Toward Semiautomated Analysis of Cerebrovascular Reserve: Enhancing Objective Comparisons Using ^99mTc-HMPAO SPECT With Acetazolamide in Moyamoya

Abstract

BACKGROUND AND OBJECTIVES:

Cerebrovascular reserve (CVR) is a key physiological mechanism allowing the brain to adapt to fluctuating perfusion, particularly relevant in the management of neurovascular disorders such as idiopathic (iMM) and syndromic moyamoya (sMM). Although ^99mTc-HMPAO SPECT with acetazolamide is commonly used for CVR assessment, it faces limitations including low spatial resolution, artifacts, and variability in interpretation. This study primarily aims to evaluate a novel, semiautomated, and more objective method for interpreting HMPAO SPECT in CVR assessment. As a secondary objective, the method is applied to a cohort of patients who underwent revascularization surgery for iMM or sMM.

METHODS:

A retrospective analysis was performed on prospectively collected data from a tertiary neuroscience center, including 20 adult patients with iMM (n = 9) or sMM (n = 11). Clinical and imaging data were reviewed. ^99mTc-HMPAO SPECT images were assessed independently by 2 nuclear medicine physicians blinded to clinical details. Images were registered to T1-weighted MRI and overlaid with an arterial territory atlas. Territories classified as healthy by experts were defined as true negatives; all others as altered. Statistical comparisons were made using Student's t-tests with false discovery rate correction.

RESULTS:

Among the 20 patients (12 females), patients with sMM were older and had more cardiovascular risk factors. The proposed method significantly discriminated between altered and healthy perfusion territories. Compared with expert interpretation, the method demonstrated specificities of 93% (iMM) and 92% (sMM), with negative predictive values of 80% and 75%%, respectively.

CONCLUSION:

Although ^99mTc-HMPAO remains a validated modality for CVR assessment, its interpretation is operator-dependent. The proposed semiautomated method offers high specificity and greater objectivity, supporting its integration into clinical workflows. Further multicenter validation is warranted.

ABBREVIATIONS:

ASL

arterial spin labeling

CVR

cerebrovascular reserve

^99mTc-HMPAO SPECT

technetium-99m hexamethylpropylene amine oxime single-photon emission computed tomography

MM

moyamoya disease or syndrome

iMM

idiopathic moyamoya

sMM

syndromic moyamoya.

Cerebrovascular reserve (CVR) is a critical physiological mechanism reflecting the brain's ability to adapt to fluctuations in cerebral perfusion pressure.^1-7 This adaptive capacity is particularly vital in the context of neurovascular disorders, where impaired CVR is associated with an increased risk of ischemic events and poor clinical outcomes such as in moyamoya disease (iMM) and syndrome (sMM).^6,7 Thus, in neurosurgical practice, accurate evaluation of CVR plays an integral role in diagnosing vascular pathologies,^3,5 stratifying stroke risk, and guiding therapeutic decisions, including revascularization strategies.^2,8-11

Among the imaging modalities used to assess CVR, technetium-99m hexamethylpropylene amine oxime (^99mTc-HMPAO) single-photon emission computed tomography (SPECT) with Diamox challenge—commonly referred to as ^99mTc-HMPAO with acetazolamide—has gained significant attention.^1-5 This technique leverages the vasodilatory effects of acetazolamide, a carbonic anhydrase inhibitor, to detect regions with diminished perfusion reserve,^6,7 offering a semiquantitative assessment of cerebrovascular reactivity. The strengths of ^99mTc-HMPAO SPECT with acetazolamide lie in its ability to provide regional insights into cerebral perfusion,⁶ its widespread availability, and its established role to differentiate impaired CVR in pathologies such as moyamoya.⁶

This study primarily aims to evaluate a novel, semiautomated and more objective method for interpreting HMPAO SPECT with acetazolamide in the assessment of cerebrovascular reserve. As a secondary objective, we seek to test this method on a specific cohort of patients who have undergone revascularization surgery for moyamoya (MM). By integrating current evidence and clinical data, this work also explores the method's relevance, limitations, and potential role in the evolving field of neurovascular diagnostics.

MATERIALS AND METHODS

Population

The patient population derives from a catchment area of approximately 4 million persons in North of France. It is the unique hospital center managing cerebrovascular activity in this area allowing exhaustive inclusion. Annually, our center treats around 5 moyamoya cases. Our study retrospectively included a series of consecutive adult patients with moyamoya (idiopathic or syndromic) who underwent revascularization surgery at our institution between January 1, 2019, and January 1, 2025 (n = 20, female to male ratio 13:7). All patients presenting with age under 18 years or without ^99mTc-HMPAO with acetazolamide available in the preoperative and postoperative course were excluded (n = 4).

Ethics

The Lille Ethics Committee and the Collège de Neurochirurgie approved the study (no. IRB00011687 Collège de neurochirurgie IRB #1: 2024/77). This study complies with the French Data Protection Act (Loi Informatique et Libertés) and the reference methodology MR-004 relating to the processing of personal data in the context of health-related research (Compliance with the General Data Protection Regulation). According to French law, this anonymous retrospective observational study did not require patient informed consent. Nonopposition from all participants was confirmed from all participants.

Availability of Data and Material

The data sets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Institutional Management in MM

Each case is discussed collaboratively during neurovascular meetings. Patients receive optimized medical management including lifelong aspirin therapy, except in cases of hemorrhagic stroke. They undergo various imaging tests, including transcranial Doppler ultrasound, computed tomodensitometry angiography (CTA), magnetic resonance angiography, and outpatient cerebral angiography, to assess Suzuki staging and cerebrovascular reserve using ^99mTc-HMPAO SPECT with acetazolamide and cerebral perfusion MRI.^8-10 These studies are reviewed in consensus.

For asymptomatic patients with intact cerebrovascular reserve, neurological follow-up is conducted. Surgical intervention is indicated for patients with impaired reserve or recurrent symptoms, prioritizing direct cerebral revascularization via superficial temporal artery–middle cerebral artery M4 bypass, combined with local indirect revascularization through encephalo-duro-myo-synangiosis.¹¹

Postoperative follow-up includes CTA, 4D arterial spin labeling (ASL) MRI, and cerebral angiography. At 6 months, a comprehensive evaluation using cerebral perfusion MRI and 99mTc-HMPAO SPECT with acetazolamide is performed to monitor intervention effectiveness and guide further management.

Data Collection and Image Processing

All clinical and imaging data were thoroughly reviewed. Each ^99mTc-HMPAO SPECT study was scheduled on a 2-day sequential protocol, 7 days apart, referred as “basal” and “acetazolamide or Diamox” SPECT. On each study, patients were placed in sensorial isolation (alone in a dark and quiet dedicated room with an eye mask) for 20 minutes before the intravenous injection of 740 MBq of ^99mTc-HMPAO. After the injection, sensorial isolation was extended for 20 more minutes. For the Diamox study, 1 g of acetazolamide was intravenously injected at the beginning of the sensorial isolation. SPECT images were acquired within 5 min after the end of sensorial isolation, on a dedicated double-head gamma camera (Symbia S, SIEMENS Healthineers) with a parallel low-energy high-resolution collimator. SPECT acquisition was performed with 4 consecutive phases of 330s each (60 step-and-shoot projections; matrix 128 × 128; zoom 1.23) and images were reconstructed with an iterative algorithm (OSEM3D; 16 iterations; 8 subsets; Gaussian filter 8 mm) and Chang's attenuation correction method. Images were reviewed by 2 nuclear medicine specialists, ensuring consistency and accuracy in interpretation but masked to clinical information. Images were displayed on a Syngo.Via workstation (SIEMENS Healthineers) in axial, coronal, and sagittal planes, fused with a recent brain MRI if available (T2-FLAIR and T1-weighted sequences) for anatomic location. Basal and Diamox images were visually interpreted using a French rainbow look-up table. To evaluate CVR, both basal and Diamox SPECT images were displayed side-by-side and the intensity of cerebellum of the 2 image sets were adjusted to the same level (as a reference region considered least involved in moyamoya disease). An uptake decrement on basal SPECT, well correlated with a cerebral vascular territory, was considered altered. An uptake decrement on Diamox SPECT, worsened compared with basal SPECT, was considered as altered CVR and described as altered. The other territories described by the nuclear medicine physicians were considered healthy.

Postprocessing

All the data, ^99mTc-HMPAO SPECT as well as anatomical MRI (T1), acquired in Digital Imaging and Communications in Medicine (DICOM) format were first converted to nifti format thanks to dcm2niix.¹² The whole brain mask obtained from the T1-weighted MRI (T1w) was computed using the online brain volumetry system (volBrain, volbrain.net/). The 4 ^99mTc-HMPAO SPECT (2 without acetazolamide and 2 with acetazolamide) were registered and realigned onto the T1w image (SPM12, fil.ion.ucl.ac.uk/spm/software/spm12/). The cerebral arterial atlas,¹³ referenced in Montreal Neurological Institute Space, was coregistered with each patient's (T1w) images by normalizing the T1w image to Montreal Neurological Institute Space using a nonlinear transformation and then applying the inverse transformation to the atlas (Advanced Normalization Tools, stnava.github.io/ANTs/). Only values within the binary brain extraction mask, derived from the T1w images processed by volBrain, were retained for analysis. This approach served 2 purposes. First, for each ^99mTc-HMPAO SPECT, we extracted the mean intensity value of the cerebellum (labels 29 and 30 in the cerebral arterial atlas) to perform intensity normalization across the rest of the brain. Second, we extracted the average intensity value of each atlas region (normalization ratio) for statistical analysis. We did not include the entire brain arterial territories atlas in our analysis. Specifically, we excluded the medial lenticulostriate territory (labels 3-4), the insular pars of middle cerebral artery territory (labels 15-16), the basilar territory (labels 25-26), and the different areas related to the cerebellum (labels 27 to 32). To reduce the number of regions for the statistics, we merged the parietal pars of middle cerebral artery (labels 9-13 on the left and 10-14 on the right) as superficial middle parietal territory; and the temporal pars of middle cerebral artery territory (labels 11-17-23 on the left and 12-18-24 on the right) as superficial middle temporal territory. We classified as true negative territories identified by the nuclear medicine specialist as healthy, and all remaining territories were considered pathological. To establish normal values of the “normalization ratio,” our methodology was applied to 12 additional subjects who underwent a ^99mTc-HMPAO SPECT with acetazolamide challenge. These patients were chosen if their examination (basal and acetazolamide) were considered as normal by the nuclear medicine specialist (Figure 1).

FIGURE 1.

Postprocessing principle.

Statistical Analysis

Statistical analysis was performed in R version 4.3.3 using packages ggplot,¹⁴ ggpubr,¹⁵ and geomtextpath for figures, and rstatix¹⁵ and tidyverse¹⁶ for the statistics tests. All the statistics were computed using a Student test corrected by false discovery rate for multiple comparisons.

RESULTS

Population

All included patients (n = 20) were of European origin. There was a predominance of females in the idiopathic moyamoya group (all cases were bilateral) compared with the syndromic moyamoya group (all cases were unilateral), where patients were generally older and had a higher prevalence of cardiovascular risk factors. Except for one individual, all patients were right-handed. The most common mode of presentation was an established stroke (n = 15/20). Previous cranial irradiation was documented in 3 patients, including one who received standard fractionated radiotherapy, one treated with CyberKnife, and another who underwent proton therapy. Groups differences are summarized in Table 1.

TABLE 1.

Populations Characteristics

Parameters	iMM (n = 9)	sMM (n = 11)
Sex ratio F:M	7F:2M	5F:6M
Age (Y/O)	37,1	52
TIA manifestation	1	2
Ischemic stroke	9	6
Hemorrhagic stroke	0	0
High blood pressure	0	5
Active smoking	3	4
Diabetes	2	3
Obesity	1	4
Cranial irradiation	0	3
Cerebral dominance	1L/8 R	11R
Side affected	Bilateral	5L/6 R
Side treated	3L/5R/1 bilateral	5L/6 R

iMM, idiopathic moyamoya; L, left-sided; R, right-sided; sMM, syndromic moyamoya.

Surgical Outcomes

Double bypass combined with encephalo-duro-arterio-myo-synangiosis (EDAMS) was successfully performed in 3 patients from each group. Indirect revascularization with EDAMS alone was carried out in 2 patients from the syndromic moyamoya group and in 1 patient from the idiopathic moyamoya group. The remaining patients underwent a single bypass combined with EDAMS. Only 1 patient required a staged double direct revascularization within a short time frame.

All anastomoses patencies were initially evaluated and confirmed on multimodal evaluation, including indocyanine green angiography, micro-Doppler, transcranial Doppler, CTA, time-of-flight MRI, and digital subtraction angiography. This patency was confirmed at the final follow-up using transcranial Doppler, CTA, time-of-flight MRI, and 4D ASL MRI.

Procedure-related complications included 2 major ischemic strokes resulting in new neurological deficits (one due to massive hypotension related to preoperative cystocath placement, which was later found to be mispositioned, leading to a perioperative septic shock) and 3 transient deficits without identifiable ischemic lesions on early and follow-up MRI. No cases of meningitis or scalp necrosis were observed. In addition, 2 lacunar strokes occurred after cerebral angiography in the syndromic moyamoya group. Regarding aspirin therapy, no hemorrhagic complications necessitating surgical revision were reported in any patient.

Comparison by Specific Arterial Territories

Based on Figure 2 and on whether the disease is unilateral or bilateral, we considered dot in between the bracket interval (defined by the mean ± SD of the control group) as a true negative if green (if health territory) and as a false negative if red (if altered territory). We can define the confusion matrices as represented in Table 2. For the iMM group compared with the sMM group, our method specificities were, respectively, 93% and 92 % and negative predictive values were, respectively, 75% and 80%. The F1 scores were, respectively, 64% and 53%. Accuracies were, respectively, 80% and 75%.

FIGURE 2.

Illustration of the simplified vascular territories exposed to acetazolamide (X-axis) according to the normalization ratio (Y-axis). Each result is presented with consideration of whether the condition is syndromic (sMM: 2 upper graphs) or idiopathic (iMM: 2 lower graphs). Altered territories (classified by the nuclear medicine specialist) is represented with a red dot. Any other territory, classified as healthy, are represented with a green dot. The brackets interval represents the normal values of the normalization ratio, extracted from our normal subjects (mean ± 2SD). ACA, anterior cerebral artery; MCA, middle cerebral artery.

TABLE 2.

Confusion Matrices Comparing the Experienced Nuclear Physician to Our Defined Method Based on Whether MM Is Idiopathic or Syndromic

Idiopathic MM		Nuclear physician
		Altered	Healthy
Our method	Altered	35	9
	Healthy	31	125

Syndromic MM		Nuclear physician
		Altered	Healthy
Our method	Altered	28	10
	Healthy	40	122

MM, moyamoya disease.

Cerebrovascular Reserve Evaluation

As demonstrated in Figure 3, significant differences were detected between altered and healthy territories in each preoperative and postoperative session with respect for basal session and acetazolamide session. These significant differences diminish in the postoperative evaluation. Only one exception was noted for postoperative session under Diamox in the sMM group. In both groups, significant difference was noted in altered regions in preoperative period between basal and acetazolamide exposition. In the sMM group, significant difference was also reported in altered region in postoperative period between basal and acetazolamide exposure.

FIGURE 3.

Illustration of the comparisons between altered and healthy territories in basal condition (preoperatively and postoperatively) and exposed to acetazolamide (preoperatively and postoperatively). Idiopathic moyamoya group results are shown in the upper part, and sMM group results are shown in the lower part of the figure. Each dot is representing a vascular territory. P-values *P < .05; **P < .01; ***P < .005 after false discovery rate corrections are displayed.

DISCUSSION

Key Results

Cerebrovascular reserve restoration in moyamoya relies on direct (bypass) and indirect (EDAMS) revascularization techniques.^2,11 However, assessing cerebrovascular reserve using ^99mTc-HMPAO SPECT remains challenging because of examiner dependency.^3,5,17 Our semiautomated method as others^6,18 has shown high specificity while reducing observer bias. Despite limitations such as small sample sizes and ethical concerns regarding radiation exposure, further validation and integration with advanced imaging techniques could enhance patient selection in moyamoya.²

Limitations

This is a retrospective analysis of prospectively collected data from single-center tertiary neuroscience. The authors concede that their method, which holds promise for automation, requires validation through a larger prospective randomized study. Our study is limited by its retrospective nature and the small sample size, attributable to the rarity of the pathology, particularly in Caucasian populations. Moreover, the necessity to acquire a normal baseline cohort rises ethics consideration because of the irradiation nature of this examination. Indeed, even if considered normal, this examination should always be prescribed in patients with impaired cerebral function not allowing us to be certain of its normality. However, its strength lies in the rigorous and reproducible methodology applied. A current limitation of the method is its reliance on the availability of a 3D T1-weighted MRI. As a future perspective, we aim to develop an alternative registration approach that would allow atlas alignment directly onto SPECT images, thereby eliminating the need for anatomic MRI. The impact of this low-spatial-resolution examination, when combined with 4D ASL MRI¹⁹ performed on the same day—both with²⁰ and without acetazolamide—could be significant in optimizing the selection of the donor vessel (superficial temporal artery or occipital artery) and determining the most suitable recipient site. This examination could be directly compared with standard perfusion sequences acquired during the same imaging session, further refining its clinical utility.

Interpretation

99mTc-HMPAO SPECT in Cerebrovascular Reserve Alteration Interpretation and Automated Analysis

The ^99mTc-HMPAO SPECT is not without limitations.^1,21 The relatively low spatial resolution of SPECT and susceptibility to artifacts present challenges in clinical interpretation.^2,3,5,6,21 Furthermore, the procedure's reliance on ionizing radiation and the logistics of administering acetazolamide add additional layers of complexity.²² These factors necessitate a careful consideration of the trade-offs when selecting ^99mTc-HMPAO SPECT with acetazolamide^3-6 over alternative modalities such as perfusion MRI,²⁰ perfusion CT,²³ or positron emission tomography (PET).^6,17,24The visual interpretation of cerebrovascular reserve using ^99mTc-HMPAO SPECT is challenging and often dependent on the examiner's expertise^2,6,21 (Figure 4). However, it is crucial to identify patients with impaired cerebrovascular reserve to ensure they do not miss the therapeutic window during which a bypass could prevent a recurrent stroke.^6-8,11,17

FIGURE 4.

Interpretation difficulty of ^99mTc-HMPAO SPECT. A and B, Patient 1: Illustration of impaired CVR. A, Preoperative assessment: Baseline SPECT imaging (left triptych) demonstrates preserved perfusion. By contrast, Diamox-challenged imaging (right triptych) reveals a significant CVR impairment, with marked hyporesponsiveness in the left parietal and posterior junctional MCA territories, appearing predominantly in yellow. Notably, this perfusion deficit extends anteriorly towards the frontoinsular region (not shown). B, Postoperative assessment after bypass: Baseline perfusion remains preserved (left triptych), with clear visualization of the left parietal subcutaneous bypass. Postoperative Diamox-challenged imaging (right triptych) demonstrates restored cerebrovascular reactivity, with homogeneous and sustained hyperemic response (red signal) across previously impaired territories. C and D, Patient 2: Subtler, more limited CVR impairment. C, Preoperative assessment: Baseline perfusion (left triptych) demonstrates mild asymmetry, with reduced signal intensity and heterogeneous red coloration in the left frontal MCA territory compared with the contralateral side. Anterior junctional regions show focal yellow signals, consistent with borderline hypoperfusion. Diamox-challenged images exhibit increased heterogeneity and extension of the perfusion deficit, although less pronounced and partially equivocal. D, Postoperative assessment after EDAMS: Baseline perfusion normalizes (left triptych), with restored symmetrical hyperemia. However, Diamox-challenged imaging (right triptych) continues to demonstrate impaired CVR in previously affected regions, now more evident when compared with normal baseline perfusion. CVR, cerebrovascular reserve; EDAMS, encephalo-duro-arterio-myo-synangiosis; MCA, middle cerebral artery; SPECT, single-photon emission computed tomography.

Some have used the classic region of interest method, in which ROIs are drawn manually, as a way to solve this limitation.^6,7 But this method has shown significant intraobserver/interobserver variations. Despite the limitations posed by our small sample size (because of the rarity of the disease in Europe), our proposed method achieves a specificity of over 90% in patients with moyamoya, while remaining independent of examiner bias. Indeed, our semiautomated method confirms the possibility of automated, reader-independent quantification of cerebrovascular reserve, similar to the human placement of ROIs in quantitative SPECT^6,7 or perfusion MRI analysis,²⁰ which have been described. We see this additional diagnostic tool as an aid for nuclear medicine physicians to assist neurologists and neurosurgeons in selecting suitable candidates for bypass surgery.

Considering the interindividual variations in cerebrovascular anatomy and the highly variable development of collateral circulation—particularly in moyamoya—a deep learning–based algorithm and its associated biomarker could serve as a valuable clinical decision-support tool, aiding physicians in detecting and assessing subtle changes in baseline perfusion and vascular reserve at diagnosis and during follow-up.²¹ Further refinement of this method could reduce the rate of false negatives and thereby lower the number of patients referred to the limited number of centers equipped with a cyclotron, where a ¹⁵O-water PET scan—offering superior spatial resolution—can be performed.¹⁷ Indeed, some studies suggest that ¹⁵O-water PET²⁴ can help detect patients with impaired cerebrovascular reserve who present with a negative ^99mTcHMPAO scan but remain symptomatic.^17,24 Nonetheless, regarding the resolution aspect, the authors must concede newer SPECT ring detectors are now available. In theory, they are capable of acquiring faster and higher-quality tomoscintigraphy. This new technology has not yet achieved broad acceptance within the nuclear imaging community, and its availability remains limited in our country. To our knowledge, no acetazolamide challenge SPECT studies using this new equipment have been published. Finally, in negative imaging cases, one case report²⁵ raised the question of pharmacological-induced vasodilation importance compared with a physiological provocation test, which seems to achieve better clinical evaluation.

Cerebrovascular Reserve Restauration in Case of Moyamoya

The direct revascularization (bypass) and indirect revascularization (EDAMS) techniques, which help restore an impaired cerebral vascular reserve, are used in the treatment of moyamoya.^{2,6,7,9-11,19,26} There is a growing trend toward combining these revascularization methods in adult patients.^9-11 Our semiautomated analysis method supports the effectiveness of these treatments in restoring progressively cerebral vascular reserve. Like Lee et al,² we show that improvement in perfusion and perfusion reserve in the matched territories was objectively observed, and residual abnormality in the perfusion reserve, despite the normalized basal perfusion, was also observed. In this regard, only one patient in the iMM group required a bilateral procedure. This finding suggests the possibility of improving overall bilateral CVR in patients with moyamoya disease, whose cerebral hemispheres are both affected but at different stages of severity. This is likely due to the increased load on the communicating arteries within the circle of Willis, which have an adaptation capacity. Indirect evidence of this is provided by the possible enlargement of the donor vessel overtime and its increasing contribution to both local and more distant CVR amelioration and then restoration. The restoration is much faster in the sMM group compared with the iMM group. Some authors proposed a mathematical model enabling the calculation of CVR while remaining unaffected by variations in the time to reach maximum cerebral blood flow response.²⁶ Their approach could be a more detailed characterization of CVR compared with conventional static measurements. If confirmed, it could serve as an objective aid in the interpretation of a test that is often highly dependent on the individual reader, which remains its biggest criticism.²² Its usefulness is reinforced by its broader potential in assessing cerebrovascular reserve in conditions such as vascular dementia and Alzheimer's disease,⁴ where this examination was more commonly performed.

Generalizability

Despite its limitations, our study could serve as a precursor to a comparative analysis using a 4D ASL MRI perfusion sequence with and without acetazolamide, the analysis of which will be fully automated. The performance of our proposed method can subsequently be compared against an experienced reader as well as a novice reader (such as a neurosurgeon or neurologist) and become a diagnostic aid.

CONCLUSION

^99mTc-HMPAO is a validated technique for assessing cerebrovascular reserve, enabling the identification of patients with impaired reserve, such as those with moyamoya, and facilitating their referral for revascularization surgery while monitoring outcomes over time. However, its visual interpretation is often reliant on the examiner's experience. Our automated method achieves above 90% specificity rate in MM. Our method is independent of examiner variability and can be used in routine practice to guide selection of good surgical candidates and to evaluate surgical outcome in terms of cerebrovascular reserve restoration. Further refinement of this technique, combined with advanced imaging modalities such as 4D ASL MRI and 4D angiography, could eventually allow for fully automated quantification of cerebrovascular reserve. These findings will need to be confirmed through prospective multicenter studies with larger patient cohorts.

Funding

This study did not receive any funding or financial support.

Disclosures

The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.