Fetterly, Kenneth A.

Abstract Aims Coronary angiography might contain clinically relevant information, beyond its traditional role in delineating coronary artery disease. We sought to explore the role of deep learning (DL) in detecting right ventricular dysfunction from routinely acquired cine images during coronary angiography. Methods and results 3D-convolutional neural networks (CNN) were trained to identify right ventricle (RV) dysfunction from diagnostic angiography of the right coronary artery (RCA). The model’s input was two cine angiograms from LAO and RAO projections. The model’s output was normal vs. abnormal RV function. Ground truth on RV function was obtained from the nearest transthoracic echocardiogram to the coronary angiogram. Two analyses were performed to identify (i) any RV dysfunction (≥mild) and (ii) significant RV dysfunction (≥mild–moderate). A total of 10 336 coronary angiograms from 9849 patients (mean age 66, 36% women) were included. The cohort was split into training (70%), validation (15%), and testing (15%) sets. In the testing sets, the models attained an area under the ROC curve (AUC), sensitivity, and specificity of 0.82 (95% CI: 0.80–0.84), 0.75, and 0.74, for detecting any RV dysfunction, and 0.83 (95% CI: 0.80–0.86), 0.82, and 0.70 for detecting significant (≥mild–moderate) RV dysfunction. Combining an ECG-driven AI model with the angio-driven model improved the AUC values to 0.83 and 0.87 for detecting any or advanced RV dysfunction, respectively. Conclusion A novel deep learning algorithm yielded acceptable classification accuracy in detecting RV dysfunction from routine cine angiography of the RCA. The model’s performance improved further by adding ECG to its inputs.

AbstractBackgroundPrior implementations of the channelized Hotelling observer (CHO) model have succeeded in assessing the performance of X‐ray angiography systems under a variety of imaging conditions. However, often times these conditions do not resemble those present in routine clinical imaging scenarios, such as having complex anthropomorphic backgrounds in conjunction with moving test objects.PurposeThis work builds up on prior established CHO methods and introduces a new approach to switch from the already established “multiple‐sample” CHO implementation to a “single‐sample” technique. The proposed implementation enables the inclusion of moving test objects upon nonuniform backgrounds by allowing only a single sample to represent the test object present condition that is to be used within the statistical test to estimate the detectability index.MethodsTo assess the proposed method, two image data sets were acquired with a clinical X‐ray angiography system. The first set consisted of a uniform background in combination with static test objects while the second consisted of an anthropomorphic chest phantom in conjunction with moving test objects. The first set was used to validate the proposed approach against the multiple‐sample method while the second was used to assess the feasibility of the proposed method under a variety of imaging conditions, including seven object sizes and seven detector target dose (DTD) levels.ResultsFor the uniform background data set, considering all detectability indices greater or equal than 1, the ratio between the detectability indices of the proposed single‐sample approach versus the multiple‐sample method was 0.997 ± 0.056 (range 0.884–1.159). The average single‐direction width of the 95% confidence intervals (CIs) of the detectability index estimates for the multiple‐sample method was 0.38 ± 0.43 (range 0.03–2.20). For the single‐sample approach, the average width was 2.52 ± 0.63 (range 1.11–5.44). For the anthropomorphic background image set, the results were consistent with classical quantum‐limited signal‐to‐noise ratio (SNR) theory. The magnitude of the detectability indices varied predictably with changes in both object size and DTD, with the highest value associated with the highest dose and the largest object size. Additionally, the proposed method was able to capture differences in the imaging performance for a given test object across the field of view, which was associated with the attenuation levels provided by different features of the anthropomorphic background.ConclusionsA new single‐sample variant of the CHO model to assess the performance of X‐ray angiography imaging systems is proposed. The new approach is consistent with quantum‐limited image quality theory and with a standard implementation of the CHO model. The proposed method enables the assessment of moving test objects in combination with complex, nonuniform image backgrounds, thereby opening the possibility to assess imaging conditions which more closely resemble those used in clinical care.

Data suggest that radiation-induced cataracts may form without a threshold and at low-radiation doses. Staff involved in interventional radiology and cardiology fluoroscopy-guided procedures have the potential to be exposed to radiation levels that may lead to eye lens injury and the occurrence of opacifications have been reported. Estimates of lens dose for various fluoroscopy procedures and predicted annual dosages have been provided in numerous publications. Available tools for eye lens radiation protection include accessory shields, drapes and glasses. While some tools are valuable, others provide limited protection to the eye. Reducing patient radiation dose will also reduce occupational exposure. Significant variability in reported dose measurements indicate dose levels are highly dependent on individual actions and exposure reduction is possible. Further follow-up studies of staff lens opacification are recommended along with eye lens dose measurements under current clinical practice conditions.

Abstract The purpose of this work was to assess technical performance of a prototype high-ratio (r29), 80 line cm −1 grid for imaging conditions which mimic those for adult cardiovascular angiography. The standard equipment r15, 80 line cm −1 grid was used as a reference. Plastic Water ® LR phantoms with thickness in the range 20–44 cm were used to simulate adult patient attenuation and scatter. Grids were tested using x-ray field of view 20 and 25 cm and x-ray source to detector distance (SID) 107 and 120 cm. The primary transmission fraction ( T P ) was measured using both narrow beam geometry and a lead beam stop (BS) technique. Scatter transmission ( T S ) was measured with the lead BS technique. The quantum signal to noise ratio improvement factor ( K SNR ) was used to describe relative grid performance. The experimental conditions required revised theory to assess grid performance. Theory to account for the detector glare and underestimation of scatter intensity by the lead BS method was developed. Also, novel K SNR theory was developed to allow direct comparison of two grids operated at different SID. Mean T P was modestly lower for the r29 versus r15 grid (0.69 versus 0.75). When tested under equivalent scatter condition, T S of the r29 grid was approximately ½ that of the r15 grid (0.18 versus 0.34). K SNR of the r29 grid at SID 120 cm compared to the r15 grid at SID 107 cm increased linearly with phantom thickness (range 1.0 to ∼1.16). Findings of this work indicate that the r29 grid used at SID 120 cm is expected to provide improved image quality (or reduced patient radiation dose) when compared to the r15 grid used at SID 107 cm for adult cardiovascular patients and that the potential benefit of the r29 grid increases with patient thickness >20 cm.

PurposeChannelized Hotelling observer (CHO) models have been implemented to assess imaging performance in x‐ray angiography systems. While current methods are appropriate for assessing unprocessed images of moving test objects upon uniform‐exposure backgrounds, they are inadequate for assessing conditions which more appropriately mimic clinical imaging conditions including the combination of moving test objects, complex anthropomorphic backgrounds, and image processing. In support of this broad goal, the purpose of this work was to develop theory and methods to automatically select a subset of task‐specific efficient Gabor channels from a task‐generic Gabor channel base set. Also, previously described theory and methods to manage detectability index (d′) bias due to nonrandom temporal variations in image electronic noise will be revisited herein.MethodsStarting with a base set of 96 Gabor channels, backward elimination of channels was used to automatically identify an “efficient” channel subset which reduced the number of channels retained in the subset while maintaining the magnitude of the d′ estimate. The concept of a pixelwise Hotelling observer (PHO) model was introduced and similarly implemented to assess the performance of the efficient‐channel CHO model. Bias in d′ estimates arising from temporally variable nonstationary noise was modeled as a bivariate probability density function for normal distributions, where one variable corresponds to the signal from the test object and the other variable corresponds to the signal from temporally variable nonstationary noise. Theory and methods were tested on uniform‐exposure unprocessed angiography images with detector target dose (DTD) of 6, 18, and 120 nGy containing static disk‐shaped test objects with diameter in the range of 0.5 to 4 mm.ResultsConsidering all DTD levels and test object sizes, the proposed method reduced the number of Gabor channels in the final subset by 63–82% compared to the original 96 Gabor channel base set, while maintaining a mean relative performance () of 95% 4% with respect to the reference PHO model. Experimental results demonstrated that the bivariate approach to account for bias due to temporally variable nonstationary noise resulted in improved correlation between the CHO and PHO models as compared to a previously proposed univariate approach.ConclusionsComputationally efficient backward elimination can be used to select an efficient subset of Gabor channels from an initial channel base set without substantially compromising the magnitude of the d′ estimate. Bias due to temporally variable nonstationary noise can be modeled through a bivariate approach leading to an improved unbiased estimate of d′.

AbstractObjectivesThe purpose of this study was to investigate the influence of simulated reduced‐dose three‐dimensional angiography (3DA) on the accuracy and precision of linear measurements derived from 3DA datasets.BackgroundThree‐dimensional angiography is performed during X‐ray guided interventional procedures to aid diagnosis and inform treatment strategies for children and adults with congenital heart disease. However, 3DA contributes substantially to patient radiation dose and may lead to an increased radiation‐induced cancer risk.MethodsReduced‐dose patient 3DA images were simulated by adding quantum noise to the 2D projection angiograms, then reconstructing the projection angiograms into the 3DA dataset. Dose reduction in the range 33–72% was simulated. Five observers performed 46 vessel diameter measurements along prespecified axes within 23 vessel segments from 11 patient 3DA datasets. Statistical tests were performed to assess the influence of radiation dose reduction on the accuracy and precision of vessel diameter measurements.ResultsVessel diameter measurements were in the range 5.9– 22.7 mm. Considering all vessel segments and observers, the influence of dose level on the accuracy of diameter measurements was in the range 0.02 – 0.15 mm (p .05–.8). Interobserver variability increased modestly with vessel diameter, but was not influence by dose level (p = .52). The statistical test for observer recall bias was negative (p = .51).ConclusionsSimulated dose reduction up to 72% did not affect the accuracy or precision of the diameter measurements acquired from 3DA images. These findings may embolden 3DA radiation dose reduction for pediatric and congenital heart disease patients.

Abstract Assessment of x-ray angiography system performance is typically performed using stationary test objects with simple geometries such as a disk on a uniform background. However, these methods do not represent realistic imaging conditions in interventional cardiology as anatomy and devices are inherently non-stationary due to cardiac motion. In this work, a novel implementation of the channelized Hotelling observer (CHO) was used to assess the influence of motion blur on object detectability. A standard CHO model assumes imaging system stationarity whereby the detectability index of a test object is independent of location. However, real angiography systems are inherently non-stationary. While vendor correction gain factors and offset maps are used to compensate for visual non-uniformities, these corrections do not restore stationarity to the images. Methods to accommodate non-stationarity and allow assessment of the influence of motion blur on test object detectability will be presented. The effect of motion blur was quantified with the relative detectability index ( ), where the for an object when moving with constant linear velocity was compared to a low velocity ‘pseudo-stationary’ condition to account for system non-stationarity. The pseudo-stationary condition was used to isolate the influences of spatial non-stationarity and motion blur. Three different test object shapes (disks, spheres and capsules) with linear velocity in the range 0–30 cm · s −1 were tested. For 1 mm diameter objects and linear velocity 30 cm · s −1 , was degraded by 37%, 33% and 42% for the disk, sphere and capsule respectively, relative to the pseudo-stationary condition. Considering all test objects with diameter greater than 2 mm and linear velocity 30 cm · s −1 , was degraded by less than 10% due to motion. In summary, this work describes a new approach to assess performance of x-ray angiography systems using the CHO model and moving test objects.