Yu, Lifeng

AbstractPurposeTo demonstrate the feasibility of simultaneous dual‐contrast imaging in a large animal using a newly developed dual‐source energy‐integrating detector (EID)‐based multi‐energy computed tomography (MECT) system.MethodsTwo imaging tasks that may have potential clinical applications were investigated: head/neck (HN) CT angiography (CTA)/CT venography (CTV) with iodine and gadolinium, and small bowel imaging with iodine and bismuth in domestic swine. Dual‐source X‐ray beam configurations of 70 kV + Au120/Sn120 kV and 70 kV + Au140/Sn140 kV were used for the HN‐CTA/CTV and small bowel imaging studies, respectively. A test bolus scan was performed for each study. The regions of interest (ROIs) in the carotid artery and jugular vein for HN‐CTA/CTV imaging and abdominal aorta for small bowel imaging were used to determine the time‐attenuation curves, based on which the timing for contrast injection and the CT scan was determined. In the HN‐CTA/CTV study, an MECT scan was performed at the time point corresponding to the optimal arterial enhancement by iodine and the optimal venous enhancement by gadolinium. In the small bowel imaging study, an MECT scan was performed at the optimal time point to simultaneously capture the mesenteric arterial enhancement of iodine and the enteric enhancement of bismuth. Image‐based material decomposition was performed to decompose different materials for each study. To quantitatively characterize contrast material separation and misclassification, two ROIs on left common carotid artery and left internal jugular vein in HN‐CTA/CTV imaging and three ROIs on superior mesenteric artery, ileal lumen, and collapsed ileum (ileal wall) in small bowel imaging were placed to measure the mean concentration values and the standard deviations.ResultsIn the HN‐CTA/CTV study, common carotid arteries containing iodine and internal/external jugular veins containing gadolinium were clearly delineated from each other. Fine vessels such as cephalic veins and branches of external jugular veins were noticeable but clear visualization was hindered by image noise in gadolinium‐specific (CTV) images, as reviewed by a neuroradiologist. In the small bowel imaging study, the mesenteric arteries and collapsed bowel wall containing iodine and the small bowel loops containing bismuth were clearly distinctive from each other in the iodine‐ and bismuth‐specific images after material decomposition, as reviewed by an abdominal radiologist. Quantitative analyses showed that the misclassifications between the two contrast materials were less than 1.7 and 0.1 mg/ml for CTA/CTV and small bowel imaging studies, respectively.ConclusionsFeasibility of simultaneous CTA/CTV imaging in head and neck with iodine and gadolinium and simultaneous imaging of arterial and enteric phases of small bowel with iodine and bismuth, using a dual‐source EID‐MECT system, was demonstrated in a swine study. Compared to iodine and gadolinium in CTA/CTV, better delineation and classification of iodine and bismuth in small bowel imaging were achieved mainly due to wider separation between the corresponding two K‐edge energies.

AbstractPurposeConventional model observers (MO) in CT are often limited to a uniform background or varying background that is random and can be modeled in an analytical form. It is unclear if these conventional MOs can be readily generalized to predict human observer performance in clinical CT tasks that involve realistic anatomical background. Deep‐learning‐based model observers (DL‐MO) have recently been developed, but have not been validated for challenging low contrast diagnostic tasks in abdominal CT. We consequently sought to validate a DL‐MO for a low‐contrast hepatic metastases localization task.MethodsWe adapted our recently developed DL‐MO framework for the liver metastases localization task. Our previously‐validated projection‐domain lesion‐/noise‐insertion techniques were used to synthesize realistic positive and low‐dose abdominal CT exams, using the archived patient projection data. Ten experimental conditions were generated, which involved different lesion sizes/contrasts, radiation dose levels, and image reconstruction types. Each condition included 100 trials generated from a patient cohort of 7 cases. Each trial was presented as liver image patches (160×160×5 voxels). The DL‐MO performance was calculated for each condition and was compared with human observer performance, which was obtained by three sub‐specialized radiologists in an observer study. The performance of DL‐MO and radiologists was gauged by the area under localization receiver‐operating‐characteristic curves. The generalization performance of the DL‐MO was estimated with the repeated twofold cross‐validation method over the same set of trials used in the human observer study. A multi‐slice Channelized Hoteling Observers (CHO) was compared with the DL‐MO across the same experimental conditions.ResultsThe performance of DL‐MO was highly correlated to that of radiologists (Pearson's correlation coefficient: 0.987; 95% CI: [0.942, 0.997]). The performance level of DL‐MO was comparable to that of the grouped radiologists, that is, the mean performance difference was ‐3.3%. The CHO performance was poorer than the grouped radiologist performance, before internal noise could be added. The correlation between CHO and radiologists was weaker (Pearson's correlation coefficient: 0.812, and 95% CI: [0.378, 0.955]), and the corresponding performance bias (‐29.5%) was statistically significant.ConclusionThe presented study demonstrated the potential of using the DL‐MO for image quality assessment in patient abdominal CT tasks.

AbstractPurposeEye‐tracking approaches have been used to understand the visual search process in radiology. However, previous eye‐tracking work in computer tomography (CT) has been limited largely to single cross‐sectional images or video playback of the reconstructed volume, which do not accurately reflect radiologists' visual search activities and their interactivity with three‐dimensional image data at a computer workstation (e.g., scroll, pan, and zoom) for visual evaluation of diagnostic imaging targets. We have developed a platform that integrates eye‐tracking hardware with in‐house‐developed reader workstation software to allow monitoring of the visual search process and reader‐image interactions in clinically relevant reader tasks. The purpose of this work is to validate the spatial accuracy of eye‐tracking data using this platform for different eye‐tracking data acquisition modes.MethodsAn eye‐tracker was integrated with a previously developed workstation designed for reader performance studies. The integrated system captured real‐time eye movement and workstation events at 1000 Hz sampling frequency. The eye‐tracker was operated either in head‐stabilized mode or in free‐movement mode. In head‐stabilized mode, the reader positioned their head on a manufacturer‐provided chinrest. In free‐movement mode, a biofeedback tool emitted an audio cue when the head position was outside the data collection range (general biofeedback) or outside a narrower range of positions near the calibration position (strict biofeedback). Four radiologists and one resident were invited to participate in three studies to determine eye‐tracking spatial accuracy under three constraint conditions: head‐stabilized mode (i.e., with use of a chin rest), free movement with general biofeedback, and free movement with strict biofeedback. Study 1 evaluated the impact of head stabilization versus general or strict biofeedback using a cross‐hair target prior to the integration of the eye‐tracker with the image viewing workstation. In Study 2, after integration of the eye‐tracker and reader workstation, readers were asked to fixate on targets that were randomly distributed within a volumetric digital phantom. In Study 3, readers used the integrated system to scroll through volumetric patient CT angiographic images while fixating on the centerline of designated blood vessels (from the left coronary artery to dorsalis pedis artery). Spatial accuracy was quantified as the offset between the center of the intended target and the detected fixation using units of image pixels and the degree of visual angle.ResultsThe three head position constraint conditions yielded comparable accuracy in the studies using digital phantoms. For Study 1 involving the digital crosshairs, the median ± the standard deviation of offset values among readers were 15.2 ± 7.0 image pixels with the chinrest, 14.2 ± 3.6 image pixels with strict biofeedback, and 19.1 ± 6.5 image pixels with general biofeedback. For Study 2 using the random dot phantom, the median ± standard deviation offset values were 16.7 ± 28.8 pixels with use of a chinrest, 16.5 ± 24.6 pixels using strict biofeedback, and 18.0 ± 22.4 pixels using general biofeedback, which translated to a visual angle of about 0.8° for all three conditions. We found no obvious association between eye‐tracking accuracy and target size or view time. In Study 3 viewing patient images, use of the chinrest and strict biofeedback demonstrated comparable accuracy, while the use of general biofeedback demonstrated a slightly worse accuracy. The median ± standard deviation of offset values were 14.8 ± 11.4 pixels with use of a chinrest, 21.0 ± 16.2 pixels using strict biofeedback, and 29.7 ± 20.9 image pixels using general biofeedback. These corresponded to visual angles ranging from 0.7° to 1.3°.ConclusionsAn integrated eye‐tracker system to assess reader eye movement and interactive viewing in relation to imaging targets demonstrated reasonable spatial accuracy for assessment of visual fixation. The head‐free movement condition with audio biofeedback performed similarly to head‐stabilized mode.

AbstractObjectiveIn X‐raycomputed tomography (CT), many important clinical applications may benefit from a fast acquisition speed. The helical scan is the most widely used acquisition mode in clinical CT, where a fast helical pitch can improve the acquisition speed. However, on a typical single‐source helical CT (SSCT) system, the helical pitch p typically cannot exceed 1.5; otherwise, reconstruction artifacts will result from data insufficiency. The purpose of this work is to develop a deep convolutional neural network (CNN) to correct for artifacts caused by an ultra‐fast pitch, which can enable faster acquisition speed than what is currently achievable.MethodsA customized CNN (denoted as ultra‐fast‐pitch network (UFP‐net)) was developed to restore the underlying anatomical structure from the artifact‐corrupted post‐reconstruction data acquired from SSCT with ultra‐fast pitch (i.e., p ≥ 2). UFP‐net employed residual learning to capture the features of image artifacts. UFP‐net further deployed in‐house‐customized functional blocks with spatial‐domain local operators and frequency‐domain non‐local operators, to explore multi‐scale feature representation. Images of contrast‐enhanced patient exams (n = 83) with routine pitch setting (i.e., p < 1) were retrospectively collected, which were used as training and testing datasets. This patient cohort involved CT exams over different scan ranges of anatomy (chest, abdomen, and pelvis) and CT systems (Siemens Definition, Definition Flash, Definition AS+, Siemens Healthcare, Inc.), and the corresponding base CT scanning protocols used consistent settings of major scan parameters (e.g., collimation and pitch). Forward projection of the original images was calculated to synthesize helical CT scans with one regular pitch setting (p = 1) and two ultra‐fast‐pitch setting (p = 2 and 3). All patient images were reconstructed using the standard filtered‐back‐projection (FBP) algorithm. A customized multi‐stage training scheme was developed to incrementally optimize the parameters of UFP‐net, using ultra‐fast‐pitch images as network inputs and regular pitch images as labels. Visual inspection was conducted to evaluate image quality. Structural similarity index (SSIM) and relative root‐mean‐square error (rRMSE) were used as quantitative quality metrics.ResultsThe UFP‐net dramatically improved image quality over standard FBP at both ultra‐fast‐pitch settings. At p = 2, UFP‐net yielded higher mean SSIM (> 0.98) with lower mean rRMSE (< 2.9%), compared to FBP (mean SSIM < 0.93; mean rRMSE > 9.1%). Quantitative metrics at p = 3: UFP‐net—mean SSIM [0.86, 0.94] and mean rRMSE [5.0%, 8.2%]; FBP—mean SSIM [0.36, 0.61] and mean rRMSE [36.0%, 58.6%].ConclusionThe proposed UFP‐net has the potential to enable ultra‐fast data acquisition in clinical CT without sacrificing image quality. This method has demonstrated reasonable generalizability over different body parts when the corresponding CT exams involved consistent base scan parameters.

PurposeMulti‐energy computed tomography (MECT) has a great potential to enable many novel clinical applications such as simultaneous multi‐contrast imaging. The purpose of this study was to implement triple‐beam MECT on a traditional energy‐integrating‐detector (EID) CT platform (EID‐MECT).MethodsThis was accomplished by mounting a z‐axis split‐filter (0.05 mm Au, 0.6 mm Sn) on Tube A of a dual‐source EID CT scanner. With the two split x‐ray beams from Tube A and the third beam from Tube B, three beams with different x‐ray spectra can be simultaneously acquired. With Tube B operated at 70 or 80 kV and Tube A at 120 or 140 kV, four different triple‐beam configurations were calibrated for MECT measurements: 70/Au120/Sn120, 80/Au120/Sn120, 70/Au140/Sn140, and 80/Au140/Sn140 kV. Iodine (I), gadolinium (Gd), bismuth (Bi) samples, and their mixtures were prepared for 2 three‐material‐decomposition tasks and 1 four‐material‐decomposition task. For each task, samples were placed in a water phantom and scanned using each of the four triple‐beam configurations. For comparison, the same phantom was also scanned using three other dual‐energy CT (DECT) or MECT technologies: twin‐beam DECT (TB‐DECT), dual‐source DECT (DS‐DECT), and photon‐counting‐detector CT (PCD‐CT), all with optimal x‐ray spectrum settings and at equal volume CT dose index (CTDIvol). The phantom for four‐material decomposition (I/Gd/Bi/Water imaging) was scanned using the PCD‐CT only (140 kV with 25, 50, 75, and 90 keV). Image‐based material decomposition was performed to acquire material‐specific images, on which the mean basis material concentrations and noise levels were measured and compared across all triple‐beam configurations in EID‐MECT and various DECT/MECT systems.ResultsThe optimal triple‐beam configuration was task‐dependent with 70/Au120/Sn120, 70/Au140/Sn140, and 70/Au120/Sn120 kV for I/Gd/Water, I/Bi/Water, and I/Gd/Bi/Water material decomposition tasks, respectively. At equal radiation dose level, EID‐MECT provided comparable or better quantification accuracy in material‐specific images for all three material decomposition tasks, compared to EID‐based DECT and PCD‐CT systems. In terms of noise level comparison, EID‐MECT‐derived material‐specific images showed lower noise levels than TB‐DECT and DS‐DECT, but slightly higher than that from PCD‐CT in I/Gd/Water imaging. For I/Bi/Water imaging, EID‐MECT showed a comparable noise level to DS‐DECT, and a much lower noise level than TB‐DECT and PCD‐CT in all material‐specific images. For the four‐material decomposition task involving I/Gd/Bi/Water, the bismuth‐specific image derived from EID‐MECT was slightly noisier, but both iodine‐ and gadolinium‐specific images showed much lower noise levels in comparison to PCD‐CT.ConclusionsFor the first time, an EID‐based MECT system that can simultaneously acquire three x‐ray spectra measurements was implemented on a clinical scanner, which demonstrated comparable or better imaging performance than existing DECT and MECT systems.

AbstractPurposeTo demonstrate that variation in chemical composition has a negligible effect on the mapping curve from relative electron density (RED) to proton stopping power ratio (SPR), and to establish the theoretical framework of using Megavoltage (MV) computed tomography (CT), instead of kilovoltage (kV) dual energy CT, to accurately estimate proton SPR.MethodsA simulation study was performed to evaluate the effect of chemical composition variation on kVCT number and proton SPR. The simulation study involved both reference and simulated human tissues. The reference human tissues, together with their physical densities and chemical compositions, came from the ICRP publication 23. The simulated human tissues were created from the reference human tissues assuming that elemental percentage weight followed a Gaussian distribution. For all tissues, kVCT number and proton SPR were obtained through (a) theoretical calculation from tissue’s physical density and chemical composition which served as the ground truth, and (b) estimation from RED using the calibration curves established from the stoichiometric method. Deviations of the estimated values from the calculated values were quantified as errors in using RED to estimate kVCT number and proton SPR.ResultsGiven a chemical composition variation of 5% (1σ) of the nominal percentage weights, the total estimation error of using RED to estimate kVCT number was 0.34%, 0.62%, and 0.77% and the total estimation error of using RED to estimate proton SPR was 0.30%, 0.22%, and 0.16% for fat tissues, non‐fat soft tissues and bone tissues, respectively.ConclusionChemical composition had a negligible effect on the method of using RED to determine proton SPR. RED itself is sufficient to accurately determine proton SPR. MVCT number maintains a superb linear relationship with RED because it is highly dominated by Compton scattering. Therefore, MVCT has great potential in reducing the proton range uncertainty.

AbstractPurposeAn increasing number of implantable or external devices can impact whether patients can receive radiological imaging examinations. This study examines and tests the Neulasta (pegfilgrastim) Onpro on‐body injector in multiple imaging environments.MethodsThe injector was analyzed for four imaging modalities with testing protocols and strategies developed for each modality. In x‐ray and computed tomography (CT), scans with much higher exposure than clinical protocols were performed with the device attached to an anthropomorphic phantom. The device was monitored until the completion of drug delivery. For magnetic resonance imaging (MRI), the device was assessed using a hand‐held magnet and underwent the magnetically induced displacement testing in a 1.5T clinical MRI scanner room. For ultrasound, magnetic field changes were measured around an ultrasound scanner system with three transducers.ResultsFor x‐ray and CT no sign of device error was identified during or after the high radiation exposure scans. Drug delivery was completed at expected timing with expected volume. For MRI the device showed significant attractive force towards the hand‐held magnet and a 50‐degree deflection angle at 50 cm from the opening of the scanner bore. No further assessment from the gradient or radiofrequency field was deemed necessary. For ultrasound the maximum magnetic field change from baseline was measured to be +11.7 μT in comparison to +74.2 μT at 4 inches from a working microwave.ConclusionsNo device performance issue was identified under the extreme test conditions in x‐ray or CT. The device was found to be MR Unsafe. Magnetic field changes around an ultrasound system met the limitation set by manufacture. Patient ultrasound scanning is considered safe as long as the transducers do not inadvertently loosen the device.

AbstractIt is important to employ radiation dose reduction techniques in pediatric computed tomography (CT) to reduce potential risks of radiation‐induced malignancy. Automatic tube potential (kV) selection tools have been developed and become available on many CT scanners, which select the optimum kV based on the patient size and clinical task to improve the radiation dose efficiency. However, its use in pediatric CT has been mostly empirical, following manufacturer’s default recommendation without solid demonstration for quality improvement. This study aimed to implement an automatic tube potential tool (CAREkV, Siemens Healthcare) into routine pediatric CT practice, using the “Plan‐Do‐Study‐Act” quality improvement process, in place of an existing kV/mAs technique chart. The design of this quality improvement project involved Plan‐Do‐Study‐Act stages. Plan and Do stages identified the criteria for optimal automatic kV selection; a range of phantoms representing typical pediatric groups were scanned on a dual‐source 128‐slice scanner using a fast‐pitch scanning mode. The identified CAREkV settings were implemented into the CT protocol and evaluated after a 6‐month period. In the Study stage, an objective evaluation of the image metrics and radiation dose for two similar patient cohorts using CAREkV and the technique‐chart, respectively, were compared. The kV selected, image quality and radiation dose determined by CAREkV were comparable to those obtained while using the technique‐chart. The CAREkV was successfully implemented into our pediatric abdominopelvic CT practice. By utilizing the “PDSA” process optimal image quality and radiation dose reduction were achieved with an automatic kV selection tool to improve CT workflow.