Utility of single-energy and dual-energy computed tomography in clot characterization: An in-vitro study

Abstract

Background and purpose

Because computed tomography (CT) is the most commonly used imaging modality for the evaluation of acute ischemic stroke patients, developing CT-based techniques for improving clot characterization could prove useful. The purpose of this in-vitro study was to determine which single-energy or dual-energy CT techniques provided optimum discrimination between red blood cell (RBC) and fibrin-rich clots.

Materials and methods

Seven clot types with varying fibrin and RBC densities were made (90% RBC, 99% RBC, 63% RBC, 36% RBC, 18% RBC and 0% RBC with high and low fibrin density) and their composition was verified histologically. Ten of each clot type were created and scanned with a second generation dual source scanner using three single (80 kV, 100 kV, 120 kV) and two dual-energy protocols (80/Sn 140 kV and 100/Sn 140 kV). A region of interest (ROI) was placed over each clot and mean attenuation was measured. Receiver operating characteristic curves were calculated at each energy level to determine the accuracy at differentiating RBC-rich clots from fibrin-rich clots.

Results

Clot attenuation increased with RBC content at all energy levels. Single-energy at 80 kV and 120 kV and dual-energy 80/Sn 140 kV protocols allowed for distinguishing between all clot types, with the exception of 36% RBC and 18% RBC. On receiver operating characteristic curve analysis, the 80/Sn 140 kV dual-energy protocol had the highest area under the curve for distinguishing between fibrin-rich and RBC-rich clots (area under the curve 0.99).

Conclusions

Dual-energy CT with 80/Sn 140 kV had the highest accuracy for differentiating RBC-rich and fibrin-rich in-vitro thrombi. Further studies are needed to study the utility of non-contrast dual-energy CT in thrombus characterization in acute ischemic stroke.

Introduction

Accurate characterization of clot composition among acute ischemic stroke patients is becoming increasingly important given the fact that a number of studies have demonstrated that thrombus composition can provide insights into stroke etiology, and even predict recanalization success following both intravenous thrombolysis and mechanical thrombectomy.^1–6 Such studies have suggested that clots that are red cell-rich clots are associated with improved recanalization rates with intravenous thrombolysis, and may be more amenable to revascularization with mechanical thrombectomy techniques. Some in-vitro studies suggest that clot composition can impact the ideal choice of techniques used during clot retriever and does, in fact, play a role in device–clot interaction.^7,8

Because computed tomography (CT) is the most commonly used imaging modality for the evaluation of acute ischemic stroke patients, developing CT-based techniques aimed at improving clot characterization could prove useful. The purpose of this in-vitro study was to determine if various single-energy or dual-energy CT techniques could discriminate between various types of clots, and to determine which energy levels provided the optimum discrimination between red blood cell (RBC)-rich and fibrin-rich clots.

Materials and methods

Clot preparation and nomenclature

The technique used to create and quantify the composition of the clots has been described previously.⁹ Blood was obtained from the jugular vein of sheep at a licensed facility (Ash Stream Ltd.). Seven different clot analog types were creased with 10 clot replicates formed for each type. For the purposes of this study, the clots are labeled according to their RBC density, as defined by the area of RBC on sectioned and hematoxylin and eosin stained samples of clots using histological techniques. For the 99% RBC clot, 62% RBC clot, 36% RBC clot, 18% RBC clot and low and high density 0% RBC clots, blood was anticoagulated in 3.2% sodium citrate solution (9:1 ratio) immediately after collection. Centrifugation was performed at 550 g for 15 minutes to separate whole blood constituents. Coagulation was initiated following the addition of 2.06% calcium chloride to the blood in a 1:9 ratio and clotted material matured for 30 minutes to 1 hour at 37℃.

The 90% RBC clot was formed in a static environment. Spontaneous coagulation was initiated by collected whole blood into a syringe which was left stationary for 12 hours at room temperature. The 99% RBC clot was formed by mixing RBCs and citrated plasma in a 1:4 ratio and then coagulated. The 62% RBC clot and 36% RBC clot were formed by mixing citrated plasma with RBCs in a ratio of 6:4 and 19:1, respectively, and then coagulated. The 18% RBC clot was prepared using a modification of the Chandler loop technique, which has been described previously. The low density 0% RBC clot (LD 0% RBC) and high density 0% RBC clot (HD 0% RBC) were prepared by coagulating citrated plasma without RBCs and buffy coat. Histopathological examples of each clot type are shown in Figure 1. All clots were stored in plastic vials in either a saline or formalin solution.

Figure 1.

Histological examples of each clot type. Hematoxylin and eosin stains of the seven types of clots that were imaged in this study. RBC: red blood cells.

Clot imaging and analysis

The clot-containing vials were arranged in the center of a water tank with lateral and anterior-to-posterior dimensions of 20 and 15 cm, respectively, mimicking the attenuation of an adult head. The samples were then scanned with a second generation dual source scanner (Siemens Definition Flash) using three single (80 kV, 100 kV, 120 kV) and two dual-energy protocols (80/Sn 140 kV and 100/Sn 140 kV, with an additional tin filter to the 140 kV beam). Each single-energy and dual-energy protocol was performed using the same dose, matched to that of a non-contrast head routine CT examination. The acquisition parameters for each protocol can be seen in Table 1. The single-energy images were reconstructed using a medium sharp kernel (J40) with iterative reconstruction (SAFIRE) at a slice thickness and increment of 1 × 0.75 mm. The dual-energy images were reconstructed using a medium smooth quantitative kernel (D30) at a slice thickness and increment of 1 × 1 mm. In dual energy, images from the low energy beam and high energy beam were blended together to generate a mixed image which uses all X-ray photons. In this study, a blending ratio of 0.4 (i.e. 40% of low kV images and 60% of high kV images) was used.

Table 1.

Imaging specifications.

Scan type	kV	Eff. mAs	Detector collimation (mm)	CTDIvol (mGy)
Single energy	80	800	128 × 0.6	39.6
Single energy	100	400	128 × 0.6	38.49
Single energy	120	250	128 × 0.6	38.25
Dual energy	80/140 Sn	500/250	32.0 × 0.6	42.49
Dual energy	100/140 Sn	250/250	32.0 × 0.6	42.76

A single neuroradiologist measured Hounsfield units (HUs) for each of the 70 clots that were imaged at the five energy levels. In total, 350 measurements were performed. A ROI of 0.05–0.15 mm² was placed over each clot and the mean and standard deviation HUs were measured. The ROI was placed to include the clot only while excluding background so the measured CT number was not averaged with the background. A representative image of our imaging set-up is provided in Figure 2.

Figure 2.

Representative image of imaging set-up. Cross-sectional computed tomography images of the 90% red blood cell clot imaged at 140/80 kV (a) and 140/100 kV (b). Images show the regions of interest drawn over each clot with the associated mean and standard deviation attenuation measurements.

Statistical analysis

For each scanning energy, the mean and standard deviation attenuation for each clot type was calculated. At each energy level, clot attenuations were compared using a Student’s t-test. For these comparisons, clots were compared to the two clots with the closest mean attenuation values to see if the attenuation characteristics differed (i.e. 90% RBC compared to 99% RBC, 99% RBC compared to 62% RBC, 62% RBC compared to 36% RBC, etc.). Receiver operating characteristic (ROC) curves were constructed and area under the curve (AUC) at each energy level to determine the accuracy of each energy level in differentiating RBC-rich clots from fibrin-rich clots. For the purpose of this analysis, RBC-rich clots were those with an RBC content greater than 50%, while fibrin-rich clots were those with an RBC content of less than 50%. All statistical analysis was performed using the SAS-based statistical software package JMP12.0 (www.jmp.com).

Results

Clot attenuation by energy level

A summary of clot attenuations at each energy level is provided in Figure 3, and P values for discriminating each clot type from the clot with a closest mean attenuation are provided in Table 2. At 80 kV, clot attenuation ranged from 15.8 ± 8.2 HUs for the LD 0% RBC clots to 61.2 ± 3.9 HUs for the 90% RBC clots. At 80 kV, each of the clots could be distinguished based on mean attenuation, with the exception of the 99% RBC and 62% RBC (P = 0.23). At 100 kV, clot attenuation ranged from 12.9 ± 6.3 HUs for the LD 0% RBC clots to 63.5 ± 6.1 HUs for the 90% RBC clots. At 100 kV, three clot types could not be distinguished from one another including the 99% RBC from the 63% RBC (P = 0.48), the 36% RBC from the 18% RBC (P = 0.17) and the LD 0% RBC from the HD 0% RBC (P = 0.06). At 120 kV, clot attenuation ranged from 10.3 ± 6.8 HUs for the LD 0% RBC clots to 61.0 ± 6.5 HUs for the 90% RBC clots. At 120 kV, all clot types could be distinguished from one another, with the exception of the 36% RBC clots and 18% RB clots (P = 0.16).

Figure 3.

Graphical representation of attenuation values at different energy levels. (a) Mean attenuations at 80 kV; (b) mean attenuations at 100 kV; (c) mean attenuations at 120 kV; (d) mean attenuations at 100/140 kV; (e) mean attenuations at 80/140 kV.

Table 2.

The P values for clot attenuation comparisons.

	Energy levels
	80 kV	100 kV	120 kV	100/140	80/140
Clots compared
90% vs. 99%	0.003	0.001	0.0007	<0.0001	0.002
99% vs. 62%	0.23	0.48	0.002	0.88	0.03
62% vs. 36%	0.001	<0.0001	0.01	<0.0001	<0.0001
36% vs. 18%	0.04	0.17	0.16	0.08	0.26
18% vs. HD 0%	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
HD 0% vs. LD 0%	0.004	0.06	0.03	0.01	0.04

For the dual-energy protocols, at 100/Sn 140 kV, clot attenuation ranged from 15.0 ± 7.8 HUs for the LD 0% RBC clots to 64.8 ± 7.8 HUa for the 90% RBC clots. At 100/Sn 140 kV, all clots could be distinguished from one another, with the exception of the 90% RBC and 63% RBC clots (P = 0.88) and the 36% RBC and 18% RBC clots (P = 0.08). At 80/Sn 140 kV, clot attenuation ranged from 18.0 ± 7.5 HUs for the LD 0% RBC clots to 62.4 ± 5.8 HUs for the 90% RBC clots. At 80/Sn 140 kV, all clots could be distinguished from one another, with the exception of the 36% RBC and 18% RBC clots (P = 0.26).

Discrimination between RBC-rich and fibrin-rich clots

ROC curves were made to determine the performance of each CT energy level in distinguishing between RBC-rich and fibrin-rich clots. The AUC for 80 kV was 0.935. The AUC for 100 kV was 0.951. The AUC for 120 kV was 0.905. The AUC for 100/140 kV was 0.97 and the AUC for 80/140 kV was 0.99.

Discussion

Our in-vitro study measuring clot attenuation at various single-energy and dual-energy CT energy levels found that, in general, CT is an excellent imaging modality for distinguishing between clots which are fibrin-rich and clots which are RBC-rich. Interestingly, we found that using a dual-energy technique in which energies of 80/Sn 140 kV which are blended at a ratio of 0.4 allowed for the best discrimination between fibrin-rich and RBC-rich clots as the AUC was 0.99 and nearly all clot types could be distinguished from one another. These findings are important as they suggest that dual-energy non-contrast CT may allow for improved characterization of clot composition, a factor which could have potential implications for stroke intervention.

A number of studies have examined the association between clot histology and imaging characteristics on CT and magnetic resonance imaging for patients with acute ischemic stroke.^1,5,10–15 Studies examining the RBC content of hyperdense versus non-hyperdense clots have consistently demonstrated a positive correlation between RBC content and attenuation.^1,5,10–15 In one recently published systematic review and meta-analysis of clot characterization, we found that hyperdense thrombi had a mean RBC % of 45.2% compared to 23.3% for non-RBC-rich clots.¹⁶ It is important to note that the definition of an RBC clot varied from study to study as do the various CT protocols that were used. Of the studies in the literature which have performed histological and CT imaging correlation of clots in stroke, none have specifically tested the ability of various CT protocols and imaging levels at differentiating RBC-rich from fibrin-rich clots.

Accurate characterization of clot characteristics on imaging is potentially important for a number of reasons. A number of studies have suggested that the histological and physical characteristics of the clot can influence recanalization rates and device–clot interactions.^3,6–8 In general, clots which are rich in RBCs are softer than those which are rich in organized fibrin, calcium or cholesterol, which can influence device–clot interactions. For example, in one recently published study, van der Marel et al. found that softer clots are equally well engaged using both unsheathing and pushing techniques when deploying a stent retriever, while stiffer clots are better engaged using a pushing technique.⁷ In general, studies have found that hyperdense and RBC-rich thrombi are associated with improved recanalization outcomes with endovascular intervention as well, which is probably a function of improved device–clot interaction.¹⁶ Clot characterization is also potentially important for intravenous thrombolysis as studies have found that clots which are RBC-rich or hyperdense are more responsive to tissue plasminogen activator.¹⁷

Other studies have used dual-energy CT to characterize thrombus characteristics for non-stroke pathologies. In general, these studies have found that dual-energy CT can be useful in characterizing bland thrombus from tumor thrombus (i.e. portal venous thrombus in hepatocellular carcinoma versus bland thrombus, left atrial myxoma from left atrial thrombus, etc.).^18–21 One key feature of these studies is that the benefit of dual-energy CT is primarily due to the fact that bland thrombus generally exhibits no or minimal contrast enhancement, thus allowing it to be distinguished from tumor thrombus, which is generally vascularized. Thus, the k-edge of iodine can be exploited to allow for improved characterization. Our study differs from these studies in that we are attempting to differentiate different types of bland thrombus based on non-contrast imaging alone.

Limitations and future directions

Our study has limitations. Because our study is an in-vitro study, our findings may not directly translate into the clinical arena. For this reason, further studies examining the use of dual-energy protocols in thrombus characterization in acute ischemic stroke patients undergoing mechanical thrombectomy would be needed to determine if dual-energy CT is indeed useful in the clinical setting. The clots that were studied were in either a formalin or normal saline solution. It is likely that placing the clots in a different type of solution (i.e. blood, contrast, etc.) could alter the attenuation characteristics. All of the clots were housed in plastic vials and there was no bone or other highly dense structure surrounding the samples. It is possible that the addition of dense bone, such as that seen in the skull, could alter the attenuation characteristics of the clots as well as the ability of the various energy protocols to distinguish between them.

Conclusions

Our in-vitro study suggests that CT is an effective imaging modality at distinguishing between RBC-rich and fibrin-rich thrombi. In general, clot attenuation correlated with the proportion of RBCs in a clot at all tested energies. Dual-energy CT with 80/Sn 140 kV had the highest accuracy for differentiating between RBC-rich and RBC-poor thrombi. These findings suggest that dual-energy CT at 80/Sn 140 kV could be useful for clot characterization in acute ischemic stroke. Further clinical studies are needed to confirm these findings.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.