Clinical Validation of a Fast MRI Method to Evaluate Brain Vascular and Parenchymal Abnormalities in Sturge‐Weber Syndrome

ABSTRACT

Background

Sturge‐Weber syndrome (SWS) is a rare neurocutaneous disorder associated with venous capillary malformations, atrophy, and calcifications. Longitudinal imaging is limited by risks of sedation and gadolinium exposure in children.

Purpose

To evaluate whether strategically acquired gradient echo (STAGE), a rapid multi‐contrast quantitative MRI method, can reliably detect vascular and parenchymal abnormalities in SWS compared with conventional pre‐/post‐contrast MRI.

Study Type

Observational cross‐sectional.

Population

Twenty‐two patients with unilateral SWS diagnosed by previous MRI (13 female; ages 2–24 years).

Field Strength/Sequence

3T/T1‐weighted (T1W) and T2‐weighted (T2W) turbo‐spin‐echo, fluid attenuated inversion recovery, and a 3D gradient echo‐based STAGE sequence providing T1, proton density (PD), T2*, and R2* maps, susceptibility‐weighted imaging (SWI), quantitative susceptibility mapping (QSM), T1W with enhanced gray matter to white matter contrast (T1WE), and synthetic images of T2W, FLAIR, and gradient echo images.

Assessment

Conventional MRI and STAGE images were reviewed in 10 patients (training group), side‐by‐side, to determine the STAGE‐derived images that identify SWS abnormalities, including leptomeningeal venous capillary malformations (LVCM), enlarged deep medullary veins, choroid plexus enlargement, cerebral atrophy, and calcifications. In the remaining test group of 12 patients, three reviewers scored these abnormalities on STAGE images and compared them with scores from conventional MRI.

Statistical Tests

Interrater reliability with intraclass correlation coefficient (ICC), Spearman's rank correlation, Wilcoxon signed‐ranked test, Mann–Whitney U‐test, Fisher's exact test. Statistical significance level was set as p < 0.05.

Results

LVCMs were visualized on STAGE with SWI and R2*. Calcifications were differentiated from venous abnormalities using PD, T1WE, synthetic gradient echo, and QSM. STAGE‐derived scores had excellent interrater reliability (ICCs > 0.90) and were similar to the conventional MRI scores despite some minor differences in some individual cases (total scores from conventional MRI vs. STAGE 8.9 vs. 8.7, p = 0.29).

Data Conclusion

STAGE provided rapid, non‐contrast, multi‐parametric imaging that reliably detected vascular and parenchymal SWS abnormalities seen on conventional MRI.

Evidence Level

2.

Technical Efficacy

Stage 3.

Plain Language Summary

Sturge‐Weber syndrome (SWS) is a rare neurocutaneous disorder with abnormal brain vessels, tissue loss, and calcium deposits. Young patients often receive repeated MRIs with contrast administration and sedation, which may have harmful effects. This study tested a fast, non‐contrast MRI method called strategically acquired gradient echo (STAGE) to evaluate venous vascular and parenchymal brain abnormalities in 22 young patients without sedation. STAGE detected abnormalities on the lobar level seen on conventional pre‐/post‐contrast MRI. Certain STAGE findings were associated with epilepsy severity. STAGE may offer a safer, quicker option for detecting and monitoring brain changes in patients with SWS.

1. Introduction

Sturge‐Weber syndrome (SWS) is a rare, non‐hereditary neurocutaneous disease characterized by a facial venous capillary malformation (port‐wine birthmark), leptomeningeal venous capillary malformation (LVCM), and ocular abnormalities such as glaucoma [1]. SWS is most commonly associated with a somatic activating mutation in the GNAQ gene, although some phenotypes have also been associated with GNA11 or GNB2 gene variants [2, 3, 4]. The estimated incidence of SWS is approximately 1 in 50,000 live births [1]. The clinical manifestations of SWS are highly variable, ranging from seizures, developmental delay, and motor impairment, and often correlate with the extent and location of brain involvement. While some patients experience only mild impairments and can live normal lives, others may experience debilitating recurrent seizures, severe intellectual disability, and/or motor and vision loss [1].

Imaging aids in the diagnosis, characterization, and management of SWS. Classic features on brain MRI include leptomeningeal contrast enhancement, choroid plexus enlargement, cerebral atrophy, and subcortical calcifications [5]. Enlargement of the deep medullary veins (EDMV) is also a common imaging finding and may represent venous remodeling to allow for compensatory vascular drainage [6]. Advanced imaging techniques, including susceptibility‐weighted imaging (SWI) and perfusion studies, can aid clinicians by providing insight into the extent of vascular involvement and parenchymal damage, often guiding both prognosis and therapeutic decisions [7].

Pre‐ and post‐contrast brain MRI is often used to diagnose SWS brain involvement, in established SWS patients when neurocognitive symptoms worsen, or to aid in surgical planning [7]. However, both lengthy sedation and contrast administration can have harmful side effects. For example, children with developmental disabilities face a higher risk of hypoxia during sedation compared to their neurotypical peers [8]. Additionally, there are concerns regarding the adverse effects of repeated gadolinium‐based contrast agent exposure, although long‐term negative clinical effects have not been clearly demonstrated [9, 10, 11, 12]. Efforts to minimize these potential harms have been made. A recent study evaluated the feasibility of performing non‐contrast/non‐sedated brain MRIs in pediatric patients diagnosed with SWS and determined that good‐quality images could be obtained in a vast majority of young patients and provided useful diagnostic information [13]. An additional area of interest is the development and clinical validation of rapidly acquired images. A multi‐institutional cross‐sectional investigation comprised of academic pediatric hospitals showed that the average turnaround time for a brain MRI with sedation was between 158 and 224 min and was 70–112 min without sedation [14]. The average image acquisition time was 38 min (SD: 14 min) for both sedated and non‐sedated MRI.

This time can be reduced substantially by applying advanced fast and ultrafast MRI acquisition protocols, which can also be important in reducing movement artifacts in non‐sedated MRI sessions [15, 16]. Application of ultrafast‐MRI has demonstrated high diagnostic accuracy for the detection of intracranial pathologies [17, 18]; however, such protocols usually do not include SWI, a superior technique to visualize details of venous abnormalities and calcifications in the human brain, which are common pathologies in SWS [5, 6, 19, 20, 21]. One recently developed method of rapidly acquired imaging known as strategically acquired gradient echo (STAGE) can provide multiple qualitative and quantitative images/maps in 7 min or less at 3 T, including SWI [22, 23, 24]. This reduced scan time is important to avoid the need for sedation in young patients, allowing more frequent and earlier low‐risk imaging, and thus making serial MRI examinations more feasible. Early serial imaging may detect the evolution of brain abnormalities, allowing researchers to better understand the pathogenesis of many neurological conditions and monitor the effects of interventions. Standardized imaging would also allow for more effective multi‐institutional studies for researching rare pediatric neurological conditions.

Thus, the aims of this study were (1) to evaluate STAGE imaging, acquired without sedation or contrast injection, for detecting typical cerebral venous vascular and parenchymal abnormalities in a cohort of young patients with SWS, (2) to compare the findings with those from conventional pre‐ and post‐contrast MRI, and (3) to assess the extent of SWS brain abnormalities relative to clinical seizure variables.

2. Methods

2.1. Participants and Clinical Assessment

This protocol was approved by the local Institutional Review Board of Wayne State University (protocol numbers: 1102009399 and 20‐01‐1765) and written informed consent was obtained from the patients or their parents prior to the study.

Inclusion criteria were: (i) unilateral SWS brain involvement established by previous conventional pre‐ and post‐contrast MRI, (ii) ages 3 months to 24 years, (iii) good or adequate quality of conventional MRI and STAGE images for review, (iv) no history of previous brain surgery; and (iv) current or previous post‐contrast brain MRI, acquired on a 3T magnet within 1 year, available for review.

Clinical seizure variables (age at seizure onset, duration of epilepsy, seizure frequency score) were obtained from medical charts and updated during the visits for the research MRI acquisitions. Severity of the patients' epilepsy was assessed by the seizure frequency score by assessing the number of clinical seizures in the year before the study, as described previously [6, 25]: 0—no seizure in the last 1 year; 1—one to 11 seizures per year (i.e., at least yearly but less than monthly seizures); 2—one to 4 seizures per month; and 3—> 4 seizures per month (i.e., at least weekly seizures, on average).

2.2. MRI Data Acquisition

All MRI studies were performed on 3T MRI scanners (MAGNETOM Verio or Cima.X, Siemens Healthineers, Erlangen, Germany) using a 32‐channel head coil. The total scanning time was up to 20 min for the conventional MRI protocol plus 7 min for the STAGE sequence. Strategies to ensure minimal motion artifacts on these non‐sedated MRIs in young subjects have been detailed in a previous study [13]. Conventional MRI acquisition included: (i) an axial T2‐weighted (T2W) turbo spin echo sequence (voxel size 0.44 × 0.44 × 4 mm³, TE/TR: 93/6000 ms; acquisition time: 1 min 14 s); (ii) an axial fluid attenuated inversion recovery sequence (FLAIR; voxel size 0.875 × 0.875 × 2 mm³, TE/TE: 128/9000 ms; 3 min 56 s); (iii) SWI (voxel size 0.5 × 0.5 × 2.0 mm³; field of view 224 × 168 × 128 mm; base resolution 448; TE 5.1 ms/18 ms, TR 30 ms/30 ms; bandwidth 410/160 Hz/pixel, 2 × accelerated GRAPPA parallel imaging with 24 reference lines, and 6/8 partial Fourier along phase encoding; 5 min); and (iv) a volumetric axial T1‐weighted (T1W) 3‐dimensional magnetization prepared rapid gradient echo sequence (MPRAGE; voxel size: 0.9 × 0.9 × 0.9 mm³, TE/TR: 3/1700 ms; 4 min 35 s). In some patients, this MPRAGE sequence was repeated after gadobutrol (Bayer Healthcare Pharmaceuticals Inc., Whippany, NJ, USA) (0.1 mL/kg) contrast administration. In addition, a STAGE acquisition was obtained prior to contrast administration. STAGE was acquired with two flip angles (6^o, 24^o) and a TR of 25 ms, each with two fully flow‐compensated echoes of 5 ms and 18 ms. The sequence was acquired with an imaging resolution of 0.6 × 0.6 × 2.0 mm³ and 64 axial slices in 3 min 30 s for each flip angle, a total scan time of 7 min.

2.3. STAGE Data Processing

An in‐house MATLAB program was used to generate the results automatically as previously described [22, 23, 24]. In brief, the first echoes (TE = 5 ms) of the two flip angles were used for T1 and proton density (PD) mapping; the second echo (TE = 18 ms) of the low flip angle scan (a PD‐weighted [PDW] image to improve cerebrospinal fluid (CSF)‐to‐veins contrast) was used for generating the SWI; and all four echoes were used for R2* (1/T2*) mapping and quantitative susceptibility mapping (QSM). After generating the quantitative maps of T1, PD, and T2*, the following synthetic images were created: synthetic FLAIR (sFLAIR), synthetic T2W (sT2W) by multiplying the T1 map with a T2* mask from the QSM data, synthetic double inversion recovery (sDIR), and synthetic gradient echo (sGRE) with arbitrary flip angles for various image contrasts from PDW to T1W. Finally, T1W with enhanced gray matter to white matter contrast (T1WE) images were created by subtracting the PDW from the T1W images [22]. Radiofrequency inhomogeneities were corrected during the processing as previously described [23].

2.4. Image Assessment

First, the acquired research MRIs were assigned to subsequent analysis if: (i) all conventional and STAGE sequences were available, and [2] the images had good or adequate image quality, as defined previously [13], i.e., they had no major movement artifacts to obscure SWS brain abnormalities. In the next step, conventional MRI results of the selected patients were reviewed, and a subgroup of 10 patients (i.e., training group), showing a wide range of SWS brain abnormalities, has been selected for a direct, side‐by‐side comparison of the conventional and STAGE images by three investigators (S.G.M., a radiologist with 4 years of experience, A.F.L., a pediatric neurologist with 15 years of experience with SWS, and C.J., an adult neurologist with 22 years of experience with SWS). The following common SWS brain pathologies were available for comparison on conventional MRI [6, 19, 20]: (1) LVCM on post‐contrast T1W, (2) EDMV on the minimum intensity projected (mIP) SWI, (3) enlarged choroid plexus on post‐contrast T1W, (4) atrophy on T2W, and (5) calcification on SWI and high‐pass filtered phase images. Once STAGE sequences best depicting these abnormalities were identified, STAGE images of the remaining patients (test group) have been scored by all three investigators independently, using a visual scoring protocol described recently for conventional MRI sequences [13]. The investigators recorded the presence (score 1) or absence (score 0) of LVCM, atrophy, and calcification in each of the four lobes in the affected hemisphere (score range: 0–4 for each). The following criteria were used to identify calcifications on STAGE‐derived images: (1) hypointense on the PDW and PD maps, (2) hyperintense or isointense to white matter on the T1WE, (3) negative on the high‐pass filtered phase images, and (4) negative susceptibility on QSM. Criteria 3 and 4 are complementary to criteria 1 and 2. This is because when there is a large mass of calcification with a complicated geometry the phase and QSM often have severe artifacts from strong susceptibility. In addition, the presence/absence of EDMV was assessed in five cerebral venous territories (frontal, central, parietal, temporal, occipital; score range: 0–5) [6], and enlarged choroid plexus was assessed in the posterior ventricular horn (score range 0–1). The STAGE‐derived average of the three reviewers was then compared to abnormalities detected on conventional MR images (i.e., the ground truth) and available from the study database, using the same scoring system. In some patients, where clinical brain CT scans within 1 year were available, the extent of calcifications defined by these CT scans was also compared to that on the STAGE‐derived images.

2.5. Statistical Analysis

Both conventional and STAGE‐derived MRI scores, specific for each abnormality and the total scores, were tabulated and evaluated by descriptive statistics, with mean and standard deviation. Inter‐rater reliability of the STAGE MRI scores from the test group (n = 12) was assessed using intra‐class correlation coefficient (ICC) with a two‐way random‐effects model for absolute agreement for average measures. Reliability was interpreted as excellent (1.00–0.90), good (0.89–0.75), moderate (0.74–0.50), or poor (< 0.50) [26]. The average scores from the three reviewers were then compared to the conventional MRI scores using the Wilcoxon signed rank test, and the MRI scores were also correlated with the clinical seizure variables using Spearman's rank correlations. If multiple scores correlated with a clinical variable, a follow‐up multivariable regression analysis was performed. Since calcifications are considered to be highly epileptogenic pathology, seizure variables were also compared between patients with versus without calcified regions, using the Mann–Whitney U‐test. p < 0.05 was considered statistically significant.

3. Results

A total of 26 patients were enrolled. Three of these (two at 10 months and one at 4 years of age) failed to complete the non‐sedated brain MRI, as they were not able to stay still despite all mitigation techniques. The success rate of the non‐sedated brain MRI was 88.4%. One participant (3 months of age) completed the non‐sedated MRI but was excluded from the analysis due to the questionable‐SWS diagnosis based on the MRI. Finally, there were 22 patients (13 females, mean age 14 years, range 2–24 years; Table 1) diagnosed with SWS who were included in the analysis. Three of these 22 patients received contrast agent during the conventional MRI acquisition of the research scan (following the STAGE acquisition). The remaining 19 patients had a recent post‐contrast T1W image available for the identification of the pial enhancement for LVCM detection. A full representative set of STAGE‐derived images and a conventional SWI are shown in Figure 1.

TABLE 1.

Clinical data of the 22 patients with SWS.

No.	Gender	Age (years)	SWS side	AAO (years)	Epilepsy duration (years)	Sz. freq. score
1	F	2	Right	0.6	1.4	1
2	M	2.5	Left	0.3	2.2	1
3	M	4	Left	3.3	0.7	1
4	M	9	Right	0.6	8.4	1
5	M	9	Right	0.8	8.2	1
6	M	10	Left	5	5	2
7	F	13	Right	0.8	12.2	1
8	F	13	Right	2.5	10.5	1
9	M	13	Left	1.5	11.5	2
10	F	13	Right	0.5	12.5	4
11	F	13	Right	1.3	11.7	1
12	F	15	Left	0.5	14.5	2
13	F	15	Right	n/a	n/a	0
14	F	17	Left	0.2	16.8	1
15	M	19	Right	0.5	18.5	2
16	F	20	Right	0.4	19.6	1
17	F	20	Left	0.1	19.9	2
18	F	20	Right	0.8	19.2	1
19	M	20	Right	1.5	18.5	1
20	F	21	Left	0.3	20.7	1
21	F	22	Right	2	20	0
22	M	24	Left	0.6	23.4	1

Abbreviations: AAO: age at onset (of epilepsy), F: female, M: male, Sz. freq.: seizure frequency.

FIGURE 1.

Representative images of STAGE versus conventional SWI from a 13‐year‐old girl with SWS. The images in the left panel were from the STAGE scan acquired in 7 min, while the SWI and phase in the right panel were from the conventional SWI scan acquired in 5 min.

3.1. STAGE Identification of All Five Types of Brain Abnormalities in SWS

The clinical variables and conventional MRI abnormalities were not different between the training and the test group (Table S1). In the training group, the STAGE‐derived images best visualizing each SWS brain abnormality were identified (summarized in Table 2 and illustrated in Figures 2, 3, 4, 5, 6, 7). Overall, LVCM and EDMV were best assessed using mIP‐SWI images (Figure 2). For the detection of small LVCMs, and in the vicinity of calcifications, R2* maps helped confirm the LVCM when SWI was equivocal (Figure 3). Choroid plexus enlargement (Figure 4) and cerebral atrophy (Figure 4) were well identified on STAGE‐derived sT2W, sFLAIR, and PDW or PD maps. Calcifications could be distinguished by the combination of a dark signal on PDW and PD maps. If doubt remained, T1WE helped confirm the determinations, as calcification appears hyperintense or isointense with white matter on the T1WE (Figure 5). A similar strategy helped in identifying subtle punctate calcified areas and differentiating them from small veins (Figure 6). QSM of the short echo (TE1) was also helpful to confirm calcification by showing negative susceptibility in calcified areas (Figure 7), versus positive susceptibility in venous structures. However, QSM was vulnerable to artifacts depending on the complexity of the geometry, especially in large, irregular calcifications in close proximity with enlarged venous structures (Figure S1).

TABLE 2.

STAGE‐derived images best visualizing SWS brain abnormalities identified on conventional MRI.

SWS abnormality	STAGE images
LVCM	SWI, R2*
EDMV	SWI (mIP, phase)
CPE	sT2W, sFLAIR, PD map
Atrophy	sT2W, sFLAIR, PD map
Calcification	PDW, PD map, T1WE, sGRE, QSM

Abbreviations: CPE: choroid plexus enlargement, EDMV: enlarged deep medullary veins, FLAIR: fluid‐attenuated inversion recovery, LVM: leptomeningeal venous capillary malformation, mIP: minimal intensity projection, PD: proton density, PDW: PD‐weighted, phase: high‐pass filtered phase images from SWI sequence, QSM: quantitative susceptibility mapping, sFLAIR: synthetic FLAIR, sGRE: synthetic gradient recalled echo, sT2: synthetic T2‐weighted, SWI: susceptibility‐weighted imaging, T1WE: T1‐weighted with enhanced gray matter to white matter contrast.

FIGURE 2.

Comparison of conventional MRI versus STAGE‐derived images for detection of various SWS brain abnormalities in a 9‐year‐old child with right posterior brain involvement. On conventional MRI, post‐contrast T1‐weighted images (a) showed a right parietal LVCM (dashed circles). SWI (b) showed a few EDVs (solid arrows) and calcification (dashed arrows), best identified on the SWI phase image (c). STAGE‐derived images showed the best depiction of the LVCM on the R2* map (d), the EDV on SWI (e), and calcification on the PD map (f).

FIGURE 3.

Detection of a small LVCM on post‐contrast conventional MRI versus by STAGE sequences. The area of LVCM is indicated by a solid circle on post‐contrast T1W (a) from the conventional MRI protocol as well as on the STAGE‐derived R2* map (b and c). The STAGE‐derived SWI mIP image (d) did not visualize the small LVCM well but detected an enlarged deep medullary vein in the same region (dotted circle).

FIGURE 4.

Detection of enlarged choroid plexus and lobar atrophy by STAGE‐derived images. Enlarged left choroid plexus (long arrows on a and b) from a patient can be visualized by synthetic FLAIR (a) and T2W (b) images. Lobar atrophy (short arrows on c and d) can also be detected in these two patients: left frontal atrophy on the PD map (c) and right parietal atrophy on the synthetic T2W image (d).

FIGURE 5.

Detection of CT‐verified calcification on non‐contrast STAGE images. Calcified areas in the right hemisphere of a 13‐year‐old girl (patient #7 in Table 1) were shown on the CT image (a) and also on various STAGE‐derived images. STAGE‐derived SWI (b) shows both the calcified lesions and numerous enlarged veins (including veins in the vicinity of calcifications) as hypointense areas, making it difficult to differentiate them. In contrast, the hypointense areas on the PD map (c) and the PD‐weighted synthetic GRE (d) are confined to the calcified regions, while the veins are only faint (and hyperintense). The calcified areas on the T1WE images (e) blend in with the cortex intensity, while veins appear as faint, hypointense structures. The first row and second row images were from two different slice locations of the same patient.

FIGURE 6.

Detection of subtle punctate calcified areas by STAGE‐derived PD map and T1WE images. Two small areas of punctate calcification (arrows) were hypointense on the PD map (a) but hyperintense on the T1WE image (b), consistent with calcification, as verified on conventional SWI (d) and phase (c) images. Small caliber veins (dot circled in the right hemisphere on c and d) appear hypointense on conventional SWI (and hyperintense on SWI phase images) but were not visible on these two STAGE‐derived images, which were from the first echo with minimal susceptibility effects. Inserts are amplified images of the calcification area.

FIGURE 7.

Calcification shown on STAGE images verified by a subsequent CT scan. A small medial laminar calcified area (arrows) can be seen on various STAGE‐derived images on a 13‐year‐old boy (patient #9 in Table 1) with left‐sided SWS brain involvement. The calcified areas are shown as hypointense on synthetic PDW (a), PD map (b), and SWI (d). In contrast, the calcified areas are hyperintense on the subtraction‐derived T1WE image (c). The phase of the first echo (e) shows a dark band which is more aliased on the second echo phase (f) due to the longer echo time and the strong susceptibility of the area. Note the calcified areas appear as negative (dark) susceptibility on the first echo QSM (g). The subsequent CT scan (h) 2 months later confirmed that the area was consistent with calcification rather than venous abnormality.

3.2. Assessment of Reproducibility of STAGE‐Derived MRI Scores

The inter‐rater reliability of the STAGE MRI scores evaluated in the test group was found to be excellent, with an ICC of 0.98 (CI: 0.96–0.99) for the total score and ICCs ranging from 0.93 to 0.98 for the abnormality subscores (Table S2).

3.3. Comparison of SWS Abnormalities on STAGE Versus Conventional MRI and CT Scans

The average STAGE‐derived scores in the test group for each of the five brain abnormalities were similar to the conventional MRI scores (Table 3). While some subtle abnormalities, especially small LVCMs and minor asymmetries of the choroid plexus, were more obvious (and confirmed) by post‐contrast T1W images, these mild discrepancies did not affect the binary lobar scores.

TABLE 3.

MRI scores of the various SWS abnormalities from the conventional MRI and STAGE images (average from three readers) in the test group (n = 12).

Pt. no.	LVCM		EDMV		CPE		Atrophy		Calcification		Total
	Conv.	STAGE	Conv.	STAGE	Conv.	STAGE	Conv.	STAGE	Conv.	STAGE	Conv.	STAGE
2	2	2	4	3.3	0	0	3	2.3	0	0	9	7.7
3	1	1	1	1	0	0	1	0.7	0	0	3	2.7
5	2	2	2	1.7	1	0.7	2	1.7	2	2	9	8
6	3	2.3	2	1.7	1	1	1	1.3	2	2	9	8.3
9	2	2	0	0	0	0	0	0.3	2	2	4	4.3
10	1	1	0	0	0	0	0	0	1	1	2	2
11	2	2	1	1.3	1	1	2	2	2	2.3	8	9
14	2	2.7	4	3.7	1	1	4	4	2	2	13	13.3
15	3	3	5	5	1	1	3	3.3	3	3	15	15
17	3	3.3	5	4.3	1	1	4	3.7	1	1	14	13.3
18	1	1	0	0	0	0	2	1.7	1	1.3	4	4
20	3	3.3	5	4.7	1	1	4	4	4	4	17	17
Mean	2.1	2.1	2.4	2.2	0.6	0.5	2.2	2.1	1.7	1.7	8.9	8.7
SD	0.8	0.8	2	1.9	0.5	0.5	1.4	1.4	1.1	1.1	5	5
p**	0.58		0.06		0.32		0.37		0.16		0.29

Note: p values refer to the comparison between the STAGE‐derived and conventional MRI scores (Wilcoxon signed rank test).

Abbreviations: Conv.: conventional MRI, CPE: choroid plexus enlargement, EDMV: enlarged deep medullary veins, LVCM: leptomeningeal venous capillary malformation, Pt. no.: patient number, SD: standard deviation.

While recent CT scans were only available in three patients, visual comparison of these scans to the STAGE‐derived images supported the accuracy of detecting the calcified brain regions, as illustrated in Figures 5 and 7 .

3.4. Correlation Between MRI Scores and Clinical Seizure Variables

Higher total conventional MRI scores were associated with lower age at seizure onset (Spearman's rho = −0.62). Of the various sub‐scores, both extensive atrophy (rho = −0.71) and high EDMV scores (rho = −0.44) correlated with lower age at seizure onset. In the follow‐up logistic regression analysis, with both atrophy and EDMV entered as independent variables, only the high atrophy scores showed a significant correlation with lower age at onset (partial r = −0.51; for EDMV scores: partial r = 0.13, p = 0.54). Higher calcification scores were associated with higher seizure frequency scores (rho = 0.42). Patients with at least one area with calcification (n = 7) had significantly higher mean seizure frequency scores than those with no calcification (1.5 vs. 0.7, respectively). Patient age or duration of epilepsy did not correlate with the MRI scores (p ≥ 0.42 and p ≥ 0.10, respectively). In the test group, STAGE‐derived atrophy (rho = −0.75) and EDMV scores (rho = −0.61) also showed a strong inverse correlation with the age at seizure onset despite the smaller sample size; seizure frequency (p ≥ 0.28) and duration of epilepsy (p ≥ 0.1) showed no correlations with the STAGE‐derived scores.

4. Discussion

This study demonstrated the feasibility and potential clinical utility of STAGE MRI, a fast, non‐contrast‐enhanced imaging method, to detect hallmark cerebral abnormalities associated with SWS in young subjects. Notably, while identification and characterization of these SWS brain abnormalities were in some instances more difficult than on conventional MRI, all the assessed features were reliably observed on STAGE‐derived images on the lobar level, demonstrating its potential to provide an optimal balance between diagnostic detail and patient safety.

Due to a lack of contrast‐enhanced images, LVCMs were among the most challenging features to visualize on STAGE but were detectable using a combination of SWI and R2* images. Similarly, subtle calcifications, which can mimic venous signals on conventional SWI, were successfully differentiated using a combination of PD map and T1WE images. Evaluating cerebral calcifications on MRI is also challenging in areas of venous malformations, particularly because these pathologies are in close proximity to each other and both show a dark signal on SWI. Although QSM can, in principle, distinguish calcification (diamagnetic; negative susceptibility) from venous abnormalities (paramagnetic; positive susceptibility), this separation can be unreliable in SWS lesions with complex geometry [27]. Large calcific deposits and LVCM generate steep susceptibility gradients and partial‐volume effects that compound the ill‐posed dipole inversion, yielding the characteristic appearance of marked hypointensity on SWI and mixed bright/dark patterns on QSM [27]. In this setting, short‐echo magnitude images, specifically PD maps and T1WE, provided complementary contrast and, when interpreted together, more consistently differentiated calcification from venous structures. Notwithstanding these limitations, QSM remains uniquely suited to assessing venous oxygenation [27]. Future work will exploit venous QSM to estimate oxygen saturation within abnormal veins in SWS, enabling quantitative characterization of venous pathophysiology. Altogether, these findings suggest the potential of STAGE to replace more lengthy MRI protocols in selected patient populations.

There are several other rapid MR acquisition techniques, sometimes referred to as synthetic MRI, which have been developed to provide clinically significant images with reduced scan time. One of these utilizes a single 3D scan with a multi‐pathway multi‐echo sequence along with machine learning to generate T1W, T2W, PDW, MPRAGE, and FLAIR images but no SWI or QSM [28]. In a recent review [29], multiple synthetic MR techniques including segmented inversion recovery‐prepared true fast imaging with steady‐state precession, multi‐dynamic multi‐echo, MR fingerprinting, and multi‐pathway multi‐echo were analyzed, all of which generated T1W, T2W, and PDW images. Additional images generated on some of these techniques include T1W‐FLAIR, T2W‐FLAIR, DIR, phase sensitive inversion recovery, short‐tau inversion recovery, and MPRAGE [28]. While these techniques produced good quality images with a reduced scan time, an advantage of STAGE is the inclusion of SWI and QSM which this study showed aided in evaluating multiple changes that occur with brain involvement in SWS such as EDMV, LVCM, and calcifications. A previous study supported the hypothesis that deep venous remodeling serves as a compensatory response to impaired superficial venous drainage—one of the pathomechanisms in SWS [6]. By enabling sensitive and comprehensive assessment of both superficial and deep venous structures, STAGE imaging may facilitate further exploration of these mechanisms across disease stages and aid in prognostic analysis. One limitation of the STAGE method is the absence of a true T2W image, as the technique relies on gradient‐echo‐based acquisitions. Instead of acquiring an actual T2W image, STAGE generates a synthetic T2W image by adding T2* weighting to the T1 map, following the principle that brain tissues with a long T1 tend to have a correspondingly long T2. Although T2W imaging is less important in SWS, which is primarily a vascular disorder, the synthetic T2W images produced by STAGE should be interpreted with caution when applied to other conditions where genuine T2W contrast is required. In addition, like most of the rapid MRI acquisition techniques, STAGE does not generate a diffusion weighted image, which is useful for detecting early white matter and ischemic changes and is used occasionally in SWS, especially after new‐onset motor weakness (i.e., stroke‐like episodes) [7]. It should also be noted that MRI with sedation may be inevitable in young children despite the shortened acquisition time, although shorter scanner time is beneficial both by limiting the amount of sedatives given and by increasing scanner throughput. The use of non‐sedate MRI may be more relevant in the research setting, especially in pediatric clinical trials with repeated scans to track early changes and monitor treatment effects, where repeated sedation could raise ethical concerns, but extended preparation time in research scanners can increase the success rate, as detailed in a previous study [13].

Consistent with prior studies on the pathophysiology of SWS, MRI abnormalities were significantly associated with clinical epilepsy severity. High atrophy scores on both conventional MRI and STAGE were correlated with earlier seizure onset, and the presence of calcifications in the whole group was associated with more frequent seizures, supporting the notion that extensive structural changes may underlie a more severe epileptogenic network in SWS [6, 20, 30, 31]. These results parallel previously reported associations between cortical atrophy, calcification, and seizure burden, and reinforce the potential of STAGE to serve as both a diagnostic and prognostic tool in clinical care without the use of a contrast agent.

Importantly, this study builds on previous studies, one of which demonstrated the feasibility of non‐sedated, non‐contrast brain MRI in pediatric SWS patients, showing that diagnostic images could be acquired using tailored protocols in awake or naturally sleeping children [13]. It also builds on studies which highlight the diagnostic value of SWI in detecting venous remodeling and associated structural abnormalities in SWS [6, 32]. In the current study, it was further shown that a single, standardized STAGE acquisition has the potential to replace multiple conventional sequences due to its ability to detect clinically relevant features.

The ability of the STAGE protocol to provide multi‐parametric imaging within a short acquisition time (< 7 min) is beneficial in pediatric neuroimaging. Avoiding both sedation and contrast administration substantially reduces risk, increases access to serial imaging, and improves patient/family acceptance of MRI follow‐up—especially relevant in young children with neurodevelopmental concerns and heightened sensitivity to sedatives [8, 13]. This capability may be particularly useful for early longitudinal monitoring, a particular need in rare neurocutaneous syndromes like SWS, where developmental progression and treatment response vary widely.

5. Limitations

Given this is a single‐center study involving a rare disease, the sample size is small which limits statistical power. Comparison of independently reviewed conventional versus STAGE images was only performed in a subset of these patients, as the first 10 cases were used for initial training to identify the optimal STAGE‐derived images for detecting each abnormality. STAGE does not provide true T2‐weighted or diffusion‐weighted images and, despite this limitation, we were able to show that SWS features could be identified on STAGE imaging. The matching abnormalities were observed on the lobar level, which does not account for potential minor intralobar differences, which were not assessed in this study. However, such minor differences on conventional versus STAGE images are unlikely to affect clinical management in SWS. Another limitation is that in most of the patients, the acquisition of the post‐contrast images was performed at a different time point from (prior to) the non‐contrast conventional and STAGE images. Pial enhancement may be absent early and emerge at some point before 2 years of age [33]. In the present study, all MRIs were done at or after 2 years of age, and the ability of STAGE imaging to detect early (including presymptomatic) SWS brain abnormalities will need to be tested in future studies. The underlying genetic variants were not available in our cases, but future studies can evaluate the use of STAGE (and other novel MRI sequences) for their utility in identifying imaging phenotypes associated with different SWS gene variants. Lastly, while we have demonstrated associations between imaging scores and seizure variables, longitudinal follow‐up is required to assess whether STAGE‐derived metrics can predict future clinical outcomes or treatment response.

6. Conclusion

STAGE MRI provided a rapid, sedation‐ and contrast‐free imaging alternative to conventional MRI with the potential to detect a broad range of clinically meaningful abnormalities in children with SWS.

Funding

Research reported in this publication was supported in part by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health (NIH) under award numbers R01NS041922 and R61NS119434. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Part of the MRI data was acquired using the Cima.X 3T MRI scanner at the MR Core Research Facility of Wayne State University, supported by NIH under award number S10OD028724.

Supporting information

Data S1: jmri70222‐sup‐0001‐FigureS1‐TableS1‐S2.docx.