Cross noise level PET denoising with continuous adversarial domain generalization.

Objective.Performing positron emission tomography (PET) denoising within the image space proves effective in reducing the variance in PET images. In recent years, deep learning has demonstrated superior denoising performance, but models trained on a specific noise level typically fail to generalize well on different noise levels, due to inherent distribution shifts between inputs. The distribution shift usually results in bias in the denoised images. Our goal is to tackle such a problem using a domain generalization technique.Approach.We propose to utilize the domain generalization technique with a novel feature space continuous discriminator (CD) for adversarial training, using the fraction of events as a continuous domain label. The core idea is to enforce the extraction of noise-level invariant features. Thus minimizing the distribution divergence of latent feature representation for different continuous noise levels, and making the model general for arbitrary noise levels. We created three sets of 10%, 13%-22% (uniformly randomly selected), or 25% fractions of events from 9718F-MK6240 tau PET studies of 60 subjects. For each set, we generated 20 noise realizations. Training, validation, and testing were implemented using 1400, 120, and 420 pairs of 3D image volumes from the same or different sets. We used 3D UNet as the baseline and implemented CD to the continuous noise level training data of 13%-22% set.Main results.The proposed CD improves the denoising performance of our model trained in a 13%-22% fraction set for testing in both 10% and 25% fraction sets, measured by bias and standard deviation using full-count images as references. In addition, our CD method can improve the SSIM and PSNR consistently for Alzheimer-related regions and the whole brain.Significance.To our knowledge, this is the first attempt to alleviate the performance degradation in cross-noise level denoising from the perspective of domain generalization. Our study is also a pioneer work of continuous domain generalization to utilize continuously changing source domains.

The role of 18F-FDG PET in minimizing variability in gross tumor volume delineation of soft tissue sarcomas.

BACKGROUND: Accurate gross tumor volume (GTV) delineation is a critical step in radiation therapy treatment planning. However, it is reader dependent and thus susceptible to intra- and inter-reader variability. GTV delineation of soft tissue sarcoma (STS) often relies on CT and MR images. PURPOSE: This study investigates the potential role of 18F-FDG PET in reducing intra- and inter-reader variability thereby improving reproducibility of GTV delineation in STS, without incurring additional costs or radiation exposure. MATERIALS AND METHODS: Three readers performed independent GTV delineation of 61 patients with STS using first CT and MR followed by CT, MR, and 18F-FDG PET images. Each reader performed a total of six delineation trials, three trials per imaging modality group. Dice Similarity Coefficient (DSC) score and Hausdorff distance (HD) were used to assess both intra- and inter-reader variability using generated simultaneous truth and performance level estimation (STAPLE) GTVs as ground truth. Statistical analysis was performed using a Wilcoxon signed-ranked test. RESULTS: There was a statistically significant decrease in both intra- and inter-reader variability in GTV delineation using CT, MR 18F-FDG PET images vs. CT and MR images. This was translated by an increase in the DSC score and a decrease in the HD for GTVs drawn from CT, MR and 18F-FDG PET images vs. GTVs drawn from CT and MR for all readers and across all three trials. CONCLUSION: Incorporation of 18F-FDG PET into CT and MR images decreased intra- and inter-reader variability and subsequently increased reproducibility of GTV delineation in STS.

Arterial spin labeled perfusion imaging with balanced steady-state free precession readout and radial sampling.

PURPOSE: To develop an arterial spin labeling (ASL) perfusion imaging method with balanced steady-state free precession (bSSFP) readout and radial sampling for improved SNR and robustness to motion and off-resonance effects. METHODS: An ASL perfusion imaging method was developed with pseudo-continuous arterial spin labeling (pCASL) and bSSFP readout. Three-dimensional (3D) k-space data were collected in segmented acquisitions following a stack-of-stars sampling trajectory. Multiple phase-cycling technique was utilized to improve the robustness to off-resonance effects. Parallel imaging with sparsity-constrained image reconstruction was used to accelerate imaging or increase the spatial coverage. RESULTS: ASL with bSSFP readout showed higher spatial and temporal SNRs of the gray matter perfusion signal compared to those from spoiled gradient-recalled acquisition (SPGR). Cartesian and radial sampling schemes showed similar spatial and temporal SNRs, regardless of the imaging readout. In case of severe B0 inhomogeneity, single-RF phase incremented bSSFP acquisitions showed banding artifacts. These artifacts were significantly reduced when multiple phase-cycling technique (N = 4) was employed. The perfusion-weighted images obtained by the Cartesian sampling scheme showed respiratory motion-related artifacts when a high segmentation number was used. The perfusion-weighted images obtained by the radial sampling scheme did not show these artifacts. Whole brain perfusion imaging was feasible in 1.15 min or 4.6 min for cases without and with phase-cycling (N = 4), respectively, using the proposed method with parallel imaging. CONCLUSIONS: The developed method allows non-invasive perfusion imaging of the whole-brain with relatively high SNR and robustness to motion and off-resonance effects in a practically feasible imaging time.

Impact of motion correction on [18F]-MK6240 tau PET imaging.

Objective. Positron emission tomography (PET) imaging of tau deposition using [18F]-MK6240 often involves long acquisitions in older subjects, many of whom exhibit dementia symptoms. The resulting unavoidable head motion can greatly degrade image quality. Motion increases the variability of PET quantitation for longitudinal studies across subjects, resulting in larger sample sizes in clinical trials of Alzheimer's disease (AD) treatment.Approach. After using an ultra-short frame-by-frame motion detection method based on the list-mode data, we applied an event-by-event list-mode reconstruction to generate the motion-corrected images from 139 scans acquired in 65 subjects. This approach was initially validated in two phantoms experiments against optical tracking data. We developed a motion metric based on the average voxel displacement in the brain to quantify the level of motion in each scan and consequently evaluate the effect of motion correction on images from studies with substantial motion. We estimated the rate of tau accumulation in longitudinal studies (51 subjects) by calculating the difference in the ratio of standard uptake values in key brain regions for AD. We compared the regions' standard deviations across subjects from motion and non-motion-corrected images.Main results. Individually, 14% of the scans exhibited notable motion quantified by the proposed motion metric, affecting 48% of the longitudinal datasets with three time points and 25% of all subjects. Motion correction decreased the blurring in images from scans with notable motion and improved the accuracy in quantitative measures. Motion correction reduced the standard deviation of the rate of tau accumulation by -49%, -24%, -18%, and -16% in the entorhinal, inferior temporal, precuneus, and amygdala regions, respectively.Significance. The list-mode-based motion correction method is capable of correcting both fast and slow motion during brain PET scans. It leads to improved brain PET quantitation, which is crucial for imaging AD.

Posterior Estimation Using Deep Learning: A Simulation Study of Compartmental Modeling in Dynamic PET.

Background: In medical imaging, images are usually treated as deterministic, while their uncertainties are largely underexplored. Purpose: This work aims at using deep learning to efficiently estimate posterior distributions of imaging parameters, which in turn can be used to derive the most probable parameters as well as their uncertainties. Methods: Our deep learning-based approaches are based on a variational Bayesian inference framework, which is implemented using two different deep neural networks based on conditional variational auto-encoder (CVAE), CVAE-dual-encoder and CVAE-dual-decoder. The conventional CVAE framework, i.e., CVAE-vanilla, can be regarded as a simplified case of these two neural networks. We applied these approaches to a simulation study of dynamic brain PET imaging using a reference region-based kinetic model. Results: In the simulation study, we estimated posterior distributions of PET kinetic parameters given a measurement of time-activity curve. Our proposed CVAE-dual-encoder and CVAE-dual-decoder yield results that are in good agreement with the asymptotically unbiased posterior distributions sampled by Markov Chain Monte Carlo (MCMC). The CVAE-vanilla can also be used for estimating posterior distributions, although it has an inferior performance to both CVAE-dual-encoder and CVAE-dual-decoder. Conclusions: We have evaluated the performance of our deep learning approaches for estimating posterior distributions in dynamic brain PET. Our deep learning approaches yield posterior distributions, which are in good agreement with unbiased distributions estimated by MCMC. All these neural networks have different characteristics and can be chosen by the user for specific applications. The proposed methods are general and can be adapted to other problems.

Super-resolution in brain positron emission tomography using a real-time motion capture system.

Super-resolution (SR) is a methodology that seeks to improve image resolution by exploiting the increased spatial sampling information obtained from multiple acquisitions of the same target with accurately known sub-resolution shifts. This work aims to develop and evaluate an SR estimation framework for brain positron emission tomography (PET), taking advantage of a high-resolution infra-red tracking camera to measure shifts precisely and continuously. Moving phantoms and non-human primate (NHP) experiments were performed on a GE Discovery MI PET/CT scanner (GE Healthcare) using an NDI Polaris Vega (Northern Digital Inc), an external optical motion tracking device. To enable SR, a robust temporal and spatial calibration of the two devices was developed as well as a list-mode Ordered Subset Expectation Maximization PET reconstruction algorithm, incorporating the high-resolution tracking data from the Polaris Vega to correct motion for measured line of responses on an event-by-event basis. For both phantoms and NHP studies, the SR reconstruction method yielded PET images with visibly increased spatial resolution compared to standard static acquisitions, allowing improved visualization of small structures. Quantitative analysis in terms of SSIM, CNR and line profiles were conducted and validated our observations. The results demonstrate that SR can be achieved in brain PET by measuring target motion in real-time using a high-resolution infrared tracking camera.

Posterior estimation using deep learning: a simulation study of compartmental modeling in dynamic positron emission tomography.

BACKGROUND: In medical imaging, images are usually treated as deterministic, while their uncertainties are largely underexplored. PURPOSE: This work aims at using deep learning to efficiently estimate posterior distributions of imaging parameters, which in turn can be used to derive the most probable parameters as well as their uncertainties. METHODS: Our deep learning-based approaches are based on a variational Bayesian inference framework, which is implemented using two different deep neural networks based on conditional variational auto-encoder (CVAE), CVAE-dual-encoder, and CVAE-dual-decoder. The conventional CVAE framework, that is, CVAE-vanilla, can be regarded as a simplified case of these two neural networks. We applied these approaches to a simulation study of dynamic brain PET imaging using a reference region-based kinetic model. RESULTS: In the simulation study, we estimated posterior distributions of PET kinetic parameters given a measurement of the time-activity curve. Our proposed CVAE-dual-encoder and CVAE-dual-decoder yield results that are in good agreement with the asymptotically unbiased posterior distributions sampled by Markov Chain Monte Carlo (MCMC). The CVAE-vanilla can also be used for estimating posterior distributions, although it has an inferior performance to both CVAE-dual-encoder and CVAE-dual-decoder. CONCLUSIONS: We have evaluated the performance of our deep learning approaches for estimating posterior distributions in dynamic brain PET. Our deep learning approaches yield posterior distributions, which are in good agreement with unbiased distributions estimated by MCMC. All these neural networks have different characteristics and can be chosen by the user for specific applications. The proposed methods are general and can be adapted to other problems.

Joint spectral quantification of MR spectroscopic imaging using linear tangent space alignment-based manifold learning.

PURPOSE: To develop a manifold learning-based method that leverages the intrinsic low-dimensional structure of MR Spectroscopic Imaging (MRSI) signals for joint spectral quantification. METHODS: A linear tangent space alignment (LTSA) model was proposed to represent MRSI signals. In the proposed model, the signals of each metabolite were represented using a subspace model and the local coordinates of the subspaces were aligned to the global coordinates of the underlying low-dimensional manifold via linear transform. With the basis functions of the subspaces predetermined via quantum mechanics simulations, the global coordinates and the matrices for the local-to-global coordinate alignment were estimated by fitting the proposed LTSA model to noisy MRSI data with a spatial smoothness constraint on the global coordinates and a sparsity constraint on the matrices. RESULTS: The performance of the proposed method was validated using numerical simulation data and in vivo proton-MRSI experimental data acquired on healthy volunteers at 3T. The results of the proposed method were compared with the QUEST method and the subspace-based method. In all the compared cases, the proposed method achieved superior performance over the QUEST and the subspace-based methods both qualitatively in terms of noise and artifacts in the estimated metabolite concentration maps, and quantitatively in terms of spectral quantification accuracy measured by normalized root mean square errors. CONCLUSION: Joint spectral quantification using linear tangent space alignment-based manifold learning improves the accuracy of MRSI spectral quantification.

Manifold Learning via Linear Tangent Space Alignment (LTSA) for Accelerated Dynamic MRI With Sparse Sampling.

The spatial resolution and temporal frame-rate of dynamic magnetic resonance imaging (MRI) can be improved by reconstructing images from sparsely sampled k -space data with mathematical modeling of the underlying spatiotemporal signals. These models include sparsity models, linear subspace models, and non-linear manifold models. This work presents a novel linear tangent space alignment (LTSA) model-based framework that exploits the intrinsic low-dimensional manifold structure of dynamic images for accelerated dynamic MRI. The performance of the proposed method was evaluated and compared to state-of-the-art methods using numerical simulation studies as well as 2D and 3D in vivo cardiac imaging experiments. The proposed method achieved the best performance in image reconstruction among all the compared methods. The proposed method could prove useful for accelerating many MRI applications, including dynamic MRI, multi-parametric MRI, and MR spectroscopic imaging.

Variational Inference for Quantifying Inter-observer Variability in Segmentation of Anatomical Structures.

Lesions or organ boundaries visible through medical imaging data are often ambiguous, thus resulting in significant variations in multi-reader delineations, i.e., the source of aleatoric uncertainty. In particular, quantifying the inter-observer variability of manual annotations with Magnetic Resonance (MR) Imaging data plays a crucial role in establishing a reference standard for various diagnosis and treatment tasks. Most segmentation methods, however, simply model a mapping from an image to its single segmentation map and do not take the disagreement of annotators into consideration. In order to account for inter-observer variability, without sacrificing accuracy, we propose a novel variational inference framework to model the distribution of plausible segmentation maps, given a specific MR image, which explicitly represents the multi-reader variability. Specifically, we resort to a latent vector to encode the multi-reader variability and counteract the inherent information loss in the imaging data. Then, we apply a variational autoencoder network and optimize its evidence lower bound (ELBO) to efficiently approximate the distribution of the segmentation map, given an MR image. Experimental results, carried out with the QUBIQ brain growth MRI segmentation datasets with seven annotators, demonstrate the effectiveness of our approach.

Deep learning-based GTV contouring modeling inter- and intra- observer variability in sarcomas.

BACKGROUND AND PURPOSE: The delineation of the gross tumor volume (GTV) is a critical step for radiation therapy treatment planning. The delineation procedure is typically performed manually which exposes two major issues: cost and reproducibility. Delineation is a time-consuming process that is subject to inter- and intra-observer variability. While methods have been proposed to predict GTV contours, typical approaches ignore variability and therefore fail to utilize the valuable confidence information offered by multiple contours. MATERIALS AND METHODS: In this work we propose an automatic GTV contouring method for soft-tissue sarcomas from X-ray computed tomography (CT) images, using deep learning by integrating inter- and intra-observer variability in the learned model. Sixty-eight patients with soft tissue and bone sarcomas were considered in this evaluation, all underwent pre-operative CT imaging used to perform GTV delineation. Four radiation oncologists and radiologists performed three contouring trials each for all patients. We quantify variability by defining confidence levels based on the frequency of inclusion of a given voxel into the GTV and use a deep convolutional neural network to learn GTV confidence maps. RESULTS: Results were compared to confidence maps from the four readers as well as ground-truth consensus contours established jointly by all readers. The resulting continuous Dice score between predicted and true confidence maps was 87% and the Hausdorff distance was 14 mm. CONCLUSION: Results demonstrate the ability of the proposed method to predict accurate contours while utilizing variability and as such it can be used to improve clinical workflow.

Free-breathing 3D cardiac T1 mapping with transmit B1 correction at 3T.

PURPOSE: To develop a cardiac T1 mapping method for free-breathing 3D T1 mapping of the whole heart at 3 T with transmit B1 ( B1+ ) correction. METHODS: A free-breathing, electrocardiogram-gated inversion-recovery sequence with spoiled gradient-echo readout was developed and optimized for cardiac T1 mapping at 3 T. High-frame-rate dynamic images were reconstructed from sparse (k,t)-space data acquired along a stack-of-stars trajectory using a subspace-based method for accelerated imaging. Joint T1 and flip-angle estimation was performed in T1 mapping to improve its robustness to B1+ inhomogeneity. Subject-specific timing of data acquisition was used in the estimation to account for natural heart-rate variations during the imaging experiment. RESULTS: Simulations showed that accuracy and precision of T1 mapping can be improved with joint T1 and flip-angle estimation and optimized electrocardiogram-gated spoiled gradient echo-based inversion-recovery acquisition scheme. The phantom study showed good agreement between the T1 maps from the proposed method and the reference method. Three-dimensional cardiac T1 maps (40 slices) were obtained at a 1.9-mm in-plane and 4.5-mm through-plane spatial resolution from healthy subjects (n = 6) with an average imaging time of 14.2 ± 1.6 minutes (heartbeat rate: 64.2 ± 7.1 bpm), showing myocardial T1 values comparable to those obtained from modified Look-Locker inversion recovery. The proposed method generated B1+ maps with spatially smooth variation showing 21%-32% and 11%-15% variations across the septal-lateral and inferior-anterior regions of the myocardium in the left ventricle. CONCLUSION: The proposed method allows free-breathing 3D T1 mapping of the whole heart with transmit B1 correction in a practical imaging time.

Evidence-based assessment of potential therapeutic effects of adjunct osteopathic medicine for multidisciplinary care of acute and convalescent COVID-19 patients.

Although the COVID-19 pandemic affects predominantly the respiratory function, epidemiological studies show that multiple systems can be affected. The severe complications of SARS-CoV-2 infection seem to be induced by an inflammatory dysregulation ("cytokine storm"), which can also induce an immunodepression. Several studies highlight beneficial effects of osteopathic medicine on inflammation and immune regulation. A careful review of evidence-based literature brings to the fore significant improvements of osteopathic manipulative treatment (OMT) in adjunction to conventional care. OMT can improve the condition of infected patients by decreasing symptoms and boosting the efficiency of conventional care. OMT might also benefit surviving patients by reducing the long-lasting consequences of the infection as well as improving their quality of life during convalescence. This review should constitute an argument in favor of multidisciplinary care, although further biological and clinical research is essential to better assess the potential beneficial contributions of adjunct osteopathic medicine to conventional care in the fight against pandemics such as COVID-19.

Motion correction for PET data using subspace-based real-time MR imaging in simultaneous PET/MR.

Image quality of positron emission tomography (PET) reconstructions is degraded by subject motion occurring during the acquisition. Magnetic resonance (MR)-based motion correction approaches have been studied for PET/MR scanners and have been successful at capturing regular motion patterns, when used in conjunction with surrogate signals (e.g. navigators) to detect motion. However, handling irregular respiratory motion and bulk motion remains challenging. In this work, we propose an MR-based motion correction method relying on subspace-based real-time MR imaging to estimate motion fields used to correct PET reconstructions. We take advantage of the low-rank characteristics of dynamic MR images to reconstruct high-resolution MR images at high frame rates from highly undersampled k-space data. Reconstructed dynamic MR images are used to determine motion phases for PET reconstruction and estimate phase-to-phase nonrigid motion fields able to capture complex motion patterns such as irregular respiratory and bulk motion. MR-derived binning and motion fields are used for PET reconstruction to generate motion-corrected PET images. The proposed method was evaluated on in vivo data with irregular motion patterns. MR reconstructions accurately captured motion, outperforming state-of-the-art dynamic MR reconstruction techniques. Evaluation of PET reconstructions demonstrated the benefits of the proposed method in terms of motion artifacts reduction, improving the contrast-to-noise ratio by up to a factor 3 and achieveing a target-to-background ratio up to 90% superior compared to standard/uncorrected methods. The proposed method can improve the image quality of motion-corrected PET reconstructions in clinical applications.

Free-Breathing Three-Dimensional T1 Mapping of the Heart Using Subspace-Based Data Acquisition and Image Reconstruction.

Mapping the longitudinal relaxation time constant (T1) of the myocardium using Magnetic Resonance Imaging (MRI) is an emerging technique for quantitative assessment of the morphology and viability of the myocardium. However, three-dimensional (3D) T1 mapping of the heart is challenging due to the high dimensionality of the signal and the presence of cardiac and respiratory motions. We propose a subspace-based method for free-breathing 3D T1 mapping of the heart without respiratory gating. The image function is represented as a high-order partially separable (PS) function to explore the inherent spatiotemporal correlations of the underlying signal. A special data acquisition scheme enabled by the high-order PS model is used for sparse sampling of the (k,t)-space, where complementary sparse datasets are acquired, each covering only a small portion of the (k,t)-space to characterize a single subspace (spatial or temporal). High-resolution dynamic MR images are reconstructed from the highly undersampled (k,t)-space using low-rank tensor and sparsity constraints. We demonstrate the feasibility of our proposed method using in vivo data obtained from healthy subjects on a 3T MR scanner. The proposed method can enable new clinical applications of T1 mapping in cardiac MR.

Quantitative analysis of focal adhesion dynamics using photonic resonator outcoupler microscopy (PROM).

Focal adhesions are critical cell membrane components that regulate adhesion and migration and have cluster dimensions that correlate closely with adhesion engagement and migration speed. We utilized a label-free approach for dynamic, long-term, quantitative imaging of cell-surface interactions called photonic resonator outcoupler microscopy (PROM) in which membrane-associated protein aggregates outcoupled photons from the resonant evanescent field of a photonic crystal biosensor, resulting in a highly localized reduction of the reflected light intensity. By mapping the changes in the resonant reflected peak intensity from the biosensor surface, we demonstrate the ability of PROM to detect focal adhesion dimensions. Similar spatial distributions can be observed between PROM images and fluorescence-labeled images of focal adhesion areas in dental epithelial stem cells. In particular, we demonstrate that cell-surface contacts and focal adhesion formation can be imaged by two orthogonal label-free modalities in PROM simultaneously, providing a general-purpose tool for kinetic, high axial-resolution monitoring of cell interactions with basement membranes.

Quantitative Imaging of Cell Membrane-associated Effective Mass Density Using Photonic Crystal Enhanced Microscopy (PCEM).

Adhesion is a critical cellular process that contributes to migration, apoptosis, differentiation, and division. It is followed by the redistribution of cellular materials at the cell membrane or at the cell-surface interface for cells interacting with surfaces, such as basement membranes. Dynamic and quantitative tracking of changes in cell adhesion mass redistribution is challenging because cells are rapidly moving, inhomogeneous, and nonequilibrium objects, whose physical and mechanical properties are difficult to measure or predict. Here, we report a novel biosensor based microscopy approach termed Photonic Crystal Enhanced Microscopy (PCEM) that enables the movement of cellular materials at the plasma membrane of individual live cells to be dynamically monitored and quantitatively imaged. PCEM utilizes a photonic crystal biosensor surface, which can be coated with arbitrary extracellular matrix materials to facilitate cellular interactions, within a modified brightfield microscope with a low intensity non-coherent light source. Benefiting from the high sensitivity, narrow resonance peak, and tight spatial confinement of the evanescent field atop the photonic crystal biosensor, PCEM enables label-free live cell imaging with high sensitivity and high lateral and axial spatial-resolution, thereby allowing dynamic adhesion phenotyping of single cells without the use of fluorescent tags or stains. We apply PCEM to investigate adhesion and the early stage migration of different types of stem cells and cancer cells. By applying image processing algorithms to analyze the complex spatiotemporal information generated by PCEM, we offer insight into how the plasma membrane of anchorage dependent cells is dynamically organized during cell adhesion. The imaging and analysis results presented here provide a new tool for biologists to gain a deeper understanding of the fundamental mechanisms involved with cell adhesion and concurrent or subsequent migration events.

Numerical Surrogates for Human Observers in Myocardial Motion Evaluation From SPECT Images.

In medical imaging, the gold standard for image-quality assessment is a task-based approach in which one evaluates human observer performance for a given diagnostic task (e.g., detection of a myocardial perfusion or motion defect). To facilitate practical task-based image-quality assessment, model observers are needed as approximate surrogates for human observers. In cardiac-gated SPECT imaging, diagnosis relies on evaluation of the myocardial motion as well as perfusion. Model observers for the perfusion-defect detection task have been studied previously, but little effort has been devoted toward development of a model observer for cardiac-motion defect detection. In this work, we describe two model observers for predicting human observer performance in detection of cardiac-motion defects. Both proposed methods rely on motion features extracted using previously reported deformable mesh model for myocardium motion estimation. The first method is based on a Hotelling linear discriminant that is similar in concept to that used commonly for perfusion-defect detection. In the second method, based on relevance vector machines (RVM) for regression, we compute average human observer performance by first directly predicting individual human observer scores, and then using multi reader receiver operating characteristic analysis. Our results suggest that the proposed RVM model observer can predict human observer performance accurately, while the new Hotelling motion-defect detector is somewhat less effective.

Generalization Evaluation of Machine Learning Numerical Observers for Image Quality Assessment.

In this paper, we present two new numerical observers (NO) based on machine learning for image quality assessment. The proposed NOs aim to predict human observer performance in a cardiac perfusion-defect detection task for single-photon emission computed tomography (SPECT) images. Human observer (HumO) studies are now considered to be the gold standard for task-based evaluation of medical images. However such studies are impractical for use in early stages of development for imaging devices and algorithms, because they require extensive involvement of trained human observers who must evaluate a large number of images. To address this problem, numerical observers (also called model observers) have been developed as a surrogate for human observers. The channelized Hotelling observer (CHO), with or without internal noise model, is currently the most widely used NO of this kind. In our previous work we argued that development of a NO model to predict human observers' performance can be viewed as a machine learning (or system identification) problem. This consideration led us to develop a channelized support vector machine (CSVM) observer, a kernel-based regression model that greatly outperformed the popular and widely used CHO. This was especially evident when the numerical observers were evaluated in terms of generalization performance. To evaluate generalization we used a typical situation for the practical use of a numerical observer: after optimizing the NO (which for a CHO might consist of adjusting the internal noise model) based upon a broad set of reconstructed images, we tested it on a broad (but different) set of images obtained by a different reconstruction method. In this manuscript we aim to evaluate two new regression models that achieve accuracy higher than the CHO and comparable to our earlier CSVM method, while dramatically reducing model complexity and computation time. The new models are defined in a Bayesian machine-learning framework: a channelized relevance vector machine (CRVM) and a multi-kernel CRVM (MKCRVM).