions less than 0.75 mL (diameter, <11 mm). For qualitative analysis of patient data, 3 experienced nuclear medicine physicians gave their preference after visual side-by-side analysis. For phantom spheres 13 mm or less, we found higher contrast-recovery coefficients and SNRs using small-voxel reconstructions. For 66 included (18)F-FDG PET-positive lesions, the average increase in SUVmean and SUVmax using the small-voxel images was 17% and 32%, respectively (P < 0.01). For lesions less than 0.75 mL (21 in total), the average increase was 21% and 44%, respectively. Moreover, averaged over all lesions, the mean and maximum SNR increased by 20% and 27%, respectively (P < 0.01). For lesions less than 0.75 mL, these values increased up to 23% and 46%, respectively. The physicians preferred the small-voxel reconstructions in 76% of cases. Supported by a phantom study, there was a visual preference toward (18)F-FDG PET images reconstructed with 2 × 2 × 2 mm voxels and a profound increase in standardized uptake value and SNR for small lesions. Hence, it is expected that small-lesion detection improves using small-voxel reconstructions.

The effect of in situ/in vitro three-dimensional quantitative computed tomography image voxel size on the finite element model of human vertebral cancellous bone.

Quantitative computed tomography-based finite element modeling technique is a promising clinical tool for the prediction of bone strength. However, quantitative computed tomography-based finite element models were created from image datasets with different image voxel sizes. The aim of this study was to investigate whether there is an influence of image voxel size on the finite element models. In all 12 thoracolumbar vertebrae were scanned prior to autopsy (in situ) using two different quantitative computed tomography scan protocols, which resulted in image datasets with two different voxel sizes (0.29 × 0.29 × 1.3 mm(3) vs 0.18 × 0.18 × 0.6 mm(3)). Eight of them were scanned after autopsy (in vitro) and the datasets were reconstructed with two voxel sizes (0.32 × 0.32 × 0.6 mm(3) vs. 0.18 × 0.18 × 0.3 mm(3)). Finite element models with cuboid volume of interest extracted from the vertebral cancellous part were created and inhomogeneous bilinear bone properties were defined. Axial compression was simulated. No effect of voxel size was detected on the apparent bone mineral density for both the in situ and in vitro cases. However, the apparent modulus and yield strength showed significant differences in the two voxel size group pairs (in situ and in vitro). In conclusion, the image voxel size may have to be considered when the finite element voxel modeling technique is used in clinical applications.

Evaluation of Diffusivity in the Anterior Lobe of the Pituitary Gland: 3D Turbo Field Echo with Diffusion-Sensitized Driven-Equilibrium Preparation

3D turbo field echo with diffusion-sensitized driven-equilibrium preparation is a non-echo-planar technique for DWI, which enables high-resolution DWI without field inhomogeneity-related image distortion. The purpose of this study was to evaluate the feasibility of diffusion-sensitized driven-equilibrium turbo field echo in evaluating diffusivity in the normal pituitary gland. First, validation of diffusion-sensitized driven-equilibrium turbo field echo was attempted by comparing it with echo-planar DWI. Five healthy volunteers were imaged by using diffusion-sensitized driven-equilibrium turbo field echo and echo-planar DWI. The imaging voxel size was 1.5 × 1.5 × 1.5 mm(3) for diffusion-sensitized driven-equilibrium turbo field echo and 1.5 × 1.9 × 3.0 mm(3) for echo-planar DWI. ADCs measured by the 2 methods in 15 regions of interests (6 in gray matter and 9 in white matter) were compared by using the Pearson correlation coefficient. The ADC in the pituitary anterior lobe was then measured in 10 volunteers by using diffusion-sensitized driven-equilibrium turbo field echo, and the results were compared with those in the pons and vermis by using a paired t test. The ADCs from the 2 methods showed a strong correlation (r = 0.79; P < .0001), confirming the accuracy of the ADC measurement with the diffusion-sensitized driven-equilibrium sequence. The ADCs in the normal pituitary gland were 1.37 ± 0.13 × 10(-3) mm(2)/s, which were significantly higher than those in the pons (1.01 ± 0.24 × 10(-3) mm(2)/s) and the vermis (0.89 ± 0.25 × 10(-3) mm(2)/s, P < .01). We demonstrated that diffusion-sensitized driven-equilibrium turbo field echo is feasible in assessing ADC in the pituitary gland.

Brown adipose tissue mapping in rats with combined intermolecular double‐quantum coherence and Dixon water–fat MRI

Brown adipose tissue (BAT) is a promising therapeutic target in obesity studies. Recently, MRI has been proposed for the mapping of BAT. However, because of the limitation of spatial resolution, similar to the existing positron emission tomography and computed tomography techniques for BAT detection, it fails to distinguish BAT cells when they are mixed with other cells. In this work, a new MRI method is proposed, combining intermolecular double-quantum coherence and the chemical shift-encoded Dixon method. Its contrast depends on the water to fat ratio at the cellular scale, which is smaller than the imaging voxel size. The feasibility of this MRI method was shown with computer simulations and phantoms, and preliminary imaging of BAT of rats at 7 T. Both computer simulations and experimental results are consistent with theoretical predictions. The method provides a novel contrast mechanism and can map BAT distribution exclusively. In particular, a mixture of BAT cells and white adipose tissue cells was detected in an older rat, which was undetectable by other noninvasive methods. This method may be applicable to a wide range of uses in BAT-related studies, including the formation and variation of BAT.

WE‐G‐WAB‐07: Commissioning of a Novel Integrated Platform for Online MRIgRT

Purpose: An integrated system for online MRIgRT comprising an MR scanner‐on‐rails proximal to a Varian TrueBeam linac is currently under installation at our institution. This study presents the development of a procedure to commission this platform along with the synthesis, testing, and validation of several unique tools/methods central to this challenging frontier. Methods: The commissioning procedure is organized into three phases: 1) the linac and a modified patient couch are commissioned in the absence of magnetic fields; 2) the MR magnet is energized, followed by experimental field mapping and verification of linac/couch performance proximal to the magnet; 3) MR imaging is evaluated and integrated guidance functionalities are tested. Novel methods/technologies embedded in these steps include: magnetic field finite element simulations to confirm magnetic decoupling of the MR scanner, linac, and couch; development of a novel harmonic distortion analysis/shim evaluation technique and associated phantom design incorporating comprehensive imaging QA; simulation of patient‐induced susceptibility distortions and pulse sequence optimization; development of a phantom/method to evaluate the accuracy/stability of MR scanner/couch travel. Results: The finite element simulations of the MRIgRT platform indicated acceptable magnetic forces on critical couch/linac components, undisturbed x‐ray beam production, and acceptable field uniformity within the MRI active volume (via active shimming). Validation of the harmonic distortion analysis demonstrated that complete volumetric mapping of the geometric distortion could be performed with an error bound by the imaging voxel size. Susceptibility‐induced distortions were simulated for various anatomical sites and can be reduced to within 1–2 mm by increasing the frequency encoding gradient strength with acceptable SNR losses. Lastly, the accuracy of translation and rotation measurements was established for evaluating MR scanner/couch travel. Conclusion: A comprehensive procedure has been established for the commissioning of an integrated MRIgRT platform, and various key tools/methods for assessing system performance have been developed and tested. J. Winter and L. Petropoulos are employees of IMRIS. J. Marle and M. Sweitzer are employees of Varian Medical Systems. IMRIS and Varian are companies co‐developing the MRgRT system. Funding for the MRgRT project at Princess Margaret Hospital was provided by the Canadian Foundation for Innovation.

Traceable micro‐CT scaling accuracy phantom for applications requiring exact measurement of distances or volumes

Volumetric x-ray microcomputed tomography (CT) can be employed in a variety of quantitative research applications such as image-guided interventions or characterization of medical devices. To ensure the highest geometric fidelity of images for these applications, a phantom and image processing algorithm have been developed to calibrate the scaling accuracy of micro-CT scanners to a traceable standard and provide corrections to image voxel sizing. The calibration phantom contains six borosilicate beads whose separations have been measured to a traceable standard. An image processing algorithm compares the known separations of the beads to their separations in micro-CT images. A least-squares solution is used to determine linear scaling correction factors along each of the three scanner axes to minimize errors in the bead separations within the images by correcting the image voxel size. The correction factors were applied to images of a similar phantom with beads at different positions to evaluate the ability of the correction factors to reduce errors at points independent of the fiducial locations in the calibration phantom. The calibration phantom was used to evaluate the scaling accuracy of five different micro-CT scanners representing four different scanner models. In two of the five scanners evaluated, the correction factors significantly reduced the mean error in bead separations in the images from 0.17% to 0.05% and from 0.37% to 0.07% of the actual bead separations, respectively. Scanners yielding similar voxel sizes possessed comparable geometric errors after correction using the phantom. Although the magnitude of the corrections is small, such corrections can be important for demanding micro-CT applications. Even if no voxel size correction is required, the phantom provides an easily implemented method to verify the geometric fidelity of micro-CT scanners to a traceable standard of measurement.