Dual-Energy Computed Tomography Imaging of Head: Virtual High-Energy Monochromatic (190 keV) Images Are More Reliable Than Standard 120 kV Images for Detecting Traumatic Intracranial Hemorrhages

Abstract

High-energy monochromatic (190 keV) images may be more reliable than standard 120 kV Images for detecting intracranial hemorrhages. We aimed to retrospectively compare virtual high monochromatic (190 keV) and standard 120 kV images from dual-energy computed tomography (CT; DECT) for the diagnosis of intracranial hemorrhages in traumatic brain injury (TBI). We analyzed admission CT studies in 100 trauma patients. Three radiologists independently reviewed four image sets: 120 kV and 190 keV (thin [1 mm] and thick [5 mm] section) images for the presence of various types of intracranial hemorrhages. The proportions of positive variables were compared and differences calculated by McNemar test and sensitivities determined by contingency tables. Randomly selected hemorrhagic lesions were analyzed for contrast index (CI). Thin-section 190 keV images were superior in the detection of subdural hematomas (SDH) (p < 0.0001), supratentorial contusions (p < 0.0001), and epidural hematomas (EDH) (p = 0.014), when compared with standard 120 kV images. However, 190 keV images were inferior to standard 120 kV images in diagnosis of subarachnoid hemorrhage (SAH) (thin-sections, p = 0.059; thick-sections, 0.0075). The 190 keV images yielded moderate increase in CI of contusions (Cohen's d > 0.53) and a large increase in CI of extra-axial hematomas (Cohen's d > 0.86). Our results indicate that virtual high monochromatic (190 keV, thin-section) images combined with standard 120 kV images may provide optimal diagnostic performance for evaluation of patients suspected of TBI.

Introduction

Approximately 6 million head and neck trauma patients are seen annually in the emergency departments of North America.¹ Prompt and reliable detection of intracranial hemorrhage (ICH) is important for treatment of traumatic brain injury (TBI) patients ( ∼1.7 million).² Computed tomography (CT) is the initial diagnostic method used to assess patients with TBI.³

Theoretically, virtual high monochromatic images from dual-energy computed tomography (DECT) data decrease beam hardening artifacts from cranial bones.^4
–6 In addition, the new monochromatic image algorithm (Monoenergetic Plus, syngo.via, version VB10B; Siemens Healthcare, Forchheim, Germany) significantly decreases image noise.⁷ The combination of these two factors allows thin-section image reconstruction, which is otherwise not possible with mixed 120 kV images due to the limitation of high image noise. Decreased beam hardening artifacts could also positively affect the display of bone–brain interface helping in detection of subdural hematomas (SDH) and epidural hematomas (EDH). As the monochromatic energy level is increased, the spectral attenuation curves demonstrate a modest decrease in attenuation of brain hemorrhages and parenchyma as Compton effect prevails in materials with low atomic numbers, whereas at the same time, the bones, due to high atomic numbers demonstrate a dramatic decrease in attenuation because of the photoelectric effect.^8,9 The suppression of background attenuation of brain parenchyma at high monochromatic energy (190 keV) can potentially increase the contrast index (CI) of hematomas and hence the perception of the contrast differences at the blood–brain interface. These factors could be used for task-specific imaging of intracranial hemorrhages.¹⁰ We have therefore included automatically generated high monochromatic (190 keV) images along with 120 kV images in our routine head CT protocol at R. Adams Cowley Shock Trauma Center.

Standard 120 kV images (5 mm slice thickness; 5-mm intervals) derived from dual-energy data mirror typical single-energy CT (SECT) images used in clinical practice. We have frequently observed intracranial hemorrhages on 190 keV images that were not detected on 120 kV images. This observation needs to be substantiated for the full range of intracranial hemorrhages in patients who present for emergency assessment of blunt TBI. We aimed to retrospectively compare high monochromatic (190 keV) and standard 120 kV images from DECT for the diagnosis of intracranial hemorrhages in TBI and to determine the effect of high keV on CI of hemorrhagic lesions.

Methods

This retrospective study was Health Insurance Portability and Accountability Act (HIPAA) compliant, and permission was obtained from our institutional review board. Informed consent was waived.

Subjects

The inclusion criteria were (a) history of blunt trauma with acquisition of admission head DECT between May 15, 2016 and September 10, 2016; (b) acquisition of follow-up head CT within 24 h of admission CT; and (c) age ≥18 years. At our institution, patients with traumatic lesions on initial CT, patients with persistent altered mental status without traumatic lesions, patients for whom CT fails to explain neurological status, and patients on anti-coagulation without traumatic lesions tend to be evaluated by follow-up CT. We selected patients with follow-up head CTs to capture a majority of TBI patients with traumatic lesions.

A search of radiology information system (RIS) yielded 113 patients with an initial DECT and with at least one follow-up head CT performed within 24 h of the initial study. The first 100 patients were selected for this study. Baseline clinical and radiological characteristics of the cohort are shown in Table 1.

Table 1.

Baseline Clinical and Radiological Characteristics

Characteristic	Total (n = 100) N (SD)
Mean age	40.2 (22.3)
Sex (male)	63
Mechanism of injury
- Motor Vehicle collisions	44
- Falls	25
- Struck pedestrians	7
- Assaults	6
- Other vehicular accidents (motorcycle crash, bicycles, all-terrain vehicles, etc.)	18
Mean admission GCS score	12.7 (3.1)
- Intubated	5.97T (2)
Pupil reactivity
- Both eyes	82
- One eye	13
- None	5
Hypoxia (PaO₂ <8kPa (60 mm Hg)/SaO₂ <90%)	7
Hypotension (Systolic BP <90 mm Hg)	19
Mean admission glucose (mg/dL)	143.1 (52.8)
Mean admission hemoglobin (g/dL)	12.96 (2.1)
Mean PT (sec)	15.4 (3.7)
Mean aPTT (sec)	28.3 (4.7)
Mean fibrinogen (mg/dL)	295.2 (97.3)
Mean platelets (10³/μL)	215.7 (77)
Marshall CT classification of TBI
- DI I	22
- DI II	51
- DI III	9
- DI IV	5
- NEML VI	13

aPTT, activated partial thromboplastin time; CT, computed tomography; DI, diffuse injury; GCS, Glasgow Coma Scale; NEML, non-evacuated mass lesion; PT, prothrombin time; SD, standard deviation; TBI, traumatic brain injury.

Imaging technique

Head CT examinations were performed with DECT (SOMATOM FORCE, Siemens Healthcare, Forchheim, Germany). Images were obtained with the X-ray tubes at 80 kV and Sn150 kV. Scan parameters were as follows: rotation time 0.5 sec; pitch 0.55. The reference mAs was 273 for the Sn150 kV and 410 for the 80 kV tube. Original DE data sets were reconstructed with an increment of 1 mm and slice thickness of 1 mm, using an adaptive iterative reconstruction algorithm (ADMIRE, Siemens Healthcare, Forchheim, Germany) at a strength level of 2. Automatic reconstruction of high monochromatic (190 keV) images at 1-mm slice thickness, 1-mm intervals and 120 kV equivalent mixed DECT images at 5-mm slice thickness, 5-mm intervals was performed and sent to picture archiving and communication system (PACS) (AGFA IMPAX; AGFA Healthcare, Greenville, SC). Automatic tube current modulation (CARE Dose 4D, Siemens Healthcare, Forchheim, Germany) was used in all patients. Mean CT dose index (CTDI) vol and dose-length product (DLP) were 31.45 SD 2.95 mGy and 609.62 SD 76.79 mGy cm, respectively.

Image analysis of DECT

Dual-energy data sets from admission head CTs were transferred to a post-processing workstation (syngo.via). Additional sets of 190 keV images at 5-mm slice thickness, 5-mm intervals and 120 kV images at 1-mm slice thickness, 1-mm intervals were retrospectively derived and sent to PACS for the study purpose. Hence, a total of four image sets were used for analysis in each patient, that is, 120 kV (thick-section), 120 kV (thin-section), 190 keV (thick-section), and 190 keV (thin-section).

Two radiologists (Reviewer 1 [R1]: 6 years of experience as attending; Reviewer 2 [R2]: 8 years of experience, performed independent reviews of each study set on the institution's PACS. Thus, there were four reading sessions and no observer saw different image sets in a particular patient within the same reading session. To avoid memory effects, it was ensured that at least 4 weeks elapsed between each reading session.

All the variables were given nominal scores based on presence (score 1) versus absence (score 0) of each of the variables. The following pathoanatomical terms for definitions of TBI lesions as recommended by the National Institute of Neurological Disorders and Stroke Common Data Elements (http://www.commondataelements.ninds.nih.gov/) were recorded: hemorrhage present or absent; if present, type(s) of hemorrhage: subarachnoid hemorrhage (SAH), subdural hematoma (SDH), epidural hematoma (EDH), intra-ventricular hemorrhage (IVH), and hemorrhagic contusions; location (frontal, parietal, temporal, or occipital lobes [800 lobes]); and side (right, left). Contusions were further discriminated based on the involvement of supratentorial lobes, cerebellum, corpus callosum, or brainstem. Separate lesions were listed as separate entries. For example a holohemispheric SDH was entered four times based on its distribution over all the four lobes.¹¹ Subsequently, reviewer 3 with 11 years of experience [R3], blinded to the discrepant variables, reviewed all the studies containing at least one variable with discrepant interpretation between R1 and R2 and also analyzed all the discordant lesions between 120 kV (thick-section) and 190 keV (thin-section) images, by measuring the largest diameter of contusions, maximum width of the SDHs and EDHs. Randomly selected were 10 contusions and 10 extra-axial hematomas (SDHs or EDHs) from different patients, which were used to determine the effect of 190 keV on the contrast of the hematomas by measuring the CI.

To determine the CI, individual lesions were identified and marked on a post-processing workstation. A region of interest (ROI) with a constant area of 2 mm² was placed over each hematoma, on virtual monochromatic series. The software calculates the attenuation (HU) values at 190 keV and 120 kV. CI was calculated by the formula: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{ \rm{CI}} = \left( {{ \rm{H}}{{ \rm{U}}_{{ \rm{HEM}}}} - { \rm{H}}{{ \rm{U}}_{{ \rm{BG}}}}} \right) / { \rm{H}}{{ \rm{U}}_{{ \rm{BG , }}}} \end{align*} \end{document}

where HU_HEM is the mean HU of the hematoma in the ROI and HU_BG is the mean HU of the background. To promote uniformity in the measurements of background attenuation, the ROI was placed over adjacent normal parenchyma at equal distance from bone as that of hematoma.

Reference standard

Reference standard was determined independently for each of the four sets of images (120 kV, and 190 keV; thin 1-mm and thick 5-mm sections). For a case to be judged as the final imaging diagnosis, either unanimous agreement between R1 and R2 regarding the presence of a variable on an individual CT scan or concordance of at least two of the three reviewers served as the reference standard. The third reviewer's (R3's) interpretation was used for adjudicating the discrepancies between R1 and R2. For the final diagnosis, presence of a specific type of hemorrhage in any of the four sets was considered as reference standard and was used for calculating the sensitivities of each set of images.

We did not use the follow-up CT scans to confirm the presence of specific type of hemorrhage as TBI lesions tend to evolve and some of the subtle lesions can disappear by the time a follow-up CT is obtained.

Statistical analysis

Statistical analysis was performed using statistical software (JMP 12 software, SAS Institute). The proportions of patients with intracranial hemorrhages on 120 kV (thick-section) images were compared with the remaining three image sets, as the primary objective of the study. The secondary objective was to determine the effect of 190 keV on CI of hemorrhagic lesions.

McNemar's paired proportion test was used to measure the concordance between 120 kV and 190 keV images for each variable. Contingency tables were used for calculating the sensitivities. The significance level was set at p < 0.05. Inter-observer agreement between the two reviewers (R1 and R2) was determined through calculation of Cohen kappa (κ) and 95% confidence intervals (95% CI) were provided. Contrast Index (CI) of the hematomas were compared by using paired t test and effect size was determined by Cohen's d.

Results

The comparison between standard 120 kV (clinical standard) image performance against the remaining three sets of image reconstructions for select CT variables is shown in Supplementary Table 1 (per lobe) and Supplementary Table 2 (per patient) (see online supplementary material at http://www.liebertpub.com). The sensitivities of each set of images to that of the reference standard are shown in Table 2 (per lobe and patient). The specificities are not provided as the results would be perfect due the methodology of reference standard derivation, that is, hemorrhage is considered to be absent, only if all the image sets were negative for specific type of intracranial hemorrhage.

Table 2.

Sensitivities from Each of the Image Sets for Different Types of Intracranial Hemorrhages Based on per Lobe and per Patient

Per Lobe (n = 800)		120 kV (5x5)	190 keV (1x1)	190 keV (5x5)	120 kV (1x1)
SDH	Sensitivity (95% CI)	85.6 (79.7, 90.4)	100 (98, 100)	92.3 (87.4, 95.7)	89.5 (84.1, 93.6)
SAH	Sensitivity (95% CI)	93.8 (88.5, 97.1)	88.3 (81.9, 93)	86.9 (80.3, 91.9)	89.7 (83.5, 94.1)
Supratentorial contusions	Sensitivity (95% CI)	77.9 (68.7, 85.4)	100 (96.5, 100)	80.8 (71.9, 87.8)	88 (79.6, 93.9)
EDH	Sensitivity (95% CI)	40 (12.2, 73.8)	100 (69.2, 100)	40 (12.2, 73.8)	70 (34.8, 93.3)
Per Patient (n = 100)
SDH	Sensitivity (95% CI)	85.4 (73.3, 93.5)	100 (93.5, 100)	89.1 (77.7, 95.9)	89.1 (77.7, 95.9)
SAH	Sensitivity (95% CI)	92 (80.8, 97.8)	92 (80.8, 97.8)	92 (80.8, 97.8)	96 (86.3, 99.5)
Supratentorial contusions	Sensitivity (95% CI)	92 (80.8, 97.8)	100 (92.9, 100)	92 (80.8, 97.8)	98 (89.4, 99.9)
EDH	Sensitivity (95% CI)	50 (15.7, 84.3)	100 (63, 100)	50 (15.7, 84.3)	75 (34.9, 96.8)
ICH	Sensitivity (95% CI)	95 (87.7, 98.6)	100 (95.5, 100)	92.5 (84.4, 97.2)	97.5 (91.3, 99.7)

1x1 mm denotes thin-sections and 5x5 mm, thick-sections.

CI, confidence interval; EDH, epidural hematoma; ICH, intracranial hemorrhage; SAH, subarachnoid hemorrhage; SDH, subdural hematoma.

Comparison of 190 keV and standard 120 kV images

Subdural hematoma

SDHs were detected in 181 of 800 lobes (23%) and 55 of 100 patients (55%) on thin-section 190 keV images. Standard 120 kV images detected SDHs only in 155 of 800 lobes (19%) and 47 of 100 patients (47%). Detection of SDHs was more frequent on thin-section 190 keV images (p < 0.0001 for lobes and p = 0.0047 for patients) (Fig. 1, Supplementary Tables 1 and 2). The mean width of SDHs in 8 patients with hematoma visualized only in 190 keV images was 4.1 mm (95% CI, 3.3–4.9 mm). Detection of SDHs was also more frequent with thick-section 190 keV images (p = 0.0005) for lobes (Supplementary Table 1).

FIG 1.

A 65-year-old male developed left frontal cortical contusion, and right frontal and parafalcine subdural hematomas after a fall. (A) Axial 120 kV (thick-section) image and (B) axial 190 keV (thick-section) image show parafalcine subdural hematoma (curved arrow), but reviewers failed to identify the right frontal subdural hematoma (arrow) and left frontal cortical contusion (arrowhead). (C) Axial 120 kV (thin-section) and (D) 190 keV (thin-section) images, where reviewers identified all the hemorrhages that include parafalcine subdural hematoma (curved arrow), right frontal subdural hematoma (arrow), and left frontal cortical contusion (arrowhead). (E) Three-dimensional (3D) surface plots of the attenuation values (HU) for 120 kV (thick-section) and (F) 190 keV (thick-section) images fail to show the attenuation peak from right frontal subdural hematoma (arrow). (G) 3D surface plot of attenuation values for 120 kV (thin-section) and (H) 190 keV (thin-section) images shows the attenuation peak from right frontal subdural hematoma (arrow), parafalcine subdural hematoma (curved arrow), and left frontal cortical contusion (arrowhead). Also note that 3D plots from 190 keV images showing relatively thinner region of calvarial attenuation, homogenously suppressed attenuation of brain parenchyma with flat peaks and troughs (main contributing factor for increase in contrast index of subdural hematomas and contusions), whereas 120 kV images show relatively thick region of attenuation caused by calvarium, wide ranges of brain tissue attenuation in the form of peaks and troughs. The attenuation plots were created using Fiji Image J software.²⁰

Supratentorial hemorrhagic contusions

Contusions were detected in 104 of 800 lobes (13%) and 50 of 100 patients (50%) on thin-section 190 keV images. On standard 120 kV images, contusions were detected in 81 of 800 lobes (10%) and 46 of 100 patients (46%). The detection of contusions was more frequent on thin-section 190 keV images (p < 0.0001 for lobes and p = 0.045 for patients) (Figs.1 and 2) (Supplementary Tables 1 and 2). In the 4 patients in whom contusions were visualized only on 190 keV, the lesions were mainly counter coup contusions that were linear or ovoid in shape, distributed at the inferior aspect of frontal and temporal lobes parallel to crista galli or petrous temporal bones. The mean diameter of the hematoma was 7.9 mm along the long axis (95% CI, 4.8–11 mm). Thick-section 190 keV images showed no superiority for contusions.

FIG 2.

A 56-year-old male developed cerebellar contusion after a presumed assault. (A) Standard 120 kV (thick-section) and (B) 120 kV (thin-section) images shows streak artifacts from dental amalgam limiting the visualization of hemorrhagic contusion (arrowhead). (C) Axial 190 keV (thin-section) images significantly reduce the streak artifacts, allowing visualization of contusion (arrowhead).

Epidural hematoma

EDHs were detected in 10 of 800 lobes (1.3%) and 8 of 100 patients (8%) on thin-section 190 keV images. EDHs were detected in only 4 of 800 lobes (0.5%) and 4 of 100 patients (4%) on standard 120 kV images. Detection of EDHs was more frequent on thin-section190 keV images (p = 0.014 for lobes and p = 0.045 for patients) (Supplementary Tables 1 and 2). The mean width of EDHs in 4 patients with hematoma visualized only in 190 keV images was 4 mm (95% CI, 2.7–5.3 mm). Thick-section 190 keV images showed no superiority for EDHs.

Subarachnoid hemorrhage

SAHs were detected in 128 of 800 lobes (16%) and 46 of 100 patients (46%) on thin-section 190 keV images. On 120 kV images, SAHs were detected in 136 of 800 lobes (17%) and 46 of 100 patients (46%). When compared with the thin-section 190 keV images, detection of SAHs was more frequent with 120 kV images (p = 0.059, for lobes), and there was no superiority (p = 1) per patient basis (Fig. 3) (Supplementary Tables 1 and 2). Superiority of 120 kV images is also demonstrated, when compared with thick-section 190 keV images (p = 0.0075 for lobes) (Supplementary Table 1).

FIG 3.

A 69-year-old female developed traumatic subarachnoid hemorrhage with blood in the right sylvian fissure after motor vehicle collision. (A) Standard 120 kV (thick-section) image shows blood extending along the sylvian fissure (curved arrow). (B) Axial 190 keV (thin-section) image fails to demonstrate blood in this location (curved arrow). (C) Axial 120 kV (thin-section) image fails to demonstrate blood in this location (curved arrow). Although the diagnosis was made due to the presence of blood in multiple locations, this example demonstrates suppressed density of subarachnoid blood on 190 keV images.

Effect of 120 kV image section thickness in detection of post-traumatic hemorrhages

There was no statistical superiority of thin-section 120 kV images over 120 kV thick-section images, although the thin-section images detected a greater number of SDHs (p = 0.09, 162 vs. 155 lobes). There was also no statistical difference in the detection rates of SAHs (p = 0.13, 130 vs. 136 lobes). However, thin-section 120 kV images were superior to thick-sections 120 kV images in detecting supratentorial hemorrhagic contusions (p = 0.0005, 93 vs. 81). Complete details are shown in Supplementary Table 1. The results from a separate analysis comparing thin-section 120 kV and 190 keV images are shown in a 2 × 2 table (Table 3). The results failed to show a superiority of one set of images over the other set (p = 0.7) in SAH detection. However, thin-section 190 keV images were superior to thin-section 120 kV images in detection of supratentorial hemorrhagic contusions (p = 0.0009).

Table 3.

Results of McNemar Test for Paired Proportions, Lobe-by-Lobe Basis

		190 keV (1x1 mm)
Test type		Test positive	Test negative	Total
SDH 120 (1x1) kV	Test positive	160	2	162
	Test negative	21	617	638
	Total	181	619	p = 0.00007
SAH 120 (1x1) kV	Test positive	115	15	130
	Test negative	13	657	670
	Total	128	672	p = 0.7
Supratentorial contusions 120 (1x1) kV	Test positive	93	0	93
	Test negative	11	696	707
	Total	104	696	p = 0.0009
EDH 120 (1x1) kV	Test positive	7	0	7
	Test negative	3	790	793
	Total	10	790	p = 0.083

Boldface for p values indicates statistical significance (p < 0.05, McNemar test).

Each variable formatted as 2x2 table; 1x1 mm denotes thin-sections.

EDH, epidural hematoma; SAH, subarachnoid hemorrhage; SDH, subdural hematoma.

There was a substantial agreement among the reviewers in the interpretation of SDH (κ = 0.73; 95% CI, 0.6–0.85), contusions (κ = 0.78; 95% CI, 0.66–0.89), and EDH (κ = 0.75; 95% CI, 0.5–0.98), whereas there was a moderate agreement for SAH (κ = 0.58; 95% CI, 0.42–0.73). R3 evaluated 204 image sets out of 400 that had a minimum of one variable with discrepant interpretation between R1 and R2.

Comparison between standard 120 kV image performance and thin-section 190 keV images for the CT variables that are less frequent in occurrences and those that did not show statistical significance are shown in Supplementary Table 3 (see online supplementary material at http://www.liebertpub.com).

Effect of 190 keV on CI

The mean CI of hemorrhagic contusions on 120 kV images was 0.7 (SD, 0.26), and 0.89 (SD, 0.44) on 190 keV, corresponding to a moderate increase in CI (Cohen's d > 0.53) at blood–brain interface. For extra-axial hemorrhages the mean was 0.58 (SD, 0.24), on 120 kV and 0.945 (SD, 0.54) on 190 keV, corresponding to a large increase in CI (Cohen's d > 0.86) at blood–brain interface.

Discussion

The major findings from our study are the following: (1) SDHs, supratentorial contusions, and EDHs were more effectively detected by thin-section 190 keV images than by standard 120 kV images, and (2) 190 keV images were inferior to standard 120 kV images for the detection of SAH.

TBI patients with positive head CT for intracranial injury are managed differently from those without any signs of injury.¹² However, one third of patients with severe TBI without traumatic lesions on initial CT may progress to show new lesions on follow-up CT.¹³ This is a medically relevant group of patients, because up to 75% of these patients have raised intracranial pressure (ICP).¹³ A subset of this group may have had subtle injuries beyond the sensitivity threshold of routine 120 kV images. Hence, the more sensitive 190 keV images might detect some of these injuries on the initial CT and provide an opportunity for earlier ICP monitoring.¹⁴ TBI is a dynamic process with evolving pathological changes of the existing structural lesions.¹⁵ Approximately 15% of diffuse injuries manifest with new lesions on follow-up, and contusions, SDHs, and EDHs tend to enlarge.^3,16
–18 The increased sensitivity of thin-section 190 keV images not only has the potential to identify additional traumatic lesions on the admission CT, allowing serial follow-up imaging, but also to help anticipate, prevent, and reverse detrimental physiological changes that can exacerbate secondary brain injury.¹⁹

The greater overall sensitivity of thin-section 190 keV images for SDHs, contusions, and EDHs (Table 2) can be attributed to the reduced beam-hardening artifacts, increased CI of hematoma at blood–brain interface, improved spatial resolution, and decreased partial volume averaging rendered by thin-section images. Although there was an increased sensitivity of thin-section 190 keV for SDHs and EDHs, thick-section 190 keV images were superior only for SDHs, despite their similar extra-axial hematoma location (Table 2). This difference may be attributable solely to the improved special resolution rendered by thin-sections in detecting small EDHs.

Thin-section 120 kV images were shown to have an increased sensitivity in detecting supratentorial hemorrhagic contusions when compared with thick-section images that can be explained by the improved visibility due to improved spatial resolution and decreased partial volume averaging. However, the main problem with thin-section reconstructions is the image noise that can impair the depiction of traumatic lesions. For contusions, the balance between image noise and improved spatial resolution may have favored the latter. However, for SAH and SDHs that did not show improved sensitivity with thin-section 120 kV, the improved spatial resolution on thin-section reconstruction images might not have been enough to compensate for the deterioration of image quality caused by the noise. Increased noise levels are expected to result in areas of spuriously increased attenuation; hence, one could anticipate SDHs or SAHs not showing sufficient contrast with surrounding tissue to be detected at a higher rate on thin-section images. As a result, image thickness should be carefully chosen to account for a balance between an improvement in spatial resolution and deterioration of quality due to image noise. Image noise is the main limiting factor for a lack of acceptance of thin-section 120 kV images in clinical practice.

Finally, 190 keV images were inferior to standard 120 kV images for SAHs. There is no clear explanation for the phenomenon of blood in cerebrospinal fluid (CSF) space behaving differently from other intracranial hemorrhages. The limitation might originate from the decrease in the attenuation difference between the small amount of subarachnoid blood in the CSF space and the adjacent parenchyma caused by high keV. Another reason may be the methods used to filter the X-ray beams for generation of the virtual monochromatic images and the algorithms used that might be more susceptible to artifacts caused by image noise compared with the methods that use the actual raw data for the generation of monochromatic images.

Limitations

This study has a retrospective single-center design, which introduces selection and institutional bias. The cohort comprises only those patients with follow-up CT studies; this was done to capture a majority of TBI patients with structural lesions. This methodology would be suitable for demonstrating the superiority of 190 keV images, as analyzed per lobe basis. However, the selection bias might have excluded TBI patients with subtle intracranial hemorrhages that radiologists fail to prospectively identify and report on initial CT (120 kV images). The results obtained from per patient analysis tend to underestimate the sensitivity differences between 190 keV and 120 kV image sets, because of such exclusion. As such, this data may not be generalizable to all trauma patients. The study evaluated only axial images; inclusion of coronal and sagittal reconstructions might impact the results as hemorrhages adjacent to the vertex and skull base are disproportionately effected by beam hardening artifacts on axial images. Finally, separate entries for presence of hemorrhage in patients with holohemispheric SDH or SDH that extends over more than one lobe might have inflated the number of detected or missed lesions in individual image sets.

Conclusion

High keV monochromatic images (thin-sections) may be able to detect intracranial hemorrhages not evident on conventional CT. Hence, it may be worthwhile to adopt high monochromatic sequence into the regular workflow for assessment of blunt TBI. One important caveat is that 120 kV images tend to perform better for SAH. Therefore the use of high keV monochromatic images combined with standard 120 kV images may provide optimal diagnostic performance for evaluation of patients suspected of TBI.

Footnotes

Acknowledgments

The authors thank Brigitte Pocta for editing the manuscript and Kuldeep Sudini for his assistance with the statistics.

Author Disclosure Statement

Activities related to the present article: no competing financial interests exist. Activities not related to the present article: U.K. Bodanapally and D. Dreizin have received a research grant from Siemens Healthineers. U.K. Bodanapally, K.L. Archer-Arroyo, and T.R. Fleiter have received payment and travel fees from Siemens Medical Solutions.

References

Stiell

I.G.

, Clement

C.M.

, Grimshaw

J.M.

, Brison

R.J.

, Rowe

B.H.

, Lee

J.S.

, Shah

, Brehaut

, Holroyd

B.R.

, Schull

M.J.

, and McKnight

R.D.

(2010). A prospective cluster-randomized trial to implement the Canadian CT Head Rule in emergency departments. CMAJ, 182, 1527–1532.

Laker

S.R.

(2011). Epidemiology of concussion and mild traumatic brain injury. PM, R3, S354–S358.

Bodanapally

U.K.

, Sours

, Zhuo

, and Shanmuganathan

(2015). Imaging of traumatic brain injury. Radiol. Clin. North Am. 53, 695–715.

, Leng

, and McCollough

C.H.

(2012). Dual-energy CT–based monochromatic imaging. AJR Am. J. Roentgenol. 199, S9–S15.

, Langan

D.A.

, Xu

, Benson

T.M.

, Pack

J.D.

, Schmitz

A.M.

, Tkaczyk

E.J.

, Leverentz

, and Licato

(2009). Monochromatic CT image representation via fast switching dual kVp. Proc. SPIE 7258, Medical Imaging 2009: Physics of Medical Imaging, 725845; doi:10.1117/12.811698

Matsumoto

, Jinzaki

, Tanami

, Ueno

, Yamada

, and Kuribayashi

(2011). Virtual monochromatic spectral imaging with fast kilovoltage switching: improved image quality as compared with that obtained with conventional 120-kVp CT. Radiology, 259, 257–262.

Grant

K.L.

, Flohr

T.G.

, Krauss

, Sedlmair

, Thomas

, and Schmidt

(2014). Assessment of an advanced image-based technique to calculate virtual monoenergetic computed tomographic images from a dual-energy examination to improve contrast-to-noise ratio in examinations using iodinated contrast media. Invest. Radiol. 49, 586–592.

Postma

A.A.

, Das

, Stadler

A.A.

, and Wildberger

J.E.

(2015). Dual-energy CT: what the neuroradiologist should know. Curr. Radiol. Rep. 3, 1–16.

Potter

C.A.

, and Sodickson

A.D.

(2016). Dual-energy CT in emergency neuroimaging: added value and novel applications. Radiographics, 36, 2186–2198.

10.

Nandigam

R.N.

, Viswanathan

, Delgado

, Skehan

M.E.

, Smith

E.E.

, Rosand

, Greenberg

S.M.

, and Dickerson

B.C.

(2009). MR imaging detection of cerebral microbleeds: effect of susceptibility-weighted imaging, section thickness, and field strength. Am. J. Neuroradiol. 30, 338–343.

11.

Grinnon

S.T.

, Miller

, Marler

J.R.

, Lu,Y., Stout

, Odenkirchen

, and Kunitz

(2012). National institute of neurological disorders and stroke common data element project–approach and methods. Clin. Trials, 9, 322–329.

12.

Carney

, Totten

A.M.

, O'reilly

, Ullman

J.S.

, Hawryluk

G.W.

, Bell

M.J.

, Bratton

S.L.

, Chestnut

, Harris

O.A.

, Kissoon

, and Rubiano

A.M.

(2017). Guidelines for the management of severe traumatic brain injury. Neurosurgery, 80, 6–15.

13.

Lobato

R.D.

, Sarabia

, Rivas

J.J.

, Cordobes

, Castro

, Muñoz

M.J.

, Cabrera

, Barcena

, and Lamas

(1986). Normal computerized tomography scans in severe head injury: prognostic and clinical management implications. J. Neurosurg. 65, 784–789.

14.

Narayan

R.K.

, Kishore

P.R.

, Becker

D.P.

, Ward

J.D.

, Enas

G.G.

, Greenberg

R.P.

, Da Silva

A.D.

, Lipper

M.H.

, Choi

S.C.

, Mayhall

C.G.

, and Lutz III

H.A.

(1982). Intracranial pressure: to monitor or not to monitor? A review of our experience with severe head injury. J. Neurosurg. 56, 650–659.

15.

Maas

A.I.R.

, Dearden

, Teasdale

G.M.

, Braakman

, Cohadon

, Iannotti

, Karimi

, Lapierre

, Murray

, Ohman

, and Persson

(1997). EBIC-guidelines for management of severe head injury in adults. Acta Neurochir. (Wien), 139, 286–294.

16.

Servadei

, Murray

G.D.

, Penny

, Teasdale

G.M.

, Dearden

, Iannotti

, Lapierre

, Maas

A.J.

, Karimi

, Ohman

, and Persson

(2000). The value of the “worst” computed tomographic scan in clinical studies of moderate and severe head injury. Neurosurgery, 46, 70–77.

17.

Chang

E.F.

, Meeker

, and Holland

M.C.

(2006). Acute traumatic intraparenchymal hemorrhage: risk factors for progression in the early post-injury period. Neurosurgery, 58, 647–656.

18.

Sullivan

T.P.

, Jarvik

J.G.

, and Cohen

W.A.

(1999). Follow-up of conservatively managed epidural hematomas: implications for timing of repeat CT. AJNR Am. J. Neuroradiol. 20, 107–113.

19.

Piek

, Chesnut

R.M.

, Marshall

L.F.

, Berkum-Clark

M.V.

, Klauber

M.R.

, Blunt

B.A.

, Eisenberg

H.M.

, Jane

J.A.

, Marmarou

, and Foulkes

M.A.

(1992). Extracranial complications of severe head injury. J. Neurosurg. 77, 901–907.

20.

Schindelin

, Arganda-Carreras

, Frise

, Kaynig

, Longair

, Pietzsch

, Preibisch

, Rueden

, Saalfeld

, Schmid

, and Tinevez

J.Y.

(2012). Fiji: an open-source platform for biological-image analysis. Nat. Methods, 9, 676–682.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

0.02 MB