Garry Gold | Stanford University (original) (raw)

Papers by Garry Gold

Recently, C-arm cone-beam CT systems have been used to acquire knee joints under weight-bearing c... more Recently, C-arm cone-beam CT systems have been used to acquire knee joints under weight-bearing conditions. For this purpose, the Carm acquires images on a horizontal trajectory around the standing patient, who shows involuntary motion. The current state-of-the-art reconstruction approach estimates motion based on fiducial markers attached to the knee. A drawback is that this method requires calibration prior to each scan, since the horizontal trajectory is not reproducible. In this work, we propose a novel method, which does not need a calibration scan. For comparison, we extended the stateof-the-art method with an iterative scheme and we further introduce a closed-form solution of the compensated projection matrices. For evaluation, a numerical phantom and clinical data are used. The novel approach and the extended state-of-the-art method achieve a reduction of the reprojection error of 94% for the phantom data. The improvement for the clinical data ranged between 10% and 80%, which is followed by the visual impression. Therefore, the novel approach and the extended state-of-the-art method achieve superior results compared to the state-of-the-art method.

American Journal of Roentgenology, Jun 1, 2021

BACKGROUND.Potential approaches for abbreviated knee MRI, including prospective acceleration with... more BACKGROUND.Potential approaches for abbreviated knee MRI, including prospective acceleration with deep learning, have achieved limited clinical implementation.OBJECTIVE.The objective of this study was to evaluate the interreader agreement between conventional knee MRI and a 5-minute 3D quantitative double-echo steady-state (qDESS) sequence with automatic T2 mapping and deep learning super-resolutionaugmentation and to compare the diagnostic performance of the two methods regarding findings from arthroscopic surgery.METHODS.Fifty-one patients with knee pain underwent knee MRI that included an additional 3D qDESS sequence with automatic T2 mapping. Fourier interpolation was followed by prospective deep learning super resolution to enhance qDESS slice resolution twofold. A musculoskeletal radiologist and a radiology resident performed independent retrospective evaluations of articular cartilage, menisci, ligaments, bones, extensor mechanism, and synovium using conventional MRI. Following a 2-month washout period, readers reviewed qDESS images alone followed by qDESS with the automatic T2 maps. Interreader agreement between conventional MRI and qDESS was computed using percentage agreement and Cohen kappa. The sensitivity and specificity of conventional MRI, qDESS alone, and qDESS plus T2 mapping were compared with arthroscopic findings using exact McNemar tests.RESULTS.Conventional MRI and qDESS showed 92% agreement in evaluating all tissues. Kappa was 0.79 (95% CI, 0.76–0.81) across all imaging findings. In 43 patients who underwent arthroscopy, sensitivity and specificity were not significantly different (p = .23 to > .99) between conventional MRI (sensitivity, 58–93%; specificity, 27–87%) and qDESS alone (sensitivity, 54–90%; specificity, 23–91%) for cartilage, menisci, ligaments, and synovium. For grade 1 cartilage lesions, sensitivity and specificity were 33% and 56%, respectively, for conventional MRI; 23% and 53% for qDESS (p = .81); and 46% and 39% for qDESS with T2 mapping (p = .80). For grade 2A lesions, values were 27% and 53% for conventional MRI, 26% and 52% for qDESS (p = .02), and 58% and 40% for qDESS with T2 mapping (p < .001).CONCLUSION.The qDESS method prospectively augmented with deep learning showed strong interreader agreement with conventional knee MRI and near-equivalent diagnostic performance regarding arthroscopy. The ability of qDESS to automatically generate T2 maps increases sensitivity for cartilage abnormalities.CLINICAL IMPACT.Using prospective artificial intelligence to enhance qDESS image quality may facilitate an abbreviated knee MRI protocol while generating quantitative T2 maps.

Basketball players represent a population with an inherently high risk of sustaining meniscal inj... more Basketball players represent a population with an inherently high risk of sustaining meniscal injuries. Evaluation of quantitative MRI (qMRI) metrics of the meniscus within these athletes may improve clinical insight. To date, no studies have longitudinally evaluated meniscal T2* values in high performance athletes. Therefore, the purpose of this study was to utilize ultra-short TE (UTE) MRI to longitudinally evaluate medial and lateral meniscal T2* values within elite weight-bearing (basketball) and non-weight bearing (swimmers) athletes. Significant differences of T2* values were found between the medial and lateral menisci. No significant difference of meniscal T2* values were found between basketball players and swimmers.

Lecture Notes in Computer Science, 2020

Involuntary motion during weight-bearing cone-beam computed tomography (CT) scans of the knee cau... more Involuntary motion during weight-bearing cone-beam computed tomography (CT) scans of the knee causes artifacts in the reconstructed volumes making them unusable for clinical diagnosis. Currently, image-based or marker-based methods are applied to correct for this motion, but often require long execution or preparation times. We propose to attach an inertial measurement unit (IMU) containing an accelerometer and a gyroscope to the leg of the subject in order to measure the motion during the scan and correct for it. To validate this approach, we present a simulation study using real motion measured with an optical 3D tracking system. With this motion, an XCAT numerical knee phantom is non-rigidly deformed during a simulated CT scan creating motion corrupted projections. A biomechanical model is animated with the same tracked motion in order to generate measurements of an IMU placed below the knee. In our proposed multi-stage algorithm, these signals are transformed to the global coordinate system of the CT scan and applied for motion compensation during reconstruction. Our proposed approach can effectively reduce motion artifacts in the reconstructed volumes. Compared to the motion corrupted case, the average structural similarity index and root mean squared error with respect to the no-motion case improved by 13-21% and 68-70%, respectively. These results are qualitatively and quantitatively on par with a state-of-theart marker-based method we compared our approach to. The presented study shows the feasibility of this novel approach, and yields promising results towards a purely IMU-based motion compensation in C-arm CT.

Proceedings of SPIE, Mar 22, 2016

C-arm-based cone-beam CT (CBCT) systems with flat-panel detectors are suitable for diagnostic kne... more C-arm-based cone-beam CT (CBCT) systems with flat-panel detectors are suitable for diagnostic knee imaging due to their potentially flexible selection of CT trajectories and wide volumetric beam coverage. In knee CT imaging, over-exposure artifacts can occur because of limitations in the dynamic range of the flat panel detectors present on most CBCT systems. We developed a straightforward but effective method for correction and detection of over-exposure for an Automatic Exposure Control (AEC)-enabled standard knee scan incorporating a prior low dose scan. The radiation dose associated with the low dose scan was negligible (0.0042mSv, 2.8% increase) which was enabled by partially sampling the projection images considering the geometry of the knees and lowering the dose further to be able to just see the skin-air interface. We combined the line integrals from the AEC and low dose scans after detecting over-exposed regions by comparing the line profiles of the two scans detector row-wise. The combined line integrals were reconstructed into a volumetric image using filtered back projection. We evaluated our method using in vivo human subject knee data. The proposed method effectively corrected and detected over-exposure, and thus recovered the visibility of exterior tissues (e.g., the shape and density of the patella, and the patellar tendon), incorporating a prior low dose scan with a negligible increase in radiation exposure.

Lecture Notes in Computer Science, 2018

Patient motion is one of the major challenges in cone-beam computed tomography (CBCT) scans acqui... more Patient motion is one of the major challenges in cone-beam computed tomography (CBCT) scans acquired under weight-bearing conditions, since it leads to severe artifacts in reconstructions. In knee imaging, a state-of-the-art approach to compensate for patient motion uses fiducial markers attached to the skin. However, marker placement is a tedious and time consuming procedure for both, the physician and the patient. In this manuscript we investigate the use of anatomical landmarks in an attempt to replace externally attached fiducial markers. To this end, we devise a method to automatically detect anatomical landmarks in projection domain X-ray images irrespective of the viewing direction. To overcome the need for annotation of every X-ray image and to assure consistent annotation across images from the same subject, annotations and projection images are generated from 3D CT data. Twelve landmarks are annotated in supine CBCT reconstructions of the knee joint and then propagated to synthetically generated projection images. Then, a sequential Convolutional Neuronal Network is trained to predict the desired landmarks in projection images. The network is evaluated on synthetic images and real clinical data. On synthetic data promising results are achieved with a mean prediction error of 8.4 ± 8.2 pixel. The network generalizes to real clinical data without the need of retraining. However, practical issues, such as the second leg entering the field of view, limit the performance of the method at this stage. Nevertheless, our results are promising and encourage further investigations on the use of anatomical landmarks for motion management.

Orthopaedic Proceedings, Oct 1, 2010

Introduction: The hip joint is usually considered a ball-in-socket. However, there have been few ... more Introduction: The hip joint is usually considered a ball-in-socket. However, there have been few studies evaluating normal hip kinematics and the contribution coming from soft tissues. Capsular laxity is at the basis of injury to the acetabular labrum (most common pathological lesion seen during hip arthroscopy). The objectives of this study were to (1) assess hip kinematics with all the soft tissues intact using a surgical navigation system, (2) assess the relative contributions of the soft tissues to hip stability and (3) assess the relative contributions of periarticular soft tissues to hip range of motion. Materials and Methods: We used 4 normal hemicorpse specimens for a total of 8 hips. A navigation system (KLEE, Orthokey) was used to acquire the kinematic data. The anatomical reference system was identified through the palpation of landmarks: (1) anterior superior iliac spines (ASIS) and (2) pelvic tubercles for the pelvis, (3) femoral head center and (4) epicondyles for femur. There were 12 passive kinematic tests repeated 3 times in 3 different limb conditions (‘intact’, ‘no-skin-muscle’, ‘labral tear’) to explore the whole kinematic range. We analysed the differences in flexion/extension, abduction/adduction, internal/external rotation ranges (Wilcoxon’s Signed Ranks Test). Results: The kinematic analysis applied on the limbs highlighted the following range of motion: (1) the F/E was 115.7 ± 2.4° (12.9 ± 1.0° in extension/101.7 ± 3.0° in flexion) in ‘intact’ limb, 139.2 ± 10.8° (14.7 ± 2.7° in extension/120.7 ± 8.6° in flexion) in ‘no-skin no-muscle’ condition, and 174.3 ± 34.1° (25.3 ± 0.5° in extension/147.4 ± 35.4° in flexion) in ‘capsule cut’ condition; all the ranges were statistically different (p < 0.05); (2) the A/A was 44.5 ± 13.7° (35.4 ± 1.5° in abduction/10.1 ± 13.4° in adduction) in ‘intact’ limb, 59.2 ± 1.8° (38.5 ± 3.2° in abduction/21.7 ± 0.7° in adduction) in ‘no-skin no-muscle’ condition, and 82.0 ± 4.6° (57.4 ± 2.5° in abduction/25.6 ± 6.8° in adduction) in ‘capsule cut’ condition; all the ranges were statistically different (p < 0.05); (3) the IR/ER was 52.2 ± 10.5° (32.0 ± 11.9° in IR/21.5 ± 1.0° in ER) in ‘intact’ limb, 59.2 ± 1.8° (36.1 ± 14.1° in IR/26.5 ± 1.2° in ER) in ‘no-skin no-muscle’ condition, and 116.4 ± 54.4° (58.2 ± 16.1° in IR/55.6 ± 36.3° in ER) in the ‘capsule cut’ condition; all the ranges were statistically different (p < 0.05), except the ranges of ‘intact’ condition and ‘no-skin no-muscles’ one (p = 0.37). Discussion: The study of the 3 different conditions highlighted the critical role of the soft tissues in hip stability and kinematics; the soft tissues do provide stability mainly in limiting hip range of motion. This study’s findings are a preliminary contribution in the understanding of the contribution of periarticular muscles, joint capsule and ligaments to hip kinematics.

We propose a method to reduce streak artifacts in cone-beam CT reconstructions that arise from th... more We propose a method to reduce streak artifacts in cone-beam CT reconstructions that arise from the edges of dense objects outside the 3D field-of-view. To this end, ramp filtering is decomposed into a derivative-and a Hilbert transform step. This allows for spectral inpainting directly in gradient domain, such that only contributions of sharp edges are removed. We applied our approach to weight-bearing knee imaging data, where plastic pipes outside the field-of-view introduce notable streak artifacts. Streak artifacts in the reconstructions are reduced and on semisimulated data the correlation coefficient could be improved from 0.88 to 0.99. The method is applicable for arbitrary object shapes and can be easily integrated into existing FDK reconstruction algorithms.

arXiv (Cornell University), Feb 24, 2021

Involuntary subject motion is the main source of artifacts in weight-bearing cone-beam CT of the ... more Involuntary subject motion is the main source of artifacts in weight-bearing cone-beam CT of the knee. To achieve image quality for clinical diagnosis, the motion needs to be compensated. We propose to use inertial measurement units (IMUs) attached to the leg for motion estimation. We perform a simulation study using real motion recorded with an optical tracking system. Three IMU-based correction approaches are evaluated, namely rigid motion correction, non-rigid 2D projection deformation and non-rigid 3D dynamic reconstruction. We present an initialization process based on the system geometry. With an IMU noise simulation, we investigate the applicability of the proposed methods in real applications. All proposed IMUbased approaches correct motion at least as good as a state-of-the-art marker-based approach. The structural similarity index and the root mean squared error between motion-free and motion corrected volumes are improved by 24-35% and 78-85%, respectively, compared with the uncorrected case. The noise analysis shows that the noise levels of commercially available IMUs need to be improved by a factor of 10 5 which is currently only achieved by specialized hardware not robust enough for the application. The presented study confirms the feasibility of this novel approach and defines improvements necessary for a real application.

WORLD SCIENTIFIC eBooks, Mar 23, 2014

Informatik aktuell, 2017

Recent C-arm CT systems allow for the examination of a patient's knees under weight-bearing condi... more Recent C-arm CT systems allow for the examination of a patient's knees under weight-bearing conditions. The standing patient tends to show involuntary motion, which introduces motion artifacts in the reconstruction. The state-of-the-art motion correction approach uses fiducial markers placed on the patients' skin to estimate rigid leg motion. Marker placement is tedious, time consuming and associated with patient discomfort. Further, motion on the skin surface does not reflect the internal bone motion. We propose a purely projection based motion estimation method using consistency conditions of X-ray projections. The epipolar consistency between all pairs of projections is optimized over various motion parameters. We validate our approach by simulating motion from a tracking system in forward projections of clinical data. We visually and numerically assess reconstruction image quality and show an improvement in Structural Similarity from 0.912 for the uncorrected case to 0.943 using the proposed method with a 3D translational motion model. Initial experiments showed promising results encouraging further investigation of practical applicability.

bioRxiv (Cold Spring Harbor Laboratory), Oct 2, 2021

The inability to detect early degenerative changes to the articular cartilage surface that common... more The inability to detect early degenerative changes to the articular cartilage surface that commonly precede bulk osteoarthritic degradation is an obstacle to early disease detection for research or clinical diagnosis. Leveraging a known artifact that blurs tissue boundaries in clinical arthrograms, contrast agent diffusivity can be derived from computed tomography arthrography (CTa) scans. We combined experimental and computational approaches to study protocol variations that may alter the CTa-derived apparent diffusivity. In experimental studies on bovine cartilage explants, we examined how contrast agent dilution and transport direction (absorption vs. desorption) influence the apparent diffusivity of untreated and enzymatically digested cartilage. Using multiphysics simulations, we examined mechanisms underlying experimental observations and the effects of image resolution, scan interval and early scan termination. The apparent diffusivity during absorption decreased with increasing contrast agent concentration by an amount similar to the increase induced by tissue digestion. Models indicated that osmotically induced fluid efflux strongly contributed to the concentration effect. Simulated changes to spatial resolution, scan spacing and total scan time all influenced the apparent diffusivity, indicating the importance of consistent protocols. With careful control of imaging protocols and interpretations guided by transport models, CTa-derived diffusivity offers promise as a biomarker for early degenerative changes.

Medical Physics, Sep 20, 2016

Springer eBooks, 2020

Analyzing knee cartilage thickness and strain under load can help to further the understanding of... more Analyzing knee cartilage thickness and strain under load can help to further the understanding of the effects of diseases like Osteoarthritis. A precise segmentation of the cartilage is a necessary prerequisite for this analysis. This segmentation task has mainly been addressed in Magnetic Resonance Imaging, and was rarely investigated on contrast-enhanced Computed Tomography, where contrast agent visualizes the border between femoral and tibial cartilage. To overcome the main drawback of manual segmentation, namely its high time investment, we propose to use a 3D Convolutional Neural Network for this task. The presented architecture consists of a V-Net with SeLu activation, and a Tversky loss function. Due to the high imbalance between very few cartilage pixels and many background pixels, a high false positive rate is to be expected. To reduce this rate, the two largest segmented point clouds are extracted using a connected component analysis, since they most likely represent the medial and lateral tibial cartilage surfaces. The resulting segmentations are compared to manual segmentations, and achieve on average a recall of 0.69, which confirms the feasibility of this approach.

Communications in computer and information science, 2017

Cone-beam C-arm CT systems allow to scan patients in weight-bearing positions to assess knee cart... more Cone-beam C-arm CT systems allow to scan patients in weight-bearing positions to assess knee cartilage health under more realistic conditions. Involuntary patient motion during the acquisition results in motion artifacts in the reconstructions. The current motion estimation method is based on fiducial markers. They can be tracked with a high spatial accuracy in the projection images, but only deliver sparse information. Further, placement of the markers on the patient's leg is time consuming and tedious. Instead of relying on a few well defined points, we seek to establish correspondences on dense surface data to estimate 3D displacements. In this feasibility study, motion corrupted X-ray projections and surface data are simulated. We investigate motion estimation by registration of the surface information. The proposed approach is compared to a motion free, an uncompensated, and a state-of-the-art marker-based reconstruction using the SSIM. The proposed approach yields motion estimation accuracy and image quality close to the current state-of-the-art, reducing the motion artifacts in the reconstructions remarkably. Using the proposed method, Structural Similarity improved from 0.887 to 0.975 compared to uncorrected images. The results are promising and encourage future work aiming at facilitating its practical applicability.

Acta Orthopaedica, Feb 4, 2021

ISMRM Annual Meeting

While deep-learning-based MRI reconstruction and image analysis methods have shown promise, few h... more While deep-learning-based MRI reconstruction and image analysis methods have shown promise, few have been translated to clinical practice. This may be a result of (1) paucity of end-to-end datasets that enable comprehensive evaluation from reconstruction to analysis and (2) discordance between conventional validation metrics and clinically useful endpoints. Here, we present the Stanford Knee MRI with Multi-Task Evaluation (SKM-TEA), a dataset of 155 clinical quantitative 3D knee MRI scans with k-space data, DICOM images, and dense tissue segmentation and pathology annotations to facilitate clinically relevant, comprehensive benchmarking of the MRI workflow. Dataset, code, and trained baselines are available at https://github.com/StanfordMIMI/skm-tea.

Quantitative Imaging in Medicine and Surgery, 2022

Background: This study investigated the utility of a 2-dimensional watershed algorithm for identi... more Background: This study investigated the utility of a 2-dimensional watershed algorithm for identifying the cartilage surface in computed tomography (CT) arthrograms of the knee up to 33 minutes after an intraarticular iohexol injection as boundary blurring increased. Methods: A 2D watershed algorithm was applied to CT arthrograms of 3 bovine stifle joints taken 3, 8, 18, and 33 minutes after iohexol injection and used to segment tibial cartilage. Thickness measurements were compared to a reference standard thickness measurement and the 3-minute time point scan. Results: 77.2% of cartilage thickness measurements were within 0.2 mm (1 voxel) of the thickness calculated in the reference scan at the 3-minute time point. 42% fewer voxels could be segmented from the 33-minute scan than the 3-minute scan due to diffusion of the contrast agent out of the joint space and into the cartilage, leading to blurring of the cartilage boundary. The traced watershed lines were closer to the location of the cartilage surface in areas where tissues were in direct contact with each other (cartilage-cartilage or cartilage-meniscus contact). Conclusions: The use of watershed dam lines to guide cartilage segmentation shows promise for identifying cartilage boundaries from CT arthrograms in areas where soft tissues are in direct contact with each other.