High-resolution neurosonographic examination of the lenticulostriate vessels in neonates with hypoxic-ischemic encephalopathy

Abstract

Objective

To assess the feasibility of visualizing lenticulostriate vessels (LV) using a linear high-resolution ultrasound probe and characterize LV morphology to determine whether morphological alterations in LV are present in neonatal hypoxic–ischemic encephalopathy (HIE) as compared to the unaffected infants.

Methods

We characterized LV by their echogenicity, width, length, tortuosity, and numbers of visualized stems/branches in neurosonographic examinations of 80 neonates. Our population included 45 unaffected (non-HIE) and 35 with clinical and/or imaging diagnosis of HIE. Of the neonates with clinical diagnosis of HIE, 16 had positive MRI findings for HIE (HIE+MRI) and 19 had negative MRI findings (HIE-MRI). Annotations were performed twice with shuffled data sets at a 1-month interval and intrarater reliability was assessed. Focused comparison was conducted between non-HIE, HIE+MRI and HIE-MRI neonates whose images were acquired with a high frequency linear transducer.

Results

Studies acquired with the two most frequently utilized transducers significantly differed in number of branches (p = 0.002), vessel thickness (p = 0.007) and echogenicity (p = 0.009). Studies acquired with the two transducers also significantly differed in acquisition frequency (p < 0.001), thermal indices (p < 0.001) and use of harmonic imaging (p < 0.001). Groupwise comparison of vessels imaged with the most frequently utilized transducer found significantly fewer branches in HIE + MRI compared to HIE-MRI negative and non-HIE patients (p = 0.005).

Conclusion

LV can be visualized in the absence of pathology using modern high-resolution neurosonography. Visualization of LV branches varies between HIE + MRI, HIE-MRI neonates and controls.

Advances in knowledge

High-resolution neurosonography is a feasible technique to assess LV morphology in healthy neonates and neonates with HIE.

Introduction

Neurosonography is a mainstay in the imaging evaluation of the neonatal brain. ¹ It is highly sensitive for the detection of hemorrhage, hydrocephalus, and periventricular leukomalacia (PVL) and is an ideal screening examination owing to its inexpensiveness, portability, minimal requirement for sedation, and absence of ionizing radiation. ^2–4 Numerous diagnoses such as lenticulostriate vasculopathy (LSV) are only established via neurosonography.

LSV (also known as thalamostriate vasculopathy and mineralizing vasculopathy) describes the sonographic identification of abnormally echogenic punctate or branching linear structures in the basal ganglia and thalamus. These echogenic structures were first reported by Grant et al ⁵ and were subsequently found to represent echogenic vessels ^6,7 which corresponded to the lenticulostriate (medial and lateral striate), thalamoperforate and thalamogeniculate vessels. These vessels comprise the terminal perforating arteries arising from the proximal vessels of the circle of Willis which supply the deep gray structures and surrounding parenchyma (Figure 1). Histopathological examination of these echogenic vessels in LSV demonstrated deposits of basophilic mineralizing material in the arterial walls (Figure 2). ^8,9 LSV has been associated with multiple disorders such as congenital malformations, ¹⁰ congenital infections, ⁶ metabolic disorders, ¹¹ chromosomal abnormalities, ⁷ maternal drug exposure, ¹² twin-twin transfusion syndromes, ^12,13 and particularly with congenital heart disease. ⁹ At the time that LSV was first described, it was thought that the LV were only sonographically distinguishable from their surrounding brain parenchyma in the presence of pathology ^14,15 ; however, we report that modern high-resolution ultrasound evaluation is allowing detailed visualization of the LV even when not pathologic and further hypothesize that subtle morphologic differences in the LV may be markers of disease. As the LV supply tissues which are frequently involved in hypoxic–ischemic encephalopathy (HIE), we hypothesized that morphologic differences in the LV may be manifest in the setting of HIE. The morphology of the LV have previously been classified in adult 7 T magnetic resonance angiography (MRA). ^16,17 Within the adult population, fewer stems and branches have been observed in hypertensive and diabetic adults and fewer branches were also observed in patients with lacunar infarcts of the basal ganglia as compared to normal controls. ^18–20 To our knowledge, our study is the first to describe grayscale sonographic LV morphology utilizing a similar classification system.

Figure 1.

Illustration demonstrating the branch vessels of the circle of Willis which supply the deep parenchymal structures of the basal ganglia, thalamus and internal capsule. Lenticulostriate vasculopathy (also known as thalamostriate vasculopathy or mineralizing vasculopathy) describes abnormally echogenic vessels in the region of the basal ganglia and thalamus which comprise the anatomic vessels depicted in this illustration including the medial, distal medial, lateral striate, thalamoperforate and thalamogeniculate arteries. (IP = image plane, tCN = tail of caudate nucleus, ACA = anterior cerebral artery, MCA = middle cerebral artery, Pcom = posterior communicating artery, PCA = posterior cerebral artery, ICA = internal carotid artery, GP = globus pallidus, p = putamen, IC = internal capsule, CN = caudate nucleus, TH = thalamus, cc = corpus callosum). Illustration ^©2021 Children’s Hospital of Philadelphia. All rights reserved.

Figure 2.

(a, left) Sagittal grayscale image of a 1-year-old male with autopsy proven mineralizing vasculopathy performed with a GE LOGIQ Ultrasound utilizing an L2-9 transducer demonstrates echogenic LV with focal punctate echogenicities (white arrows). (b, right) Histologic examination shows a small vessel (V) within the basal ganglia with perivascular mineralization (arrow). Hematoxylin and eosin stain, 400x magnification.

Methods and materials

Patients and study selection

This study was approved by the Institutional Review Board at our institution (Children’s Hospital of Philadelphia), informed consent exemption was granted. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were used. ²¹ Our study population was established through a query of our institution’s radiologic imaging database for neonates who received neurosonographic examinations between January 2018 and December 2020. Owing to the difficulty of finding completely normal controls within the population for whom neurosonography is indicated, non-HIE patients were defined by normal sonographic examinations and were excluded if there was presence of clinical documentation indicating the presence of either severe neurologic disease (e.g. seizures, COVID meningitis, infarcts, intraventricular hemorrhage) or diseases which have been associated with LSV (e.g. congenital heart diseases including hypoplastic left heart syndrome, transposition of the great arteries, Tetralogy of Fallot, twin-twin transfusion syndrome, TORCH infections, etc.). Patients with HIE were identified by a database query for terms related to HIE. Presence of clinical HIE was verified by documentation of clinicians at our institution who utilized the Sarnat Grading Scale. For inclusion, HIE patients were also required to have had both MRI and neurosonographic examinations such that the populations could be further classified depending on whether MRI findings compatible with HIE were present (HIE+MRI) or absent (HIE-MRI). When MRI findings compatible with HIE were present, the territories of parenchymal involvement were also noted. Our overall HIE population included all identified patients who met inclusion criteria within this timeframe and non-HIE patients were included until a number that was proportionate to the number of HIE patients was accrued.

Neurosonographic studies were included for review if a cine series of at least one hemisphere was acquired in the parasagittal plane. Images were acquired utilizing either a Phillips EPIQ ultrasound with mC12-3, L12-5, or eL18-4 transducers or a GE LOGIQ ultrasound using an L2-9 transducer. Studies were excluded if neurosonographic cine imaging was not performed in the parasagittal plane or if there was an artifact obscuring the LV ( Figure 3).

Figure 3.

Patient selection flowsheet demonstrating process of identification and exclusion criteria for non-HIE, HIE + MRI and HIE-MRI neonates. HIE, hypoxic–ischemic encephalopathy.

Imaging

Classification

LV morphology has previously been evaluated with 7 T time-of-flight (TOF) MRA in adults and classified with respect to number of stems, number of branches, vessel length, and tortuosity. ^16,17 Using a similar approach, we classified vessels by the number of visible stems and branches, vessel width and echogenicity, normalized average length, and tortuosity.

Branches and stems

Stems were defined as the longest independent echogenic structures extending from the vessel base into the deep gray parenchyma of the basal ganglia/thalamus (Figure 4, middle row). Branches were defined as shorter echogenic structures originating from the stems (Figure 4, middle row).

Figure 4.

Original (top row) and annotated (middle and bottom row) right paramedian images from a sagittal cine sequence of a 1-month-old male acquired with a Phillips EPIQ using an eL18-4 transducer. Middle row, stems (yellow lines) were defined as linear echogenicities extending from the base and branches (red lines) were defined as linear echogenicities branching from a stem. Lengths of stems were normalized based on the distance from the vessel based to the superior pia (L_BP, blue line) Bottom left, width and echogenicity of the LV (yellow box) were compared to that of an available pial surface (white box). Bottom right, tortuosity was calculated via a distance metric dividing the length of the stem/branch (yellow line) by a line connecting the ends of the stem/branch (green line). LV, lenticulostriate vessels.

Length

Length was defined by the average normalized length of all observable stems. As measured length in pixels could be influenced by gestational age, magnification during acquisition and annotation, we chose to assess a normalized length (L_Norm) which was produced by adjusting the vessel lengths by the largest distance from the base of the vessel to a point on the superior pial surface on the same image (L_BP) as in the equation below (Figure 4, middle row) using an approach based conceptually on previous methods of anatomic standardization. ²²

$${\displaystyle L_{Norm}=\frac{\sum _{i=1}^n\frac{L{V}_n}{L_{BP}}}{n}}$$

Width and echogenicity

Echogenicity was classified as imperceptible in the absence of any visible stems (1), perceptible though less echogenic than the pia (2), or as echogenic or greater than the pia (3) (Figure 4, bottom left). Width was classified in a similar manner with imperceptibility (1), width smaller than the pia (2), and width equal to or greater than the pia (3).

Tortuosity

Tortuosity was defined by the distance metric described by Bullitt et al. ²³ The total lengths of each stem and branch were divided by the length of a line connecting the beginning and endpoints of the stems and branches (Figure 4, bottom right).

Vessel annotation and analysis

Images were manually annotated in PowerPoint (Microsoft PowerPoint; Microsoft; Redmond, WA) then exported and processed in MATLAB (MATLAB v. R2018a, The MathWorks Inc.; Natick, MA). Lengths of the vessels as well as the line from the LV base to the superior pial surface were measured in number of pixels from which normalized lengths and tortuosity measures were produced (Figure 5). A second set of annotations was produced by the same rater using a shuffled data set at an interval of 1 month from the first annotation. Both sets of annotations were processed and intrarater reliability was assessed.

Figure 5.

Raw (left), annotated (middle) and plotted analysis output (right) of the LV from a parasagittal image of a 41-week-old male acquired with a Phillips L12-5 linear transducer in the parasagittal plane. Vessels (yellow) and distance from the vessel base to the superior pial surface (green) were annotated in PowerPoint exported for analysis in MATLAB. Pixel lengths (P), normalized pixel length (nP) and tortuosity (T) measures were produced. LV, lenticulostriate vessels.

Statistical analysis

Statistical analysis was performed in SPSS software v. 26 (IBM SPSS Statistics for Windows, v. 26, IBM Corp., Armonk, NY). Multiple univariate analyses were performed via χ²/Fisher’s exact for categorical/ordinal data and Kruskal–Wallis/one-way ANOVA/Independent samples t-test as appropriate for continuous data. Intrarater reliability was calculated with intraclass correlation coefficients. Multiple bivariate linear, logistic, and ordinal regressions of vessel morphology were performed as appropriate. A p-value of 0.05 was selected as the threshold for statistical significance.

Results

Overall population characteristics

80 neonates were examined of whom 45 were non-affected controls (non-HIE) and 35 had clinical diagnoses of HIE. Of the neonates with HIE, 16 had positive findings on brain MRI (HIE+MRI) while 19 had negative MRI findings (HIE-MRI) (Table 1). All 16 HIE+MRI patients demonstrated a pattern of parenchymal injury which included the central deep gray structures and 8 patients demonstrated injury to both central and peripheral structures. 10 patients within the non-HIE group also underwent MR examination at any point in their available medical record at an average of 18 days following the sonographic examination. The most frequently supplied indication for these MRIs in the non-HIE population was surveillance and no MRI performed in the non-HIE population demonstrated findings compatible with HIE.

Table 1.

Overall characteristics of all patients in the study divided by clinical group

	Total patients (n = 80)
	Non-HIE (n = 45)	HIE (n = 35)
		HIE-MRI (n = 19)	HIE + MRI (n = 16)
Sex, Female, n (% within category)	14 (31%)	5 (26.3%)	5 (30%)
Gestational age at birth, mean (SD), weeks	36 (6)	38 (2)	38 (3)
Gestational age at examination, mean (SD) weeks	39 (5)	39 (2)	39 (4)
APGAR at 1,5 min, mean (SD)	5.8 (2.3), 7.6 (1.5)	1.6 (0.9), 3.8 (2.0)	1.8 (2.1), 3.2 (2.6)
Most common additional diagnoses, n (% within category)a
Congenital heart disease Meconium aspiration Congenital diaphragmatic hernia Necrotizing enterocolitis Respiratory failure	0 (0.0%) 6 (13.3%) 11 (24.4%) 2 (4.4%) 16 (35.6%)	1 (5.3%) 3 (15.8%) 0 (0%) 0 (0%) 5 (26.3%)	1 (6.3%) 0 (0%) 1 (6.3%) 1 (6.3%) 4 (25.0%)
Transducers utilized, n (% within category)
Phillips eL18-4 Phillips L12-5 GE L2-9 Phillips mC12-3	32 (71.1%) 10 (22.2%) 2 (4.4%) 1 (2.2%)	15 (78.9%) 4 (21.1%) 0 (0%) 0 (0%)	12 (75%) 3 (18.8%) 1 (6.3%) 0 (0%)

HIE, hypoxic–ischemic encephalopathy.

^aDiagnoses with a prevalence <3 are not listed for comparison. These additional diagnoses included COVID, myelomeningocele, ASA lyase deficiency, multicystic dysplastic kidney, Turner’s syndrome, Hirschsprung’s disease, Jejunal volvulus, gastroesophageal reflux, etc.

Analysis of variance (ANOVA) did not demonstrate significant differences between the three populations in gestational age at birth (F(2, 77)=2.73, p = 0.72) or at examination (F(2, 77)=0.289, p = 0.75) nor was there any difference in proportion of premature patients in each group (X2(2,N = 80)=3.0, p = 0.22). The non-HIE population was defined by presence of normal sonographic examinations and absent clinical evidence of neurologic disease or disease previously associated with LSV. The most common diagnoses within the non-HIE population included respiratory failure, congenital diaphragmatic hernia, and meconium aspiration syndrome. The most frequently supplied indications for ultrasonography within the non-HIE population included prematurity, surveillance, evaluation for hemorrhage, and increased head circumference. Two patients within the HIE population had diagnoses of congenital heart disease; however, no patients within the non-HIE population had documented evidence of disease traditionally associated with LSV. Review of documented neurologic examinations was performed for each patient in the study and, except for one patient with congenital hypothyroidism, all patients within the non-HIE group were documented to have unremarkable examinations either on the day of sonographic examination or prior to sedation if neurologic examination could not be assessed due to sedation.

Comparison by transducer type and sonographic acquisition parameters

The LV were rated imperceptible in 2 of 80 total examinations which included one non-HIE patient and one HIE-MRI patient. Of the 80 examinations, 76 were acquired using either the Phillips eL18-4 (n = 59) or the L12-5 (n = 17) transducer (Table 1). Independent samples t-test comparison of LV features visualized by these two transducers demonstrated a greater number of visualized branches (p = 0.002) as well as greater average vessel echogenicity (p = 0.007) and thickness (p = 0.009) in patients scanned with the eL18-4 transducer. Studies acquired with the eL18-4 transducer were also more likely to employ a lower frequency (p < 0.001) with a higher thermal index (p < 0.001) and tissue harmonic imaging (p < 0.001) compared to the studies acquired with the L12-5 transducer. Multiple bivariate linear and ordinal regressions were performed comparing vessel morphology to frequency, thermal index, mechanical index, dynamic range, and frequency band within studies acquired with the same transducer. The effect of tissue harmonics could not be assessed as no studies acquired with the L12-5 probe utilized harmonic imaging and all but one study acquired with the eL18-4 transducer employed harmonics. Within studies acquired with the L12-5 transducer, the dynamic range was weakly negatively associated with number of visualized stems (Exp(B)=−0.33, p = 0.04). No significant associations were observed within studies acquired with the eL18-4 transducer although increasing thermal index was nearly significantly positively associated with increased number of visualized stems (Exp(B)=1.16, p = 0.056) and branches (Exp(B)=2.33, p = 0.077).

Comparison of vessel morphology by HIE group

To limit effects associated with the transducer, groupwise comparison of vessel morphology was limited to patients scanned with our higher resolution eL18-4 transducer and studies scanned with other transducers were excluded from analysis. The remaining population was similar to the overall population and no significant differences in gestational age at examination (F(2,56)=0.262, p = 0.77), birth (F(2,56)=2.345, p = 0.105) or sex (X2(2,N = 59)=0.309, p = 0.857)) were detected between the three clinical groups. One-way ANOVA revealed a significant difference in number of branches between the HIE + MRI, HIE-MRI, and non-HIE populations (p = 0.004, Figure 6). Post-hoc analysis showed that HIE + MRI neonates demonstrated fewer branches than both HIE-MRI (p = 0.03) and non-HIE neonates (p = 0.003). No other significant differences were observed in vessel characteristics or image acquisition parameters between the three populations (Table 2). ROC analysis of HIE-MRI positivity by number of branches demonstrated an area under the curve of 0.804 (p = 0.001, Figure 7).

Figure 6.

Bar chart comparing mean (95% CI) LV morphology between the HIE + MRI, HIE-MRI and non-HIE neonates scanned with the Phillips eL18-4 transducer along with matched representative still images from each category on the left. HIE + MRI patients demonstrated fewer branches compared to both HIE-MRI and non-HIE patients. *p < 0.05. HIE, hypoxic-ischemic encephalopathy; LV, hypoxic-ischemic encephalopathy.

Table 2.

LV characteristics of HIE + MRI, HIE-MRI and non-HIE neonates

	Patients scanned with the eL18-4 transducer (n = 59)
	HIE + MRI(n = 12)	HIE-MRI(n = 15)	Non-HIE(n = 32)	p- value
Number of stems, mean (SD)	2.4 (0.9)	2.7 (1.2)	2.8 (0.9)	0.523
Number of branches, mean (SD)	1.0 (0.9)	3.0 (2.8)	3.3 (1.9)	0.005*
Width				0.67
Imperceptible (n, % in category)	0 (0%)	1 (6.7%)	0 (0%)
Less than pia (n, % in category)	10 (83.3%)	12 (80%)	24 (75%)
Equal or greater than pia (n, % in category)	2 (16.7%)	3 (20%)	9 (33.3%)
Echogenicity				0.67
Imperceptible (n, % in category)	0 (0%)	1 (6.7%)	0 (0%)
Less than pia (n, % in category)	10 (83.3%)	11 (73.3%)	23 (72%)
Equal or greater than pia (n, % in category)	2 (16.7%)	3 (20%)	9 (28.1%)
Normalized Length, mean (SD)	0.36 (0.08)	0.40 (0.07)	0.39 (0.06)	0.327
Tortuosity, mean (SD)	1.06 (0.02)	1.07 (0.03)	1.08 (0.04)	0.499
Transducer settings
Frequency, MHz, mean (SD)	27.5 (2.5)	25.9 (2)	26.5 (4.5)	0.155
Mechanical index, mean (SD)	0.5 (0.1)	0.6 (0.1)	0.5 (0.1)	0.707
Thermal index, mean (SD)	0.4 (0.2)	0.4 (0.2)	0.5 (0.2)	0.070
Dynamic range, mean (SD)	73.2 (5.4)	73.1 (3.8)	70.3 (6)	0.155
Frequency band, n (% in category)				0.563
Gen	12 (100%)	12 (80%)	29 (90.6%)
Pen	0 (0%)	2 (13.3%)	2 (6.3%)
Res	0 (0%)	1 (6.7%)	1 (3.1%)
Harmonic, n (% in category)	12 (100%)	15 (100%)	31 (96.9%)	0.651

HIE, hypoxic–ischemic encephalopathy; LV, lenticulostriate vessels.

LV characteristics of HIE+MRI, HIE-MRI and non-HIE neonates who were scanned with the Phillips eL18-4 transducer. *statistical significance.

Figure 7.

ROC curve evaluating HIE-MRI positive state by number of branches visualized with the Phillips eL18-4 transducer. Area under the curve was 0.804 (p = 0.001). HIE, HIE, hypoxic–ischemic encephalopathy; ROC, receiver operating characteristic.

Intrarater reliability

Intrarater reliability was computed using intraclass correlations coefficients utilizing an average measurement, absolute agreement, two-way mixed effects model. There was excellent intrarater reliability for number of stems (ICC = 0.94, 95% CI = 0.91, 0.96, p < 0.001) and branches (ICC = 0.94, 95% CI = 0.91, 0.96, p < 0.001). Good reliability was also observed regarding width (ICC = 0.88, 95% CI = 0.81, 0.92, p < 0.001) and echogenicity (ICC = 0.87, 95% CI = 0.79, 0.92, p < 0.001). Normalized length annotations were moderately reliable overall (ICC = 0.65, 95% CI = 0.45–0.78, p < 0.001) and somewhat increased in patients whose LV were rated as thick and echogenic as the pia (n = 12, ICC = 0.78, 95% CI = 0.26;0.94, p = 0.001). Tortuosity measurements were not significantly reproducible (ICC = 0.22, 95% CI = −0.23, 0.5 p = 0.143).

Correlation of vessel morphology with additional clinical parameters

Limiting analysis to studies acquired with the eL18-4 transducer, multiple bivariate linear, logistic, and ordinal regressions were performed as appropriate to compare reliably assessed vessel morphologic features (stems, branches, width, and echogenicity) to clinical characteristics such as APGAR scores, GA at birth and examination, and sex. Compared to HIE + MRI patients, HIE-MRI (Exp(B)=2.339, p = 0.013) and non-HIE (Exp(B)=1.49, p = 0.005) patients were more likely to demonstrate an increased number of visible branches. No other statistically significant associations were identified (Table 3).

Table 3.

Multiple bivariate regression analyses of LV morphology by various clinical characteristics

	All studies acquired with eL18-4 (n = 59)
	Stems		Branches		Echogenicity		Width
Clinical characteristic	B or Exp(B)	p	B or Exp(B)	P	B or Exp(B)	p	B or Exp(B)	P
Sex (Male)	Exp (B) = 0.835	0.542	Exp (B) = 0.914	0.518	Exp (B) = 2.064	0.264	Exp (B) = 1.702	0.417
GA at birth	B = 0.019	0.508	B = 0.063	0.323	Exp (B) = 1.043	0.489	Exp (B) = 1.031	0.624
GA at examination	B = 0.014	0.633	B = 0.03	0.642	Exp (B) = 1.021	0.721	Exp (B) = 1.024	0.692
APGAR at 1 min	B = 0.051	0.313	B = 0.168	0.11	Exp (B) = 1.114	0.271	Exp (B) = 1.098	0.352
APGAR at 5 min	B = 0.074	0.156	B = 0.176	0.101	Exp (B) = 1.217	0.078	Exp (B) = 1.193	0.124
HIE grouping (compared to HIE+MRI)
Non-HIE	Exp (B) = 1.495	0.254	Exp (B) = 2.339	0.005 a	Exp (B) = 1.748	0.438	Exp (B) = 1.54	0.552
HIE-MRI	Exp (B) = 1.293	0.517	Exp (B) = 2.173	0.013 a	Exp (B) = 1.132	0.882	Exp (B) = 0.8	0.796

GA, gestational age; HIE, hypoxic-ischemic encephalopathy; HIE+MRI, clinical HIE with positive MRI findings of HIE;HIE-MRI, clinical HIE with MRI findings negative for HIE.

*p<0.05.

^ap<0.01

Discussion

High-resolution neurosonography has proven to be useful to evaluate the presence of LV and their morphology in neonates without any underlying neurological impairment. Our study contrasts with the prevalent thought that LV visibility in brain ultrasound is necessarily associated with pathology. ^15,24 These findings could potentially help determine differences in the LV morphology between healthy patients and patients with a suspected neurological disorder to better understand early markers of disease that could be readily identifiable at the bedside, adding value to this non-invasive, relatively inexpensive modality. ^4,25

In our study, we visualized the LV in 78 of 80 total examinations which included 44 of 45 non-HIE controls (Table 1) who were defined by the presence of normal sonographic examinations without clinical evidence of severe neurologic disease or disease which had traditionally been associated with LSV (such as congenital heart disease, twin-twin transfusion, TORCH infections, etc).

Indeed, the prevalence of LSV has steadily increased from an initial prevalence of 0.2–4.2% ^7,9 to up to 29.1% ²⁶ more recently which has been thought to reflect a combination of advances in ultrasonography, changes in patient population, and more attention being paid to this finding. ²⁷ Our study supports suggestions that differences in ultrasound technology may be contributing to this increased frequency of LV visualization as we observed significant differences in visualized LV morphology which were associated with the type of transducer and acquisition settings employed. ²⁷

We found that LV visualized with the eL18-4 transducer demonstrated a greater number of branches with greater width and echogenicity compared to the L12-5 transducer. Within studies acquired with the same transducer, correlation of morphologic characteristics found a weakly significant negative association between number of stems and dynamic range in the L12-5 transducer and no relationship in studies acquired with the eL18-4 transducer. The effect of harmonic imaging was of interest; however, it could not be assessed as no study acquired with the L12-5 employed harmonics, and all but one study acquired with the eL18-4 utilized harmonic imaging. Harmonic imaging is thought to offer superior visibility and diagnostic confidence due to decreased side lobe effect and decreased reverberation clutter; however, the technique is limited by reduced penetration and lower signal-to-noise ratio. ²⁸ Our findings indicate that the difference in visualized morphology between studies acquired with the two transducers may be due to intrinsic characteristics of the transducers themselves, the use of harmonic imaging, or a combination of factors. Further investigation isolating the contribution of the transducer and specific acquisition settings may better elucidate the relationship between sonographic parameters and the degree of LV visualization.

We also found evidence that differences in visualized LV morphology are associated with states of disease. We observed fewer visible grayscale LV branches in HIE+MRI neonates compared to HIE-MRI and non-HIE patients. The LV are thought to be particularly vulnerable to hypoxic–ischemic injury as they lack a rich capillary network and supply territories with high metabolic demand. ^29–31 Our HIE+MRI neonates all had MRI evidence of HIE involving the deep gray structures and demonstrated fewer LV branches on neurosonography compared to non-affected and HIE-MRI patients. These findings suggest that the decreased number of visible LV branches may be related to hypoxic–ischemic microvascular injury. The reduced number of branches in our HIE+MRI patients also evokes comparison to 7 T MRA examinations of the LV in adults in whom diminished number of LV branches were seen in the settings of hypertension, diabetes, and lacunar infarcts of the basal ganglia. ^18–20 Further evaluation is warranted to determine whether additional disease states are associated with changes in LV morphology and whether there are neurocognitive developmental associations.

Interestingly, Zheng et al ³² compared pulsed arterial spin labeling perfusion in HIE + MRI, HIE-MRI, and normal neonatal controls and demonstrated regional patterns of hyperperfusion and hypoperfusion in both HIE + MRI and HIE-MRI patients compared to controls. Although our assessment failed to detect a difference in LV morphology between HIE-MRI and controls, it is possible that discriminating features may exist in characteristics that were not well evaluated in our study (such as tortuosity), and further evaluation of these populations is indicated.

Limitations and future steps

A completely normal age-matched control population was not available within the population which underwent neurosonography. Non-affected patients were thus defined by normal sonographic examinations and absence of clinical documentation indicating presence of either severe neurologic disease or diseases which have been associated with LSV. The presence of nearly uniformly unremarkable documented neurologic examinations within the non-HIE population supports the validity of these patients serving as a relatively unaffected control population although there are inherent limitations to a retrospective analysis of documented diagnoses and clinical neurologic examinations. Future evaluations may benefit from a prospective selection of a more homogeneous normal control population and integration of detailed neurologic examination and neurocognitive developmental follow-up within the study design.

Our evaluation of LV morphology was retrospective and primarily utilized two linear high-resolution transducers. The use of a specific linear transducer was at the discretion of the technologist who may have chosen to only scan a single hemisphere in the parasagittal plane. Although no clear evidence of population bias was seen, it is possible that studies which were evaluated utilizing particular linear transducers were more abnormal at baseline or that the transducer was disproportionately utilized within specific populations. We reduced imaging heterogeneity by limiting groupwise analysis to studies acquired with a single transducer; however, this too introduced a limitation in that the results and specific thresholds separating HIE + MRI, HIE-MRI and non-HIE neonates were somewhat related to the transducer and specific sonographic acquisition settings. For example, a lower resolution examination might only visualize the branches infrequently or not at all, and so there may be no discernible difference between groups. Conversely, a future higher resolution examination may theoretically be able to visualize a wider spectrum of branches and vessel features such that differences between groups may be more prominent in some other vessel characteristics (such as tortuosity which could not be reliably assessed in our examination).

Subsequent attempts to characterize the normal grayscale sonographic range of LV may benefit from long-term prospective evaluation utilizing a single ultrasound machine, transducer, and scan protocol with progressive optimization for visualization of the LV.

Intrarater agreement was excellent for number of stems and branches, good for vessel echogenicity and width, moderate for normalized length, and unreliable for tortuosity. We attribute the decreased reliability for length to human measurement error. The tortuosity metric was even more susceptible to measurement error as it was a metric calculated from the specific manually annotated vessel course.

We chose to perform manual annotation because the LV were sometimes faint and not amenable to automatic segmentation. It is possible that future advances in grayscale sonographic technology or the use of advanced Doppler techniques utilizing background suppression algorithms such as MicroFlow imaging would make these vessels amenable to more reliable or possibly automated segmentation and analysis. ³³

Conclusion

Our findings demonstrate that modern sonographic visualization of the LV is not necessarily pathologic. With high-resolution linear ultrasonography, the LV were visualized in nearly all cases and the degree of LV visualization varied with the specific transducer and acquisition settings employed. Interestingly, neurosonographic comparison of LV morphology in neonates with HIE and non-affected controls demonstrated fewer visible branches in HIE+MRI neonates. These findings suggest that features of LV morphology may serve as a valuable new imaging biomarker in the evaluation of the neonatal brain. Further research into the diagnostic and prognostic utility of LV morphology in various neurologic conditions affecting neonates is warranted.