Computed tomography angiography with pulmonary artery thrombus burden and right-to-left ventricular diameter ratio after pulmonary embolism

Abstract

Purpose

Computed tomography angiography is used for quantifying the significance of pulmonary embolism, but its reliability has not been well defined.

Methods

The study cohort comprised 10 patients randomly selected from a 150-patient prospective trial of ultrasound-facilitated fibrinolysis for acute pulmonary embolism. Four reviewers independently evaluated the right-to-left ventricular diameter ratios using the standard multiplanar reformatted technique and a simplified (axial) method, and thrombus burden with the standard modified Miller score and a new, refined Miller scoring system.

Results

The intraclass correlation coefficient for intra-observer variability was .949 and .970 for the multiplanar reformatted and axial methods for estimating right-to-left ventricular ratios, respectively. Inter-observer agreement was high and similar for the two methods, with intraclass correlation coefficient of .969 and .976. The modified Miller score had good intra-observer agreement (intraclass correlation coefficient .820) and was similar to the refined Miller method (intraclass correlation coefficient .883) for estimating thrombus burden. Inter-observer agreement was also comparable between the techniques, with intraclass correlation coefficient of .829 and .914 for the modified Miller and refined Miller methods.

Conclusions

The reliability of computed tomography angiography for pulmonary embolism was excellent for the axial and multiplanar reformatted methods for quantifying the right-to-left ventricular ratio and for the modified Miller and refined Miller scores for quantifying of pulmonary artery thrombus burden.

Keywords

Pulmonary embolism computed tomography pulmonary angiography right-to-left ventricular ratio Miller Score reliability

Introduction

Venous thromboembolic events in the United States occur annually in 1 to 2 individuals per 1000 population or in approximately 300,000–600,000 cases.^1,2 Pulmonary embolism (PE) is the presenting event in about one-third of these patients and has a high case fatality rate.³ Right ventricular (RV) failure due to PE may result in hemodynamic collapse, cardiogenic shock, and early mortality.⁴ Ultrasound-facilitated, catheter-directed thrombolysis, which alleviates RV dysfunction, was recently approved by the US Food and Drug administration for the treatment of PE.^5–8

Effective therapy for PE requires rapid, accurate diagnosis and risk stratification, which is facilitated by computed tomography angiography (CTA).^9,10 Concomitant RV dysfunction can be estimated by CTA with measurement of the right-to-left ventricular (RV/LV) diameter ratio.¹¹ Imaging findings help clinicians to stratify patients to anticoagulation alone plus advanced therapy, such as catheter-directed thrombolysis.^12,13 Further, CTA is useful to quantitate thrombus dissolution and RV normalization during and after treatment.¹⁴

While CTA is valuable in the diagnosis and management of PE, the methodology for measuring the RV/LV ratio is not well standardized. Measurement techniques vary, rendering problematic comparisons among clinical trials, individual patients, and longitudinal examinations of a single patient. Our project assesses the reliability of different CTA measures for quantifying the functional and anatomic significance of PE, assessing the intra-observer agreement and inter-observer agreement of CTA endpoints in patients with PE.

Materials and methods

The study cohort was a randomly selected 10-patient subset of the SEATTLE II trial, a 150-patient prospective multicenter single-arm study of ultrasound-facilitated, catheter-directed thrombolysis for acute massive and submassive PE (NCT01513759).¹⁵ Patients aged ≥18 years of age with symptoms ≤14 days in duration were eligible for inclusion after a contrast-enhanced computed tomography imaging study demonstrated evidence of proximal PE. Thrombus was required to be present in at least one main or segmental pulmonary artery with an RV/LV diameter ratio of ≥0.9. The study was approved by the institutional review board of each investigational site, and each written informed consent was from obtained from each patient.

Each patient was required to have technically adequate CTA studies that passed internal quality assurance procedures. Each CTA was a high-quality non-gated contrast-enhanced chest computed tomographic study with adequate visualization of the ventricles and both lung fields. Images were reconstructed at ≤3.0 mm slices at ≤1.5-mm increments. CTA studies were reviewed using Aquarius iNtuition software (TeraRecon, Foster City, CA, USA). The cohort evaluated in the current study was randomly selected from the group of patients with imaging studies meeting these criteria. A random number between 0 and 1 was assigned to each of the 139 patients with technically adequate pre- and post-procedure CTA imaging studies (Microsoft Excel 2013, Redmond, WA). These patients were sequenced from low to high by the random numbers, and every 10th patient was chosen until 10 subjects were identified for the analytic subset.¹⁶

While the CTA studies were received from the investigational sites in an anonymized form, each study was secondarily de-identified to remove the study subject unique identifier. Two copies of each CTA study were generated, and each copy was assigned different identifiers, one for the first read and one for the second. In this manner, complete blinding of the readers was assured for the analysis of intra-observer agreement between two reads of the same study. In all, the 10 studies (six baseline pre-intervention and four post-treatment studies) were assessed by four independent readers. Among the readers, two were core laboratory technicians, one was the core laboratory director, and one was a physician reader. Intra-observer agreement analyses were performed on the measurements of the two reads for each study, performed at least 60 days apart, to reduce the chance that a reader might recognize the images from one review to the next.

RV/LV diameter ratio

Two different methods were used to determine the RV/LV ratio. The first was a method that used non-reformatted CTA axial images without manipulation in any vector (“axial method”). The second was modified from the method reported by Quiroz that used multiplanar reformatted reconstructions (MPR) from CTA datasets to measure the short axis of each ventricle (MPR method).¹⁷

In the axial method, the diameters of the right and left ventricles were measured in the cross-sectional transverse slices along the straight z-axis (craniocaudal axis), without manipulating the CTA datasets to generate images in other planes. The best single transverse slice was visually identified, scrolling through the slices to find the slice with adequate views of each ventricle approximating their largest diameters. The right and left ventricle diameters were measured on the same non-reformatted slice, accepting the discrepancies inherent in unadjusted, non-orthogonal planes. The maximum perpendicular distance from the free-wall ventricular endocardium to the interventricular septum was identified from these views.

In MPR (modified Quiroz) technique, the RV/LV ratio was measured in three-dimensional (3D) reformatted reconstructions, adjusting the axial, coronal, and sagittal planes to view the largest diameter of each ventricle as measured perpendicular to the interventricular septum (Figure 1).¹⁷ In the MPR method, the obliquities were adjusted to visualize and measure the diameters in a plane with the short axis perpendicular to the interventricular septum and the long axis parallel to a line defined by the center of the mitral valve and apex of the heart. As in the axial method, the diameters of the right and left ventricles were measured from the ventricular endocardium of the free wall to the ventricular endocardium of the septum. In contrast to the axial method, however, diameters of the right and left ventricles diameters were measured on separate MPR reconstructions, adjusting the planes to obtain obliquities appropriate for each ventricle.

Figure 1.

Right ventricle (RV) and left ventricle (LV) diameter determination from the three-dimensional reformatted views. Insets A and C are axial views and insets B and D are sagittal views. The axial planes (represented by the red line), coronal planes (represented by the green line) and sagittal planes (represented by the blue line) were adjusted to view the widest part of the ventricle perpendicular to the interventricular septum. In insets A and C, the diameter of each ventricle was taken in the plane with the short axis (blue line) perpendicular to the interventricular septum, and the long axis parallel to the line going through the center of the mitral valve and apex of the heart. The LV diameter (inset A, yellow line) and RV diameter (inset C, yellow line) were measured from the ventricular endocardium to the interventricular septum.

Pulmonary artery thrombus burden

Thrombus burden was determined with the modified Miller (MM) score as described by Qanadli et al. (Table 1).¹⁸ The method accounts for the anatomy of the pulmonary vasculature observed in surgical specimens, distinct from the anatomic configuration employed in the original pulmonary angiography-based Miller score.^19,20 Each lung is regarded as having 10 segmental arteries; three to the upper lobe, two to the middle and lingula, and five to the lower lobe. Cross-sectional reformatted views of the involved pulmonary arteries are viewed to assess the degree of diameter reduction of major and segmental pulmonary arteries from intraluminal thrombus. A three-point ordinal scale is used in the MM scoring system as follows: 0 for no obstruction, 1 for partial obstruction, and 2 for complete obstruction. Each major artery (main pulmonary artery, left and right pulmonary arteries, left and right interlobar arteries, left and right basal trunks) is given a weighted score based on the number of segmental arteries arising distally. The weighted score of the major artery and the sum of the subsequent segmental arteries are compared. The larger number is taken as the score for that respective artery. For example, if a major artery is partially occluded, but its segmental branches are each fully occluded, the segmental score is recorded. A cumulative score is calculated by summing the scores for all arteries. Scores can range from 0 to 40 (maximum of 20 per side).

Table 1.

The pulmonary arteries and assignment of modified Miller (MM) and refined Miller (RM) scores.

Pulmonary arteries	Order	MM scores	RM scores	Weighting factor	Maximum score	Observed MM score^a	Observed RM score^a
Main pulmonary artery	0	0, 1, 2	0, 0.5, 1.0, 1.5, 2	20	40	0.0 ± 0.0	0.0 ± 0.0
Right and left pulmonary artery	First	0, 1, 2	0, 0.5, 1.0, 1.5, 2	10	20	0.2 ± 0.4	0.2 ± 0.5
Apical anterior segmental	Third	0, 1, 2	0, 1, 2	1	2	0.8 ± 0.5	0.8 ± 0.5
Apical posterior segmental	Third	0, 1, 2	0, 1, 2	1	2	0.6 ± 0.6	0.8 ± 0.5
Apical segmental	Third	0, 1, 2	0, 1, 2	1	2	0.6 ± 0.6	0.6 ± 0.6
Interlobar artery	Second	0, 1, 2	0, 0.5, 1.0, 1.5, 2	7	14	0.7 ± 0.6	0.6 ± 0.6
Medial segmental^b	Third	0, 1, 2	0, 1, 2	1	2	0.9 ± 0.6	0.9 ± 0.6
Lateral segmental^b	Third	0, 1, 2	0, 1, 2	1	2	0.8 ± 0.6	0.8 ± 0.6
Superior basal segmental	Third	0, 1, 2	0, 1, 2	1	2	0.7 ± 0.6	0.7 ± 0.6
Basal trunk	Second	0, 1, 2	0, 0.5, 1.0, 1.5, 2	4	8	0.9 ± 0.5	0.9 ± 0.6
Lateral basal segmental	Third	0, 1, 2	0, 1, 2	1	2	0.7 ± 0.6	0.7 ± 0.6
Posterior basal segmental	Third	0, 1, 2	0, 1, 2	1	2	0.8 ± 0.6	0.8 ± 0.6
Medial basal segmental	Third	0, 1, 2	0, 1, 2	1	2	0.6 ± 0.6	0.6 ± 0.6
Anterior basal segmental	Third	0, 1, 2	0, 1, 2	1	2	0.7 ± 0.6	0.7 ± 0.6

Mean ± Standard Deviation.

On the left side, these are replaced by the superior and inferior lingular arteries as the third-order branches.

To account for a wide spectrum of major pulmonary artery obstruction, a refined Miller (RM) scoring system was developed as a modification of the MM score. Like the MM score, the RM score regards each lung as having 10 segmental arteries, and the degree of obstruction of the segmental pulmonary arteries is measured using an ordinal scale. In contrast to the MM, however, the RM score allows for a finer distinction of partial obstruction in seven major arteries; the main pulmonary artery, left and right pulmonary artery, left and right interlobar arteries, left and right basal trunks. Partial obstruction is classified in three categories based upon cross-sectional diameter reduction: 0.5 for 1–33%, 1.0 for 34–66%, and 1.5 for 67–99% obstruction (Figure 2). As in the MM scoring system, the weighted score of the major supplying artery is compared with the sum of its tributaries, and the larger number is recorded. For example, if thrombus obstructed the lumen of a basal trunk vessel by 50% (score = 1.0 × 7), but its four segmental tributaries were fully occluded (score = 2.0 × 4), the higher value of 8 would be recorded as the RM score. A cumulative score is calculated by summing the scores for all arteries. Similar to the MM score, the RM score can range from 0 to 20 per lung, for a maximum of 40 bilaterally.

Figure 2.

(a) CTA with thrombus in left basal trunk with a refined Miller score of 1.5 and modified Miller score of 1. (b) Thrombus in right basal trunk with a refined Miller score of 0.5 and modified Miller score of 1.

Statistical analysis

Data analysis was performed with SPSS version 22 (International Business Machines Corporation, Armonk, New York, USA). All values are expressed as means ± standard deviation (SD), median, and range. p Values were considered significant when the two-tailed alpha level was less than 0.05. Analyses that do not involve the first and second reads are represented by data from the first read.

The RV/LV diameter ratios and thrombus burden scores were assessed as continuous variables. Intra-observer agreement was evaluated by calculating the intraclass correlation coefficients (ICC) between the first and second measurements for each reviewer. Correlations were calculated with Spearman’s rho. Reliability of each measure was estimated with ICC and respective 95% confidence intervals (CI).²¹ ICC were calculated on a scale of 0 to 1, where 1 represents perfect reliability and no measurement error and 0 indicates no reliability. Intra-observer agreement was calculated as the ICC between the first and second reads of the four reviewers. Inter-observer agreement was calculated as the ICC between the measurements for the four reviewers on each of the ten CTA images. Bland–Altman plots were constructed as scatter plots in which the Y axis represented the difference between two paired measurements, and the X axis represented the average of these measurements.²² The graphical modification of Jones was used for assessing agreement for multiple observers.²³ Horizontal lines depict the mean difference and two standard deviations above and below the mean.

Results

No patient in the study cohort had thrombus in the main pulmonary artery. The RV and LV diameter measurements and RV/LV ratios were well correlated between the site-reported and the core laboratory values, whether the MPR or axial core laboratory methodology was used (Table 2). The site-determined values, however, slightly underestimated the ventricular diameters; RV to a greater extent than LV (p < .001 or all companions). Differences between the site-reported and core laboratory RV/LV values were not statistically distinct.

Table 2.

Site-reported and core laboratory pre-treatment right-to-left ventricular diameter measurements and ratios in the 150-patient SEATTLE-II trial.

Measure	Method	Mean ± SD	p vs. site-reported measurement^a
RV diameter	Site-reported	49 ± 7 mm	–
	Core laboratory MPR	53 ± 8 mm	<.001
	Core laboratory axial	54 ± 7 mm	<.001
LV diameter	Site-reported	33 ± 7 mm	–
	Core laboratory MPR	35 ± 7 mm	<.001
	Core laboratory axial	36 ± 8 mm	<.001
RV/LV ratio	Site-reported	1.55 ± 0.42	–
	Core laboratory MPR	1.58 ± 0.43	.153
	Core laboratory axial	1.59 ± 0.43	.069

Two-tailed paired t-test between the site-reported and core-laboratory measurements.

LV: left ventricle; MPR: multiplanar reformatted reconstruction; RV: right ventricle.

RV/LV ratio

Using the MPR method for RV/LV ratio, the mean RV/LV ratio was 1.14 ± 0.26 (median 1.11, range 0.83–1.94) on the first read and 1.16 ± 0.26 (median 1.14, range 0.81–1.83) on the second read. The intra-observer agreement was high, with an ICC of .949 (CI .906–.973). The inter-observer agreement was also excellent, with an ICC of .969 (CI .922–.991) (Table 3, Figure 3).

Figure 3.

Bland–Altman plot of inter-rater reliability for right ventricle–left ventricle (RV/LV) diameter ratio as measured by the axial method, without multiplanar reformatting of CTA images.

Table 3.

Intraclass correlation coefficients (ICC) for right ventricle (RV) and left ventricle (LV) diameter measurements and RV/LV ratios measured with the multiplanar reformatting (MPR) and Axial techniques.

Measure	Method	Intra-observer ICC	Inter-observer ICC
RV diameter	MPR	.872 (.771–.930)	.963 (.907–.990)
	Axial	.960 (.927–.979)	.965 (.912–.990)
LV diameter	MPR	.905 (.828–.949)	.973 (.932–.992)
	Axial	.983 (.968–.991)	.962 (.905–.989)
RV/LV ratio	MPR	.949 (.906–.973)	.969 (.922–.991)
	Axial	.970 (.944–.984)	.976 (.939–.993)

Note: None of the differences in ICC between the MPR and Axial methods were statistically significant.

When the RV/LV ratios were measured using the axial method, results were similar to those measured with the MPR method. The mean RV/LV ratio was 1.12 ± 0.24 (median 1.07, range 0.83–1.77) on the first read and 1.13 + 0.26 (median 1.09, range 0.84–1.84). The intra-observer agreement of the axial method was as high or higher than observed using the MPR method, with an ICC of .970 (CI .944–.984). The inter-observer agreement was also excellent using the axial method, with an ICC of .976 (CI .939–.993) (Figure 4). When the individual measurements for RV and LV diameter was assessed, it appeared that the reliability of measurements for the LV diameter was slightly improved over those for the RV, but these differences did not attain statistical significance.

Figure 4.

Bland–Altman plot of inter-rater reliability for right ventricle–left ventricle (RV/LV) diameter ratio as measured by the MPR (multiplanar reformatted reconstruction) method.

There was a high degree of correlation between the MPR and axial methods of measuring the RV/LV ratio (Spearman’s rho = .874, p < .001, Figure 5). However, a small difference between the two methods was observed. On average, the MPR estimate was 0.02 ± 0.08 greater than the axial estimate on the first read (p = .093) and 0.04 ± 0.11 greater on the second read (p = .048). The correlation was slightly higher for the LV measurement than the RV measurement (Spearman’s rho = .907 and .867, respectively).

Figure 5.

Correlation between the two methods of measuring the right-to-left ventricle (RV/LV) diameter ratio. On average, the more simplistic axial method underestimated the RV/LV ratio by only 0.02.

Pulmonary artery thrombus burden

The MM score for pulmonary artery thrombus burden averaged 18.4 ± 7.1 (median 20, range 3–32) on the first read and 18.2 ± 6.6 (median 19, range 1–29) on the second read. Intra-observer agreement was good, with an ICC of .820 (CI .685–.900). Inter-observer agreement was also good, with an ICC of .829 (CI .629–.948) (Figure 6).

Figure 6.

Bland–Altman plot of inter-rater reliability for the modified Miller (MM) scores for pulmonary artery thrombus burden.

Intra-observer agreement appeared better with the RM scoring system than with the MM method. The repeated observation ICC for the RM score was .883 (CI .789–.936). Inter-observer agreement also appeared to be slightly improved with the RM system, with an ICC of .914 (CI .796–.975) (Figure 7).

Figure 7.

Bland–Altman plot of inter-rater reliability for the refined Miller (RM) scores for pulmonary artery thrombus burden, using five categories of major pulmonary artery obstruction. In addition to zero and 1 for no thrombus and complete obstruction, three intermediate categories included 0.5 for 1–33%, 1.0 for 34–66%, and 1.5 for 67–99% reduction in luminal diameter. MM: modified Miller score.

Thrombus burden calculated by the MM score and the RM scores was highly correlated (Spearman’s rho = .937, Figure 8). The MM score tended to be slightly lower for a given patient, with a mean difference of −0.2 ± 2.6 and −0.1 ± 1.8 on the first and second reads, respectively The differences between the two methods for scoring pulmonary artery thrombus did not attain statistical significance (p = .581 and .664, respectively).

Figure 8.

Correlation between the two methods of measuring the pulmonary artery thrombus burden.

Correlation between RV/LV diameter ratio and thrombus burden

There was a correlation between RV/LV ventricular diameter ratios and the pulmonary artery thrombus burden. The correlations between the RV/LV ratio as determined by the MPR method and the MM and RM pulmonary artery thrombus scores were .526 (p < .001) and .509 (p = .001), respectively (Spearman rho). The correlations between the RV/LV ratio as determined by the axial method and the MM and RM scores were less striking but still statistically significant, with Spearman coefficients of .341 (p = .032) and .334 (p = .035), respectively.

Discussion

The current analysis suggests that the axial measurement of the RV/LV ratio yields results similar to the more complex and more time-consuming MPR measurement. Furthermore, the RM score modification of the standard MM score results in slightly improved diagnostic reliability. Contrary to the findings of others, the RV/LV ratio is highly correlated with the degree of pulmonary artery obstruction.

The binary diagnostic performance of CTA as an imaging test to discriminate between patients with and without PE does not encompass the full value of the test. In an era of increasing use of pharmacologic, mechanical, and combined methods to reduce the pulmonary artery thrombus burden, CTA helps to risk stratify, target the anatomic locations of occlusive thrombi, and gauge the response to therapy. Confidence in the precision of a CTA measurement can only be assured if the variability of measurements among readers is low. Reliability attains importance in the conduct of a clinical trial, but also in the real-world management of patients with PE. Evaluation of the intra- and inter-observer agreement of CTA is a crucial factor in assessing the comparative value of measurement protocols.

Intra-rater and inter-rater agreement for the MM score were acceptable, with ICC of .82 and .83, respectively, indicating that just under 20% of the variation was caused by measurement error. RM score, intended to be a refinement of the MM score with an increased ability to detect finer grades of major pulmonary artery obstruction, was associated with slightly improved intra- and inter-observer agreement, with ICC of 0.89 and 0.91, respectively. Compared with MM score, the RM score allows a finer discrimination of non-occlusive major pulmonary artery obstruction; adding two additional categories of differentiation.

The reproducibility of CTA in detecting RV dysfunction in acute PE was assessed by Kang et al.²⁴ While ICC was not calculated in this study, the investigators found a mean difference of .014 ± .195 in axial RV/LV ratios measured by two observers. Furlan et al. used a semi-automated method of quantifying the volume of thrombus in 30 patients with PE.²⁵ The intra- and inter-observer variability was extremely low in this series, with ICC of approximately .99. On Bland–Altman analyses, the inter-observer agreement was approximately 5%, with 95% of the readings within approximately 10% of one another.

Apart from reproducibility, a valuable imaging measurement technique must be rapid and easy to perform if it is to be readily incorporated into the routine diagnostic testing used in real-world clinical practice. In this regard, we compared the straightforward axial method of measuring the RV/LV ratio as an alternative to the more complex MPR technique. The axial method can be performed in less than a minute with unreconstructed images, while the MPR method requires sophisticated image analysis software applications, with attendant skills and time to perform the multiplanar reconstructions. Despite its relative simplicity, the axial method was as reliable as the MPR method. This finding is in agreement with the observations of Lu et al., where axial CTA images were as accurate as reformatted four-chamber views in the prediction of 30-day mortality after PE.²⁶ While reliability alone does not constitute the ultimate value of a test, it was also reassuring that the axial and MPR methods were highly correlated with one another. These observations suggest that the axial method for estimating the RV/LV ratio is acceptable at least in the context of clinical practice.

The current study is limited by its relatively small sample size, although the design is strengthened by the use of four individual readers. The study population included only those patients in the SEATTLE-II trial, where an RV/LV ratio >0.9 was a criterion for eligibility. Whether the findings are generalizable to patients with less severe PE or to those without PE cannot be established from the current data. The analysis performed by our dedicated imaging core laboratory may not be easy to replicate at most clinical centers, thereby limiting the able to generalize our findings. The current study did not evaluate the validity of the CTA measurement protocols or of the indices themselves.

In summary, the current study documents excellent reliability of the MM and RM scores for the quantification of pulmonary artery thrombus and for the axial and MPR methods of quantifying the RV/LV ratio. These measures should be considered when quantification of pulmonary artery thrombus and its effects on the right ventricle are important in the research or clinical setting.

Footnotes

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The employer of authors KO, RL and YL received research funds from the sponsor of the study.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by Ekos Corporation, a BTG International Group company. Ekos reviewed and commented on drafts of the manuscript but was not involved in imaging data collection, data analysis, or the drafting of the manuscript.

ClinicalTrials.gov Identifier

NCT01513759.

References

Beckman

Hooper

Critchley

. Venous thromboembolism: a public health concern. Am J Prev Med 2010; 38: S495–S501.

Horlander

Mannino

Leeper

. Pulmonary embolism mortality in the United States, 1979–1998: an analysis using multiple-cause mortality data. Arch Intern Med 2003; 163: 1711–1717.

Laporte

Mismetti

Decousus

. Clinical predictors for fatal pulmonary embolism in 15,520 patients with venous thromboembolism: findings from the Registro Informatizado de la Enfermedad TromboEmbolica venosa (RIETE) Registry. Circulation 2008; 117: 1711–1716.

Piazza

. Submassive pulmonary embolism. JAMA 2013; 309: 171–180.

Bagla

Smirniotopoulos

van Breda

. Ultrasound-accelerated catheter-directed thrombolysis for acute submassive pulmonary embolism. J Vasc Interv Radiol 2015; 26: 1001–1006.

McCabe

Huang

Riedl

. Usefulness and safety of ultrasound-assisted catheter-directed thrombolysis for submassive pulmonary emboli. Am J Cardiol 2015; 115: 821–824.

Engelberger

Kucher

. Ultrasound-assisted thrombolysis for acute pulmonary embolism: a systematic review. Eur Heart J 2014; 35: 758–764.

Kucher

Boekstegers

Muller

. Randomized, controlled trial of ultrasound-assisted catheter-directed thrombolysis for acute intermediate-risk pulmonary embolism. Circulation 2014; 129: 479–486.

Meinel

Nance

Jr Schoepf

. Predictive value of computed tomography in acute pulmonary embolism: systematic review and meta-analysis. Am J Med 2015; 128: 747–759.

10.

Zhang

Meinel

. Computed tomography of acute pulmonary embolism: state-of-the-art. Eur Radiol 2015; 25: 2547–2457.

11.

Furlan

Aghayev

Chang

. Short-term mortality in acute pulmonary embolism: clot burden and signs of right heart dysfunction at CT pulmonary angiography. Radiology 2012; 265: 283–293.

12.

Ferretti

Collomb

Ravey

. Severity assessment of acute pulmonary embolism: role of CT angiography. Semin Roentgenol 2005; 40: 25–32.

13.

Abrahams-van Doorn

Hartmann

. Cardiothoracic CT: one-stop-shop procedure? Impact on the management of acute pulmonary embolism. Insights Imaging 2011; 2: 705–715.

14.

Kipfmueller

Quiroz

Goldhaber

. Chest CT assessment following thrombolysis or surgical embolectomy for acute pulmonary embolism. Vasc Med 2005; 10: 85–89.

15.

Piazza

GHB

Jaff

Ouriel

. A Prospective, Single-Arm, Multicenter Trial of Ultrasound-Facilitated Catheter-Directed Low-Dose Fibrinolysis for Acute Massive and Submassive Pulmonary Embolism (SEATTLE II). JACC Cardiovasc Interv 2015; 24; 8: 1382–1392.

16.

Mélard

. On the accuracy of statistical procedures in Microsoft Excel 2010. Comput Stat 2014; 29: 1095–1128.

17.

Quiroz

Kucher

Schoepf

. Right ventricular enlargement on chest computed tomography: prognostic role in acute pulmonary embolism. Circulation 2004; 109: 2401–2404.

18.

Qanadli

El Hajjam

Vieillard-Baron

. New CT index to quantify arterial obstruction in pulmonary embolism: comparison with angiographic index and echocardiography. AJR Am J Roentgenol 2001; 176: 1415–1420.

19.

Cory

Valentine

. Varying patterns of the lobar branches of the pulmonary artery. A study of 524 lungs and lobes seen at operation of 426 patients. Thorax 1959; 14: 267–280.

20.

Van Der Spuy

. The surgical anatomy of the pulmonary vessels. Thorax 1953; 8: 189–194.

21.

Hallgren

. Computing inter-rater reliability for observational data: an overview and tutorial. Tutor Quant Methods Psychol 2012; 8: 23–34.

22.

Bland

Altman

. Measurement error. BMJ 1996; 313: 744–744.

23.

Jones

Dobson

O'Brian

. A graphical method for assessing agreement with the mean between multiple observers using continuous measures. Int J Epidemiol 2011; 40: 1308–1313.

24.

Kang

Ramos-Duran

Schoepf

. Reproducibility of CT signs of right ventricular dysfunction in acute pulmonary embolism. AJR Am J Roentgenol 2010; 194: 1500–1506.

25.

Furlan

Patil

Park

. Accuracy and reproducibility of blood clot burden quantification with pulmonary CT angiography. AJR Am J Roentgenol 2011; 196: 516–523.

26.

Demehri

Cai

. Axial and reformatted four-chamber right ventricle-to-left ventricle diameter ratios on pulmonary CT angiography as predictors of death after acute pulmonary embolism. AJR Am J Roentgenol 2012; 198: 1353–1360.