Voxel-based Estimation Research Articles

Hypoxia Informed RBE-Weighted Beam Orientation Optimization for Intensity Modulated Proton Therapy Using [18F]-FMISO-PET Estimation of pO2

Variable relative biological effectiveness (RBE) models have previously informed proton therapy dose optimization algorithms, but few models have incorporated hypoxia's increase on radioresistance. Here, we obtain voxel-based estimation of partial oxygen pressure to weigh RBE values in a single biologically informed beam orientation optimization (BOO) algorithm. Four brain cancer patients with [18F]-FMISO-PET/CT images were selected from an HCP database. Oxygen values were derived from tracer uptake using a non-linear least squares curve fitting. RBE dose was then weighted using oxygen enhancement ratios (OER) for each structure and substituted into the dose fidelity term of our BOO algorithm. The nonlinear optimization problem was solved using a split-Bregman approach, with FISTA as the solver. This method (HypRBE) was compared dose fidelity terms using the Rorvik RBE model (RegRBE), without OER. Tumor homogeneity index (HI), Dmax, and D95% were evaluated along with worst-case statistics after normalization to normal tissue isotoxicity. Compared to RegRBE, HypRBE increased tumor [HI, Dmax, D95%] on average by [0.5%, 2.0%, 2.5%] and improved worst-case tumor [HI, Dmax, D95%] by [5.3%, 16.2%, 9.6%]. HypRBE shows an increase in therapeutic ratio, and is notably robust against uncertainty scenarios. We have developed an optimization algorithm whose dose fidelity term is weighted by hypoxia informed RBE values. We have shown that HypRBE selects beams that are better suited to protect low RBE, well-oxygenated normal tissue while maintaining high dose to high RBE, hypoxic tumor cells.

Quantifying stand-level clumping of boreal, hemiboreal and temperate European forest stands using terrestrial laser scanning

Clumping is critical for quantifying the radiation regime of forest canopies, but challenging to measure. We developed a method to measure clumping index (CI) of forest stands using voxel-based estimates of leaf area density from terrestrial lidar data. Our method uses the principle of the silhouette to total area ratio (STAR), a commonly used shoot clumping correction approach. We adapted the concept to forest stands, and derived that STAR at canopy scale (STARf) is no longer simply a clumping index, but a summary variable for forest structure in general. CI can be calculated from STARf when leaf area index is known.We measured CI and STARf of 38 forest stands in Finland, Estonia, and Czechia to study the natural range of these variables, their relationships to other forest variables, and to Landsat 8 OLI surface reflectance. CI did not include clumping below voxel scale (20 cm), and ranged from 0.6 to 0.9, with the lowest values (i.e., the most clumped canopies) in conifer forests and temperate oak forests, and the highest CI values (i.e., the most random canopies) in boreal broadleaved forests. CI was closely correlated with surface reflectance in conifer forests, which may be explained by contradicting influence of clumping that decreases canopy reflectance, but increases visibility of the forest floor.From the viewpoint of forest reflectance modeling, STAR is a useful variable due to its close relationship with the photon recollision probability, i.e., the probability that a photon will interact with a canopy element after being scattered from another canopy element. The photon recollision probability is used to model the influence of forest structure on reflectance. Our method provides a physically-based means of measuring STARf, and thus the photon recollision probability, hence contributing to the development of new methods for interpreting forest canopy structure from optical remote sensing data.

LiDAR Voxel-Size Optimization for Canopy Gap Estimation

Terrestrial laser scanning of forest structure is used increasingly in place of traditional technologies; however, deriving physical parameters from point clouds remains challenging because LiDAR returns do not have defined areas or volumes. While voxelization methods overcome this challenge, estimation of canopy gaps and other structural attributes are often performed by reducing the point cloud to two-dimensions, thus decreasing the fidelity of the data. Furthermore, relatively few studies have evaluated voxel-size effects on estimation accuracy. Here, we show that voxelized laser-scanning data can be used for canopy-gap estimation without performing dimensionality reduction to the point cloud. Both airborne and terrestrial LiDAR were used to estimate canopy gaps along six vertical transects and four height intervals. Voxel-based estimates were evaluated against hemispherical photography and a sensitivity analysis was performed to identify an optimal voxel size. While the results indicate that our approach can be used with both airborne and terrestrial LiDAR, voxel size has a considerable influence on canopy-gap estimation. Results from our sensitivity analysis indicate that TLS estimation performs best when using 10 cm voxels, yielding canopy gaps ranging from 32–78%. The optimal voxel size for ALS estimation was obtained with 25 cm voxels, yielding estimates ranging from 25–68%.

Estimating average single-neuron visual receptive field sizes by fMRI

The noninvasive estimation of neuronal receptive field (RF) properties in vivo allows a detailed understanding of brain organization as well as its plasticity by longitudinal following of potential changes. Visual RFs measured invasively by electrophysiology in animal models have traditionally provided a great extent of our current knowledge about the visual brain and its disorders. Voxel-based estimates of population RF (pRF) by functional magnetic resonance imaging (fMRI) in humans revolutionized the field and have been used extensively in numerous studies. However, current methods cannot estimate single-neuron RF sizes as they reflect large populations of neurons with individual RF scatter. Here, we introduce an approach to estimate RF size using spatial frequency selectivity to checkerboard patterns. This method allowed us to obtain noninvasive, average single-neuron RF estimates over a large portion of human early visual cortex. These estimates were significantly smaller compared with prior pRF methods. Furthermore, fMRI and electrophysiology experiments in nonhuman primates demonstrated an exceptionally good match, validating the approach.

Parietal lobe critically supports successful paired immediate and single-item delayed memory for targets

The parietal lobe is important for successful recognition memory, but its role is not yet fully understood. We investigated the parietal lobes’ contribution to immediate paired-associate memory and delayed item-recognition memory separately for hits (targets) and correct rejections (distractors). We compared the behavioral performance of 56 patients with known parietal and medial temporal lobe dysfunction (i.e. early Alzheimer’s Disease) to 56 healthy control participants in an immediate paired and delayed single item object memory task. Additionally, we performed voxel-based morphometry analyses to investigate the functional-neuroanatomic relationships between performance and voxel-based estimates of atrophy in whole-brain analyses. Behaviorally, all participants performed better identifying targets than rejecting distractors. The voxel-based morphometry analyses associated atrophy in the right ventral parietal cortex with fewer correct responses to familiar items (i.e. hits) in the immediate and delayed conditions. Additionally, medial temporal lobe integrity correlated with better performance in rejecting distractors, but not in identifying targets, in the immediate paired-associate task. Our findings suggest that the parietal lobe critically supports successful immediate and delayed target recognition memory, and that the ventral aspect of the parietal cortex and the medial temporal lobe may have complementary preferences for identifying targets and rejecting distractors, respectively, during recognition memory.

Disaster debris estimation using high-resolution polarimetric stereo-SAR

This paper addresses the problem of debris estimation which is one of the most important initial challenges in the wake of a disaster like the Great East Japan Earthquake and Tsunami. Reasonable estimates of the debris have to be made available to decision makers as quickly as possible. Current approaches to obtain this information are far from being optimal as they usually rely on manual interpretation of optical imagery. We have developed a novel approach for the estimation of tsunami debris pile heights and volumes for improved emergency response. The method is based on a stereo-synthetic aperture radar (stereo-SAR) approach for very high-resolution polarimetric SAR. An advanced gradient-based optical-flow estimation technique is applied for optimal image coregistration of the low-coherence non-interferometric data resulting from the illumination from opposite directions and in different polarizations. By applying model based decomposition of the coherency matrix, only the odd bounce scattering contributions are used to optimize echo time computation. The method exclusively considers the relative height differences from the top of the piles to their base to achieve a very fine resolution in height estimation. To define the base, a reference point on non-debris-covered ground surface is located adjacent to the debris pile targets by exploiting the polarimetric scattering information. The proposed technique is validated using in situ data of real tsunami debris taken on a temporary debris management site in the tsunami affected area near Sendai city, Japan. The estimated height error is smaller than 0.6m RMSE. The good quality of derived pile heights allows for a voxel-based estimation of debris volumes with a RMSE of 1099m3. Advantages of the proposed method are fast computation time, and robust height and volume estimation of debris piles without the need for pre-event data or auxiliary information like DEM, topographic maps or GCPs.

Voxel-Based Estimation of Kinetic Model Parameters of the l-[1-11C]Leucine PET Method for Determination of Regional Rates of Cerebral Protein Synthesis: Validation and Comparison with Region-of-Interest-Based Methods

We adapted and validated a basis function method (BFM) to estimate at the voxel level parameters of the kinetic model of the L-[1-(11)C]leucine positron emission tomography (PET) method and regional rates of cerebral protein synthesis (rCPS). In simulation at noise levels typical of voxel data, BFM yielded low-bias estimates of rCPS; in measured data, BFM and nonlinear least-squares parameter estimates were in good agreement. We also examined whether there are advantages to using voxel-level estimates averaged over regions of interest (ROIs) in place of estimates obtained by directly fitting ROI time-activity curves (TACs). In both simulated and measured data, fits of ROI TACs were poor, likely because of tissue heterogeneity not taken into account in the kinetic model. In simulation, rCPS determined from fitting ROI TACs was substantially overestimated and BFM-estimated rCPS averaged over all voxels in an ROI was slightly underestimated. In measured data, rCPS determined by regional averaging of voxel estimates was lower than rCPS determined from ROI TACs, consistent with simulation. In both simulated and measured data, intersubject variability of BFM-estimated rCPS averaged over all voxels in a ROI was low. We conclude that voxelwise estimation is preferable to fitting ROI TACs using a homogeneous tissue model.

Wavelet denoising in voxel-based parametric estimation of small animal PET images: a systematic evaluation of spatial constraints and noise reduction algorithms

Voxel-based estimation of PET images, generally referred to as parametric imaging, can provide invaluable information about the heterogeneity of an imaging agent in a given tissue. Due to high level of noise in dynamic images, however, the estimated parametric image is often noisy and unreliable. Several approaches have been developed to address this challenge, including spatial noise reduction techniques, cluster analysis and spatial constrained weighted nonlinear least-square (SCWNLS) methods. In this study, we develop and test several noise reduction techniques combined with SCWNLS using simulated dynamic PET images. Both spatial smoothing filters and wavelet-based noise reduction techniques are investigated. In addition, 12 different parametric imaging methods are compared using simulated data. With the combination of noise reduction techniques and SCWNLS methods, more accurate parameter estimation can be achieved than with either of the two techniques alone. A less than 10% relative root-mean-square error is achieved with the combined approach in the simulation study. The wavelet denoising based approach is less sensitive to noise and provides more accurate parameter estimation at higher noise levels. Further evaluation of the proposed methods is performed using actual small animal PET datasets. We expect that the proposed method would be useful for cardiac, neurological and oncologic applications.

Comparison of tumor histology to dynamic contrast enhanced magnetic resonance imaging-based physiological estimates

The purpose of this study was to compare histologically determined cellularity and extracellular space to dynamic contrast-enhanced magnetic resonance imaging (DCE MRI)-based maps of a two-compartment model's parameters describing tumor contrast agent extravasation, specifically tumor extravascular extracellular space (EES) volume fraction (ve), tumor plasma volume fraction (vp) and volume-normalized contrast agent transfer rate between tumor plasma and interstitium (KTRANS/VT). Obtained ve, vp and KTRANS/VT maps were estimated from gadolinium diethylenetriamine penta-acetic acid DCE T1-weighted gradient-echo images at resolutions of 469, 938 and 2500 microm. These parameter maps were compared at each resolution to histologically determined tumor type, and the high-resolution 469-microm maps were compared with automated cell counting using Otsu's method and a color-thresholding method for estimated intracellular (Vintracellular) and extracellular (Vextracellular) space fractions. The top five KTRANS/VT values obtained from each tumor at 469 and 938 microm resolutions are significantly different from those obtained at 2500 microm (P<.0001) and from one another (P=.0014). Using these top five KTRANS/VT values and the corresponding tumor EES volume fractions ve, we can statistically differentiate invasive ductal carcinomas from noninvasive papillary carcinomas for the 469- and 938-microm resolutions (P=.0017 and P=.0047, respectively), but not for the 2500-microm resolution (P=.9008). The color-thresholding method demonstrated that ve measured by DCE MRI is statistically similar to histologically determined EES. The Vextracellular obtained from the color-thresholding method was statistically similar to the ve measured with DCE MRI for the top 10 KTRANS/VT values (P>.05). DCE MRI-based KTRANS/VT estimates are not statistically correlated with histologically determined cellularity. DCE MRI estimates of tumor physiology are a limited representation of tumor histological features. Extracellular spaces measured by both DCE MRI and microscopic analysis are statistically similar. Tumor typing by DCE MRI is spatial resolution dependent, as lower resolutions average out contributions to voxel-based estimates of KTRANS/VT. Thus, an appropriate resolution window is essential for DCE MRI tumor diagnosis. Within this resolution window, the top KTRANS/VT values with corresponding ve are diagnostic for the tumor types analyzed in this study.

Estimation of the partial volume effect in MRI

The partial volume effect (PVE) arises in volumetric images when more than one tissue type occurs in a voxel. In such cases, the voxel intensity depends not only on the imaging sequence and tissue properties, but also on the proportions of each tissue type present in the voxel. We have demonstrated in previous work that ignoring this effect by establishing binary voxel-based segmentations introduces significant errors in quantitative measurements, such as estimations of the volumes of brain structures. In this paper, we provide a statistical estimation framework to quantify PVE and to propagate voxel-based estimates in order to compute global magnitudes, such as volume, with associated estimates of uncertainty. Validation is performed on ground truth synthetic images and MRI phantoms, and a clinical study is reported. Results show that the method allows for robust morphometric studies and provides resolution unattainable to date.

Fast formation of statistically reliable FDG parametric images based on clustering and principal components

Formation of parametric images requires voxel-by-voxel estimation of rate constants, a process sensitive to noise and computationally demanding. A model-based clustering method for a two-parameter model (CAKS) was extended to the FDG three-parameter model. The concept was to average voxels with similar kinetic signatures to reduce noise. Voxel kinetics were categorized by the first two principal components of the tissue time–activity curves for all voxels. k2 and k3 were estimated cluster-by-cluster, and K1 was estimated voxel-by-voxel within clusters. When CAKS was applied to simulated images with noise levels similar to brain FDG scans, estimation bias was well suppressed, and estimation errors were substantially smaller—1.3 times for Ki and 1.5 times for k3—than those of conventional voxel-based estimation. The statistical reliability of voxel-level estimation by CAKS was comparable with ROI analysis including 100 voxels. CAKS was applied to clinical cases with Alzheimer's disease (ALZ) and cortico basal degeneration (CBD). In ALZ, the affected regions had low Ki(K1k3/(k2 + k3)) and k3. In CBD, Ki was low, but k3 was preserved. These results were consistent with ROI-based kinetic analysis. Because CAKS decreased the number of invoked estimations, the calculation time was reduced substantially. In conclusion, CAKS has been extended to allow parametric imaging of a three-compartment model. The method is computationally efficient, with low bias and excellent noise properties.