Portable Ultra‐Low‐Field MRI in Outpatient Neurology: An Examination of Clinical Performance and Patient Experience

ABSTRACT

Background and Purpose

Brain magnetic resonance imaging (MRI) is an essential component for outpatient neurological evaluation, though access to timely imaging at the point of care is not feasible for most outpatient neurology practices. Portable ultra‐low‐field MRI systems offer a potential solution, but studies describing their clinical performance in routine outpatient neurology practice remain limited. This study evaluated clinical concordance and patient experience of portable MRI (pMRI) compared to standard‐of‐care MRI (sMRI) in independent neurology practices.

Methods

In this prospective, multicenter study, adults presenting to outpatient neurology clinics and requiring clinically indicated brain MRI completed imaging with both pMRI (0.064 T) and sMRI (91% 3 T, 9% 1.5 T) in the neurology clinic. All examinations were independently reviewed by board‐certified neuroradiologists in a blinded, randomized fashion without access to clinical history. The primary endpoint was patient‐level concordance between modalities for the presence or absence of abnormal findings. Discordant cases underwent post hoc unblinded paired review with clinical history to assess clinical significance. Secondary endpoints included patient‐reported experience, assessed using a structured questionnaire evaluating noise, comfort, claustrophobia, anxiety, and overall experience on 10‐point Likert scales, as well as modality preference.

Results

Among 125 participants imaged for common outpatient indications, including headache (40%), cognitive impairment or dementia (16%), multiple sclerosis follow‐up (16%), and tumor surveillance (16%), portable and sMRI demonstrated 92% concordance on blinded review. Following clinically informed post hoc review, concordance increased to 98%. pMRI was rated more favorably across all patient experience domains (p < 0.001), and participants preferred pMRI (61%) over sMRI (14%) by a 4:1 ratio.

Conclusions

In outpatient neurology practices, pMRI images demonstrated high clinical concordance with sMRI for identifying the presence or absence of structural brain abnormalities and were strongly preferred by patients. These findings support the use of pMRI as a practical point‐of‐care imaging tool for neurology patients, enabling timely access to structural neuroimaging during the clinical encounter.

1. Introduction

Magnetic resonance imaging (MRI) is indispensable in outpatient neurology, serving as the primary diagnostic tool for evaluating headaches, cognitive decline, dizziness, and monitoring known lesions [1, 2, 3, 4]. Despite this reliance on imaging, few independent neurology practices have instant access to MRI. Unlike cardiology, where office‐based imaging is routine, most neurologists must refer patients to hospitals or imaging centers for MRI. After waiting more than a month for an initial neurology visit, patients may then wait days to weeks for imaging, fragmenting the diagnostic process and delaying clinical decision‐making [5, 6]. Approximately half of outpatient MRI studies are completed later than intended, with average waiting time exceeding 18 days [7]. For patients requiring serial imaging, such as those with multiple sclerosis, additional barriers such as transportation or scheduling conflicts and cost frequently lead to missed or rescheduled examinations or even delayed treatment [8]. Neurologists assume both clinical responsibility and medico‑financial liability for diagnostic and management decisions, without access to neuroimaging at the point of care.

The separation of imaging from outpatient clinical care contrasts with inpatient practice, where neurologists routinely review neuroimaging in time‐sensitive settings such as stroke, where treatment decisions depend on immediate image review [9]. Unlike cardiology, which has numerous accredited cardiovascular imaging fellowship programs, neurology has not developed standardized training pathways in imaging, and dedicated neuroimaging fellowships remain limited [9]. As a result, outpatient neurology practices depend heavily on imaging to guide clinical decisions yet have little control over imaging access, timing, or workflow.

The limited use of office‐based brain imaging reflects practical economic and operational realities. Conventional MRI systems require multimillion‐dollar capital investment, radiofrequency (RF)‐shielded rooms, substantial physical space, specialized cooling infrastructure, service contracts, and MRI technologists with specialized training [5, 10, 11]. For most independent neurology practices, the upfront costs and imaging volume required to break even make on‐site neuroimaging economically unviable.

Portable ultra‐low‐field MRI (0.064 T), such as the Hyperfine Swoop System used in this study (Figure 1), addresses several of these barriers and offers a more feasible path to in‐office neuroimaging. These systems cost substantially less than conventional MRI, run on standard electrical power, do not require RF shielding, and can be operated by existing clinical staff with minimal training [5, 10, 11, 12, 13]. Advances in AI‐based image reconstruction have further improved image quality by reducing noise and artifacts, narrowing the gap between ultra‐low‐field and conventional MRI [14, 15, 16]. The ultra‐low‐field strength also reduces projectile risk, minimizes susceptibility artifacts, and allows imaging of select patients with metallic implants that may be contraindicated at higher field strengths [16, 17].

FIGURE 1.

Hyperfine Swoop portable MR imaging system operated within a standard outpatient exam room. The figure demonstrates positioning of a subject for a brain scan and highlights the system's ability to provide point‐of‐care neuroimaging utilizing standard exam room without the need for shielding or specialized power infrastructure. Image courtesy of Senolytix West Palm Beach, Florida. Used with permission.

Prior hospital‐based studies demonstrated that portable MRI (pMRI) can detect intracerebral hemorrhage, midline shift, hydrocephalus, and ischemic stroke lesions as small as 2.8 mm [18, 19, 20, 21]. Subsequent work has extended to chronic neurologic conditions, including multiple sclerosis [12, 22] and dementia [23, 24]. Additional studies have demonstrated feasibility in community and rural environments [25, 26, 27], with high patient acceptance in pediatric and adult populations [7, 28, 29]. Studies have shown the integration and adoption of pMRI in the inpatient hospital setting [30], significantly reduce turnaround time by 83% (from 22 to 3.7 h), simplify clinical workflows, and demonstrate cost savings [31]. However, these investigations have focused primarily on feasibility, image quality, or disease‐specific detection. The present study builds on this foundation by evaluating pMRI across the mix of conditions seen in routine outpatient neurology practices compared to same‐day high‐field‐strength MR.

To address this gap, we conducted a prospective, multicenter study comparing pMRI (0.064 T) with standard‐of‐care MRI (sMRI) (1.5 or 3 T) in patients undergoing clinically indicated brain imaging at two outpatient neurology practices. We evaluated whether pMRI could provide clinically meaningful structural information sufficient to support routine outpatient neurologic decision‑making, and how patients rated this experience as compared to sMRI.

2. Methods

2.1. Study Design

The NEURO PMR study was a prospective, multicenter study conducted under a single protocol. The study was approved by the WCG IRB (Western Copernicus Group Institutional Review Board), and all participants provided written informed consent prior to enrollment and completion of any study‐related activities. Two independent neurology practices with an established MRI service participated.

All participants underwent both pMRI and sMRI, with pMRI performed prior to sMRI on the same day. Imaging studies were de‐identified at each site and transferred in DICOM format to a HIPAA‐compliant cloud PACS for centralized review. Participants filled out a paper questionnaire after completing both scans. Demographic information, baseline characteristics, patient experience, and all non‐imaging study data were collected on source case report forms and entered into a validated electronic data capture system (Medrio Inc., San Francisco, CA) in accordance with study procedures.

2.2. Participants

Eligible participants were adults (≥18 years) scheduled to undergo brain MRI (CPT 70551 or 70553, pre‐contrast). Participants were excluded if they had MRI safety contraindications (including pacemakers/defibrillators or metallic foreign bodies in the brain or eye), if their body weight exceeded 200 kg, if they were unable to lie still or be appropriately positioned for either examination, or if they were not cleared for sMRI on the basis of routine MR safety screening.

To reflect common outpatient neurologic indications for brain MRI, recruitment targeted at least 20 participants in each of four clinical categories: (1) headache or migraine, (2) mild cognitive impairment or dementia, (3) multiple sclerosis or other demyelinating disease, and (4) tumors or neoplasms. Cases were selected on the basis of the treating neurologist's clinical evaluation and appropriate use of neuroimaging.

2.3. Image Acquisition

pMRI examinations were acquired using the Swoop Portable MR Imaging System (0.064 T; Model 2). sMRI examinations were acquired on each site's clinically available scanners, with 91% acquired at 3 T (Philips Ingenia Elition X or GE DISCOVERY MR750w) and 9% acquired at 1.5 T (GE SIGNA Voyager or GE Signa HDxt).

pMRI scan protocol included the following sequences: sagittal T1‐weighted imaging, axial T1‐weighted gray–white matter imaging, axial T2‐weighted imaging, axial fluid‐attenuated inversion recovery (FLAIR), and diffusion‐weighted imaging (DWI). sMRI examinations were required to include corresponding non‐contrast sequences for comparison. Beyond these core sequences, sMRI protocols were not harmonized across sites, reflecting routine clinical practice rather than optimized research acquisition.

Due to its ultra‐low‐field magnet strength, pMRI offers a more limited sequence set and lower spatial resolution than high‐field systems. Susceptibility‐weighted imaging (SWI) is not practical at 0.064 T, as susceptibility contrast depends strongly on magnetic field intensity. Conversely, lower field strength reduces susceptibility‐related artifacts near metallic implants, which may improve visualization in selected clinical scenarios [30].

2.4. Image Review

Two board‐certified neuroradiologists with more than 25 years of combined experience independently reviewed all imaging studies in a blinded, randomized fashion. Reviewers were provided the clinical indication for each examination (e.g., headache, dementia) but were not provided detailed clinical history or information regarding imaging field strength. Images were presented in randomized order, and reviewers did not have access to the paired pMRI and sMRI images. Each examination was assessed for the presence or absence of structural brain abnormality and for specific pathology findings.

Each examination was classified as normal or abnormal on the basis of the presence of any intracranial structural abnormality identified by the interpreting neuroradiologist. Specific pathology findings were also recorded. When readers disagreed, a third independent neuroradiologist with 15 years of experience adjudicated the final interpretation.

The primary endpoint was patient‐level concordance between pMRI and sMRI for the presence or absence of abnormal findings. Cases classified as discordant on blinded review underwent post hoc unblinded paired review with clinical history by a neuroimaging neurologist, and neuroradiologist with over 60 years of combined experience. This post hoc analysis assessed whether discrepancies reflected clinically meaningful differences or interpretive variations that would not alter clinical management.

2.5. Patient Experience Assessment

Patient experience was assessed using a structured questionnaire administered after completion of both imaging examinations. The 10‐point Likert scale was used to gauge the patient's experience in both standard MRI machines and the Hyperfine Swoop System. Participants rated overall experience, perceived noise, comfort, claustrophobia, and anxiety using 10‐point Likert scales, with higher scores indicating a more favorable experience and lower scores indicating a poor experience. The mean, standard deviation, and p values were recorded for each domain and are presented in Table 3, Table S1. Participants also indicated their preferred modality for future imaging. Responses were entered into the electronic data capture system. Mean scores and standard deviations were calculated for each domain. Paired t‐tests compared pMRI and sMRI ratings, and modality preference was summarized descriptively.

TABLE 3.

Patient‐reported experience scores for portable magnetic resonance imaging (pMRI) vs. standard‐of‐care magnetic resonance imaging (sMRI).

Patient question	pMRI Mean (SD)	sMRI Mean (SD)	p value
OVERALL EXPERIENCE: 1 (very poor experience) to 10 (great experience)	8.72 (±1.62)	7.35 (±2.17)	<0.0001
How much NOISE you experienced: 1 (very noisy) to 10 (not at all noisy)	7.29 (±2.23)	3.87 (±3.04)	<0.0001
How COMFORTABLE you were: 1 (very uncomfortable) to 10 (very comfortable)	7.85 (±2.12)	6.86 (±2.31)	<0.0001
How CLAUSTROPHOBIC you felt: 1 (very claustrophobic) to 10 (not at all claustrophobic)	8.52 (±2.40)	6.46 (±3.24)	<0.0001
Your level of ANXIETY: 1 (very anxious) to 10 (not at all anxious)	8.65 (±2.26)	7.22 (±3.02)	<0.0001

Note: Mean ± standard deviation of patient‐reported experience scores across evaluated domains, recorded on a 10‐point Likert scale. Higher scores indicate a more favorable experience.

2.6. Statistical Analysis

Agreement between modalities was summarized using patient‐level percent concordance, consistent with clinical classification used in routine neurologic decision‐making. The study was not designed to evaluate sensitivity or specificity for individual lesion detection. The primary endpoint analysis was performed using SAS version 9.4. The secondary endpoint and all other statistical analyses were performed in Python (3.14) using the NumPy (2.4.0), Pandas (2.3.3), SciPy (1.16.3), and Matplotlib (3.10.8) libraries. A two‐sided p value of 0.05 was considered statistically significant.

3. Results

3.1. Participant Characteristics

A total of 131 participants were enrolled between April and September 2025. Of these, 125 participants completed all study procedures and were included in the per‐protocol analysis. Six participants were excluded due to incomplete or incorrect imaging examinations.

The per‐protocol cohort had a mean age of 54.9 ± 17.3 years (range, 19–85 years), with 84 females (67%) and 41 males (33%). Participants were imaged for common outpatient neurologic indications, including headache, cognitive impairment, multiple sclerosis follow‐up, and tumor surveillance. Participant characteristics are summarized in Table 1.

TABLE 1.

Participant characteristics (per‐protocol analysis cohort).

Category	Total (N = 125)
Age (years) a
Mean (SD)	54.86 (17.31)
Min, Max	19.0, 85.0
Sex, n (%)
Male	41 (32.8)
Female	84 (67.2)
Reason for MRI exam, n (%)
Headache/Migraine	50 (40)
Cognitive impairment/Dementia	20 (16)
Multiple sclerosis/Demyelinating disease	20 (16)
Tumor/Neoplasm	20 (16)
Other	15 (12)
sMRI field strength, n (%)
1.5 T	11 (8.8)
3 T	114 (91.2)

Abbreviations: MRI, magnetic resonance imaging; SD, standard deviation; sMRI, standard‐of‐care MRI.

^aAge is calculated as (date of screening − date of birth) + 1/365.25.

3.2. Primary Endpoint: Patient‐Level Concordance for Presence/Absence of Abnormal Finding

On blinded neuroradiological review, pMRI and sMRI demonstrated patient‐level concordance for classification as normal or abnormal in 115 of 125 cases (92%), with 10 cases (8%) classified as discordant. Post hoc unblinded, paired review with clinical history reclassified 8 of these 10 cases as concordant, resulting in concordance in 123 of 125 cases (98%) when imaging was interpreted in clinical context. In these eight cases, abnormal findings were present on both modalities but were initially overlooked or interpreted differently on blinded review and became evident when images were reviewed in parallel with relevant clinical information.

Two cases remained discordant after clinically informed review. One involved a 2 mm cavernous malformation with associated developmental venous anomaly identified on sMRI but not visualized on pMRI, consistent with the absence of SWI at ultra‐low‐field. The second involved a subtle subcortical white matter lesion detected on sMRI but not on pMRI, likely due to lower spatial resolution. In both cases, the discordant findings were incidental and did not alter clinical management.

Patient‐level concordance assessed using both blinded radiologic interpretation and post hoc clinically informed review is summarized in Table 2.

TABLE 2.

Patient‐level concordance between portable magnetic resonance imaging (pMRI) and standard‐of‐care magnetic resonance imaging (sMRI) for abnormal findings.

Analysis approach	Concordant cases, n (%)	Discordant cases, n (%)
Blinded radiologic review	115/125 (92)	10/125 (8)
Post hoc clinically informed review	123/125 (98)	2/125 (2)

Note: Blinded radiologic review assessed patient‐level concordance for normal vs. abnormal findings without access to clinical history. Post hoc clinically informed review incorporated paired imaging comparison and relevant clinical context to assess the clinical significance of discordant findings.

3.3. Radiologic Findings by Clinical Indication

Radiologic findings are summarized by clinical indication to characterize the types of structural abnormalities identified on pMRI. When specific findings were compared between pMRI and sMRI, a high level of agreement was observed across clinical indications, with emphasis on findings relevant to routine clinical decision‑making.

3.3.1. Headache/Migraine

Among patients imaged for headache or migraine (n = 50), most examinations were either normal or demonstrated incidental findings typical of outpatient headache populations [32]. The most common abnormalities identified on pMRI included mild white matter changes involving subcortical (22%) or periventricular (16%) regions and paranasal sinus inflammatory disease (24%). Less frequent findings included structural abnormalities such as Chiari malformation and empty sella. The overall distribution and clinical relevance of findings identified on pMRI were similar to those observed in sMRI, supporting its utility for evaluating headache in routine outpatient practice (Figures 2, 3, 4).

FIGURE 2.

A 56‐year‐old male who presented with progressive tunnel vision and papilledema. Lumbar puncture opening pressure was 40 cm H20. A thrombosis was found in the distal superior sagittal sinus. There was evidence of empty sella and prominence of the optic nerve sheath consistent with intracranial hypertension. He has been maintained with anticoagulation and acetazolamide with resolution of headaches and papilledema. pMRI FLAIR axial confirms a “delta sign” consistent with a superior sagittal sinus (SSS) thrombosis; 3 T imaging on the same day confirms the diagnosis. Axial FLAIR (A) and sagittal T1W (B) on a pMRI and an FLAIR axial on an sMRI (C) confirm a posterior sagittal sinus chronic thrombosis. Sagittal T1W sequence is consistent with empty sella (large arrow) and the diagnosis of secondary intracranial hypertension.

FIGURE 3.

Headache patient was found to have a 1 cm extra‐axial left‐sided falx mass consistent with a presumed meningioma. On pMRI, (A) FLAIR, (B) T1 sagittal, and (C) T2 axial sequences show no evidence of mass effect or vasogenic edema.

FIGURE 4.

Bilateral retro‐orbital headaches in 47‐year‐old. On T2 weighted pMRI (A) and sMRI (B), there is evidence of extensive bilateral maxillary sinus “opacification” associated with inflammatory disease within the mastoids (arrow).

3.3.2. Cognitive Impairment/Dementia

Among patients imaged for cognitive concerns (n = 20), pMRI identified the key structural findings relevant to outpatient assessment: cerebral atrophy (60%) and periventricular white matter disease (50%), consistent with sMRI findings (Figure 5).

FIGURE 5.

A 76‐year‐old female with a family history of dementia suffering from mild cognitive impairment on donepezil and memantine. pMRI (A) T2 and (B) FLAIR show significant prominence of the superior frontal sulcus (yellow arrow) and central sulcus (white arrow). Volumetric studies on 3 T show that whole brain is in the <1 normative percentile, and (C) widening of the sulci on a T1W sequence is well seen.

3.3.3. Multiple Sclerosis/Demyelinating Disease

Among patients imaged for multiple sclerosis or other demyelinating disease (n = 20), pMRI identified pericallosal white matter hyperintensity (85%) consistent with demyelination in the majority of abnormal examinations. Subcortical white matter lesions (15%) were also commonly observed. These findings reflected the dominant lesion burden relevant to routine outpatient surveillance and longitudinal assessment, with overall distribution similar to that observed on sMRI (Figure 6).

FIGURE 6.

A 57‐year‐old male with relapsing remitting multiple sclerosis since 1995. Complains of urinary urgency, paresthesias, lower extremity spasticity. Has lesions in the spinal cord also. Presently on a CD20 monoclonal antibody every 6 months. (A) 3 T sagittal MRI shows extensive T2 hyperintensities within the corpus callousum and so‐called Dawson's fingers. The following three images were obtained on pMRI: (B) sagittal and axial (C) confirm extensive periventricular plaques on FLAIR. (D) T1W axial sequence confirms the presence of so‐called black holes.

3.3.4. Tumor

In the tumor and neoplasm surveillance subgroup (n = 20), pMRI identified postoperative changes and residual or recurrent lesions. Both intra‐axial and extra‐axial lesions were visualized, with differences in lesion conspicuity consistent with known field‐strength and sequence limitations (Figures 7, 8, 9).

FIGURE 7.

A 65‐year‐old female after a mild closed head injury and migraines was found to have an incidental T2 hypointense lesion that has not changed in over 5 years, consistent with a calcified tentorial meningioma. pMRI (A) and sMRI (B) show similar findings of a calcified mass without associated with edema on the T2 sequence.

FIGURE 8.

A 56‐year‐old with a diagnosis of glioblastoma diagnosed 9 years ago. No evidence of residual or recurrent disease on axial T2 (A) and FLAIR (B) on pMRI and sMRI (C). Large postoperative cavity (arrow) posterior to post‐treatment encephalomalacia.

FIGURE 9.

A 22‐year‐old female with episodes of left‐sided numbness, including the tongue. A right frontal gyral FLAIR and T2 hyperintensity (arrows) is equally well seen on pMRI and sMRI (3 T). This may represent cortical dysplasia or low‐grade glioma. (A) sMRI FLAIR, (B) sMRI T2, (C) pMRI FLAIR, and (D) pMRI T2.

3.4. Secondary Endpoint: Patient‐Reported Experience

Patient‐reported experience favored pMRI compared with sMRI across all evaluated domains on 10‐point Likert scales (all p < 0.001). The largest differences were observed for perceived noise (7.3 ± 2.2 vs. 3.9 ± 3.0) and claustrophobia (8.5 ± 2.4 vs. 6.5 ± 3.2), representing 3.4‐ and 2.0‐point differences, respectively. Mean scores and standard deviations for each domain are summarized in Table 3. Ratings for pMRI also demonstrated less variability, with narrower score distributions and fewer extreme negative responses, particularly for noise, claustrophobia, and anxiety.

When asked about future imaging preference, participants favored pMRI by a 4:1 margin. Of 125 participants, 76 (61%) preferred pMRI, 18 (14%) preferred sMRI, and 31 (25%) reported no preference.

4. Discussion

This study demonstrates that pMRI can be meaningfully incorporated into routine outpatient neurology workflow, achieving high concordance with sMRI for determining the presence or absence of structural abnormalities (92% on blinded review, 98% with clinical context), while also providing a superior patient experience. Importantly, the pMRI image quality has limitations, such as lower resolution or certain artifacts, that require proper training and experience with the modality. Unlike current outpatient workflows, where imaging is often obtained days to weeks after the clinical encounter [6, 7], pMRI provides access to structural imaging at the point of care. This model parallels cardiology and cardiovascular imaging, where in‐office imaging is routinely incorporated into clinical decision‑making, and highlights the potential for neuroimaging to play a more immediate role in outpatient neurologic care.

In outpatient practice, neurologists assume both clinical responsibility and medico‑financial liability for diagnostic and management decisions, despite limited control over imaging access and timing. Integrating point‑of‑care MRI into the neurology clinic may help align diagnostic responsibility with imaging workflow, reducing delays that contribute to fragmented care and prolonged diagnostic uncertainty.

The clinical performance observed in this study builds on prior hospital‐based evaluations [18, 19, 20, 21, 22, 30] and demonstrates that ultra‐low‐field imaging can capture structural abnormalities most commonly encountered in outpatient neurology. Findings, such as periventricular and subcortical white matter disease, demyelinating lesions, cerebral atrophy, and postoperative neoplasms, were readily identified. The eight cases reclassified from discordant to concordant after unblinded clinical review reflect real‑world outpatient practice, where imaging is interpreted alongside the clinical history, neurologic examination, and prior imaging rather than in isolation. Discrepancies were attributable to expected technical limitations or interpretive variability, and none altered clinical management. This is consistent with outpatient practice, where the central question is not precise lesion characterization, but whether actionable structural pathology is present that is directly related to patient's symptoms and treatment plan. From this perspective, pMRI appears well suited as an initial point‑of‑care imaging modality, allowing neurologists to efficiently triage patients by providing reassurance when findings are normal or expected and facilitating timely referral for high‑field imaging when further characterization is needed.

pMRI is not intended to replace sMRI in all clinical scenarios [33]. The absence of SWI limits assessment of microhemorrhages, including those associated with cerebral amyloid angiopathy or ARIA‑H (amyloid‐related imaging abnormalities‐hemosiderin) in Alzheimer's disease. Similarly, in multiple sclerosis and other demyelinating diseases, the inability to image the spine limits the role of pMRI to brain surveillance (including exclusion of treatment‐related comorbidities such as progressive multifocal leukoencephalopathy) in established disease rather than initial diagnosis.

The efficiency of in‐office neuroimaging is particularly relevant for common presentations such as headache, which accounts for approximately 12.8 million outpatient visits and 1%–4% of emergency room visits [32, 34, 35]. In headache medicine, the goal is to keep patients out of the emergency department and minimize the diagnostic uncertainty of waiting for off‐site imaging [6]. Timely access to structural imaging in the office may both ease patient anxiety and reduce burden on the health system. The benefits of in‐office imaging also extend to patients requiring serial imaging, such as those with multiple sclerosis, postoperative tumors, or other chronic neurologic conditions for whom repeated appointments at external facilities are often difficult to coordinate and can interfere with recommended monitoring intervals [7, 28].

Consistent with prior studies showing high patient acceptance [7, 24, 27, 28], participants, in this study, strongly favored pMRI, with a 4:1 preference over sMRI. Participants rated pMRI more favorably across all experience domains—particularly noise and claustrophobia, both well‑recognized barriers to MRI completion. Given that 1%–15% of patient's experience MRI‑related claustrophobia and up to 3% are unable to complete conventional scans [36, 37], performing imaging in a familiar outpatient setting with an open, quieter pMRI system may improve tolerability and reduce incomplete examinations, especially among older adults and patients with cognitive impairment, given that a caregiver can stand next to the patient and comfort them during the procedure.

pMRI offers practical advantages for office use: safety, affordability, and operational simplicity. Operating at 0.064 T, the system eliminates projectile hazards and requires neither RF shielding nor dedicated MRI technologists, although care must be taken to prevent patients with left ventricular assist devices from approaching the scanner [38]. Its compact footprint allows installation in a typical exam room, and existing clinic staff can operate it after a short training. Brain MRI studies performed on portable systems are reimbursed using the same CPT codes as conventional MRI, supporting a sustainable model for practices seeking to add ancillary services. From a health economics perspective, shifting initial MRI examinations from higher cost hospital‐based settings to independent neurology practices has the potential to reduce health system expenditures and lower out‐of‐pocket costs for patients [37]. With brain MRI demand projected to grow 40% through 2055 [40], pMRI provides a practical and affordable pathway to expand neuroimaging access in community and underserved settings.

This study has several limitations. We assessed patient‐level concordance on the presence or absence of structural abnormalities rather than lesion‐level sensitivity or specificity. pMRI may not detect small lesions within a case still classified as abnormal, which could overestimate concordance at the lesion level; the endpoint was designed to answer whether structural pathology requiring further workup was present, and no discordant case altered clinical management. Although neuroradiologists were blinded to field strength, image characteristics inherent to ultra‐low‐field MRI may have allowed readers to infer modality, which should be considered when interpreting concordance estimates. pMRI examinations were performed without contrast in accordance with current device labeling, although ongoing studies and case reports have shown diagnostically meaningful gadolinium enhancement across multiple pathologies [30, 41]. Patient experience was assessed after same‐day sequential imaging with pMRI performed first. Although the fixed order may have introduced fatigue or novelty effects, the consistency of findings across all domains and alignment with prior studies suggest these influences were minimal. The post hoc clinical review of discordant cases was exploratory and not prespecified. Both participating sites were neurology practices with on‐site MRI, which may limit generalizability to smaller practices. Future studies should evaluate pMRI implementation in community outpatient practices without existing imaging infrastructure.

In summary, pMRI demonstrated high concordance with sMRI for detecting structural brain abnormalities in routine outpatient practice and was strongly preferred by patients. It can serve as a point‐of‐care imaging option for initial evaluation when the primary question is whether structural pathology is present. pMRI makes point‐of‐care structural neuroimaging a practical reality for outpatient neurology workflow. Broadly deployed, it has the potential to reduce diagnostic delays, ease health system burden, and expand access to neuroimaging in community and underserved settings—fundamentally changing how neurological care is delivered.

Funding

Research support and funding for this study were provided by Hyperfine Inc.

Conflicts of Interest

L.M. reports being the director of the Dent Neuroimaging Center and a past president of the American Society of Neuroimaging. He also serves as a paid consultant for Hyperfine Inc. J.D. serves as a paid consultant for Hyperfine Inc. S.A.A. reports no direct disclosures; however, he reports that immediate family members are Hyperfine Inc. shareholders. N.P. serves as a paid consultant for Hyperfine Inc. E.K. is an employee of Hyperfine Inc. and holds stock and options in the company. M.A.N. and G.S. declare no conflicts of interest.

Supporting information

Supporting File 1: Jon70143‐sup‐0001‐SuppMat.docx.docx