Deep Learning Applications for Acute Stroke Management

Brain imaging is essential to the clinical care of patients with stroke, a leading cause of disability and death worldwide. Whereas advanced neuroimaging techniques offer opportunities for aiding acute stroke management, several factors, including time delays, inter-clinician variability, and lack of systemic conglomeration of clinical information, hinder their maximal utility. Recent advances in deep machine learning (DL) offer new strategies for harnessing computational medical image analysis to inform decision making in acute stroke. We examine the current state of the field for DL models in stroke triage. First, we provide a brief, clinical practice-focused primer on DL. Next, we examine real-world examples of DL applications in pixel-wise labeling, volumetric lesion segmentation, stroke detection, and prediction of tissue fate postintervention. We evaluate recent deployments of deep neural networks and their ability to automatically select relevant clinical features for acute decision making, reduce inter-rater variability, and boost reliability in rapid neuroimaging assessments, and integrate neuroimaging with electronic medical record (EMR) data in order to support clinicians in routine and triage stroke management. Ultimately, we aim to provide a framework for critically evaluating existing automated approaches, thus equipping clinicians with the ability to understand and potentially apply DL approaches in order to address challenges in clinical practice. ANN NEUROL 2022.

Portable, low-field magnetic resonance imaging enables highly accessible and dynamic bedside evaluation of ischemic stroke

Brain imaging is essential to the clinical management of patients with ischemic stroke. Timely and accessible neuroimaging, however, can be limited in clinical stroke pathways. Here, portable magnetic resonance imaging (pMRI) acquired at very low magnetic field strength (0.064 T) is used to obtain actionable bedside neuroimaging for 50 confirmed patients with ischemic stroke. Low-field pMRI detected infarcts in 45 (90%) patients across cortical, subcortical, and cerebellar structures. Lesions as small as 4 mm were captured. Infarcts appeared as hyperintense regions on T2-weighted, fluid-attenuated inversion recovery and diffusion-weighted imaging sequences. Stroke volume measurements were consistent across pMRI sequences and between low-field pMRI and conventional high-field MRI studies. Low-field pMRI stroke volumes significantly correlated with stroke severity and functional outcome at discharge. These results validate the use of low-field pMRI to obtain clinically useful imaging of stroke, setting the stage for use in resource-limited environments.

Abstract WP111: Deployment Of Portable, Bedside, Low-field Magnetic Resonance Imaging In The Emergency Department To Evaluate Patients With Acute Stroke

Background and Aims: MRI is critical for diagnosing acute stroke and guiding candidate selection for potential reperfusion therapy. However, rapid stroke evaluation using MRI is often dissuaded by the time required for patients to travel to access-controlled, high-field (1.5-3T) systems. Advances in low-field MRI enable the acquisition of clinically valuable images at the bedside. We report neuroimaging in patients presenting to the Emergency Department (ED) with stroke symptoms using a low-field portable MRI (pMRI) device.

Methods: A 64mT pMRI device was deployed in the Yale-New Haven Hospital ED from August 2020 to July 2021. Patients presenting as a “Stroke Code” or “Intracranial Hemorrhage Alert” with no MRI contraindications were scanned. Exams were performed at the bedside, in the vicinity of ED room equipment. Research staff acquired imaging via tablet, with images available immediately after acquisition. Sequences obtained and axial scan times (in minutes) included T1-weighted imaging (4:54), T2-weighted imaging (7:03), fluid-attenuated inversion recovery imaging (9:31), and diffusion-weighed imaging with apparent diffusion coefficient mapping (9:04). Patients’ demographic information, hours from the time of patients' last known normal (LKN) to time of scan, and discharge diagnoses (determined from final imaging interpretation) were assessed.

Results: pMRI exams were obtained on 54 patients (28 females, 51.9%; median age 71 years, 20-98 years). Discharge diagnoses included ischemic stroke (42.6%) no intracranial abnormality (31.5%), intraparenchymal hemorrhage (7.4%), atherosclerosis (7.4%), tumor (5.6%), subdural hematoma (3.7%), and intraventricular hemorrhage (1.9%). Patient LKN times ranged from 2 to 144 hours (median of 12 hours; 3 patients with no LKN excluded). The pMRI did not interfere with ED equipment and no significant adverse events occurred.

Conclusion: We report the use of a pMRI for bedside neuroimaging in the ED. This approach suggests that pMRI may be viable for supporting rapid diagnosis and treatment candidate selection in patients presenting with stroke symptoms to the ED.

Bedside detection of intracranial midline shift using portable magnetic resonance imaging

Neuroimaging is crucial for assessing mass effect in brain-injured patients. Transport to an imaging suite, however, is challenging for critically ill patients. We evaluated the use of a low magnetic field, portable MRI (pMRI) for assessing midline shift (MLS). In this observational study, 0.064 T pMRI exams were performed on stroke patients admitted to the neuroscience intensive care unit at Yale New Haven Hospital. Dichotomous (present or absent) and continuous MLS measurements were obtained on pMRI exams and locally available and accessible standard-of-care imaging exams (CT or MRI). We evaluated the agreement between pMRI and standard-of-care measurements. Additionally, we assessed the relationship between pMRI-based MLS and functional outcome (modified Rankin Scale). A total of 102 patients were included in the final study (48 ischemic stroke; 54 intracranial hemorrhage). There was significant concordance between pMRI and standard-of-care measurements (dichotomous, κ = 0.87; continuous, ICC = 0.94). Low-field pMRI identified MLS with a sensitivity of 0.93 and specificity of 0.96. Moreover, pMRI MLS assessments predicted poor clinical outcome at discharge (dichotomous: adjusted OR 7.98, 95% CI 2.07–40.04, p = 0.005; continuous: adjusted OR 1.59, 95% CI 1.11–2.49, p = 0.021). Low-field pMRI may serve as a valuable bedside tool for detecting mass effect.

Methodology for Low-Field, Portable Magnetic Resonance Neuroimaging at the Bedside

Neuroimaging is a critical component of triage and treatment for patients who present with neuropathology. Magnetic resonance imaging and non-contrast computed tomography are the gold standard for diagnosis and prognostication of patients with acute brain injuries. However, these modalities require intra-hospital transport to strict, access-controlled environments, which puts critically ill patients at risk for complications and secondary injuries. A novel, portable MRI (pMRI) device that can be deployed at the patient's bedside provides a needed solution. In a dual-center investigation, Yale New Haven Hospital has obtained regular neuroimaging on patients using the pMRI as part of routine clinical care in the Emergency Department and Intensive Care Unit (ICU) since August of 2020. Massachusetts General Hospital has begun using pMRI in the Neuroscience Intensive Care Unit since January 2021. This technology has expanded the population of patients who can receive MRI imaging by increasing accessibility and timeliness for scan completion by eliminating the need for transport and increasing the potential for serial monitoring. Here we describe our methods for screening, coordinating, and executing pMRI exams and provide further detail on how to scan specific patient populations.

Assessment of Brain Injury Using Portable, Low-Field Magnetic Resonance Imaging at the Bedside of Critically Ill Patients

IMPORTANCE

Neuroimaging is a key step in the clinical evaluation of brain injury. Conventional magnetic resonance imaging (MRI) systems operate at high-strength magnetic fields (1.5-3 T) that require strict, access-controlled environments. Limited access to timely neuroimaging remains a key structural barrier to effectively monitor the occurrence and progression of neurological injury in intensive care settings. Recent advances in low-field MRI technology have allowed for the acquisition of clinically meaningful imaging outside of radiology suites and in the presence of ferromagnetic materials at the bedside.

OBJECTIVE

To perform an assessment of brain injury in critically ill patients in intensive care unit settings, using a portable, low-field MRI device at the bedside.

DESIGN, SETTING, AND PARTICIPANTS

This was a prospective, single-center cohort study of 50 patients admitted to the neuroscience or coronavirus disease 2019 (COVID-19) intensive care units at Yale New Haven Hospital in New Haven, Connecticut, from October 30, 2019, to May 20, 2020. Patients were eligible if they presented with neurological injury or alteration, no contraindications for conventional MRI, and a body habitus not exceeding the scanner's 30-cm vertical opening. Diagnosis of COVID-19 was determined by positive severe acute respiratory syndrome coronavirus 2 polymerase chain reaction nasopharyngeal swab result.

EXPOSURES

Portable MRI in an intensive care unit room.

MAIN OUTCOMES AND MEASURES

Demographic, clinical, radiological, and treatment data were collected and analyzed. Brain imaging findings are described.

RESULTS

Point-of-care MRI examinations were performed on 50 patients (16 women [32%]; mean [SD] age, 59 [12] years [range, 20-89 years]). Patients presented with ischemic stroke (n = 9), hemorrhagic stroke (n = 12), subarachnoid hemorrhage (n = 2), traumatic brain injury (n = 3), brain tumor (n = 4), and COVID-19 with altered mental status (n = 20). Examinations were acquired at a median of 5 (range, 0-37) days after intensive care unit admission. Diagnostic-grade T1-weighted, T2-weighted, T2 fluid-attenuated inversion recovery, and diffusion-weighted imaging sequences were obtained for 37, 48, 45, and 32 patients, respectively. Neuroimaging findings were detected in 29 of 30 patients who did not have COVID-19 (97%), and 8 of 20 patients with COVID-19 (40%) demonstrated abnormalities. There were no adverse events or complications during deployment of the portable MRI or scanning in an intensive care unit room.

CONCLUSIONS AND RELEVANCE

This single-center series of patients with critical illness in an intensive care setting demonstrated the feasibility of low-field, portable MRI. These findings demonstrate the potential role of portable MRI to obtain neuroimaging in complex clinical care settings.

Abstract WP413: Portable, Bedside, Point of Care Magnetic Resonance Imaging in an Intensive Care Setting for Intracranial Hemorrhage

Background: Radiographic diagnosis of intracranial hemorrhage (ICH) is a critical determinant of stroke care pathways requiring patient transport to a neuroimaging suite. Advances in low-field MRI have made it possible to obtain clinically useful imaging at the point of care (POC).

Aim: The aim of this study was to obtain preliminary data regarding the ability of a bedside POC MRI scanner to detect ICH.

Methods: We studied 36 patients with a diagnosis of ICH (n=18) or ischemic stroke (n=18). Five blinded readers independently evaluated T2W and FLAIR exams acquired prospectively on a 64 mT, portable bedside MRI system (Hyperfine Research, Inc). Kappa coefficients (κ) were calculated to determine inter-rater agreement. Ground truth was obtained from the clinical report of the closest conventional imaging study (17.9 ± 10.4 hours) and verified by a core reader. For each exam, majority consensus among raters was used to determine sensitivity.

Results: ICH volume ranged from 4 to 101 cc (median of 13 cc). Exams were acquired within 7 days of symptom onset (51.1 ± 28.8 hours). A pathologic lesion was identified on every exam with 100% sensitivity. Sensitivity for distinguishing any hemorrhage was 89% and specificity was 83%. The mean sensitivity and specificity for individual raters was 79% and 69%, respectively. When limited to supratentorial hemorrhage, consensus sensitivity was 94%. For ICH cases detected by all raters (n=9), there was 100% accuracy for localizing the bleed (lobar vs. non-lobar) with perfect agreement among raters (κ = 1, p <0.0001). There was substantial agreement for identifying intraventricular hemorrhage (IVH) (κ = 0.72, p < 0.0001). Sensitivity for IVH was 100% based on rater consensus. Figure 1 shows a POC exam with an ICH and IVH.

Conclusions: These data suggest that low-field, POC MRI may be used to detect hemorrhagic stroke at the bedside. Further work is needed to evaluate this approach in the hyperacute setting and across a wide range of ICH characteristics.

Abstract 57: Deployment of Portable, Bedside, Low-Field Magnetic Resonance Imaging for Evaluation of Stroke Patients

Background: Magnetic resonance imaging (MRI) is a powerful modality for diagnosing stroke. Conventionally, patients must travel to the location of a high-field MRI device. Advances in low-field MRI have enabled acquisition of clinically useful images using a portable device at the bedside. The feasibility of using point of care (POC) MRI in a clinical stroke setting is unknown.

Objective: To determine the safety and feasibility of portable, bedside, low-field MRI in a clinical setting.

Design/Methods: POC MRI exams were performed in Yale’s Neuroscience Intensive Care Unit (NICU) from July 2018 to August 2019. Images were acquired at the bedside using a standard 110V, 15A power outlet. The environment included the bedside vitals monitor, ventilators and intravenous infusion pumps. Exams were performed by research staff trained to operate the POC scanner in the absence of a trained MRI technician. No special precautions were necessary to remove ferrous metals from the room. Scan parameters were controlled using a tablet computer interface, and images were available immediately after acquisition.

Results: POC MRI was obtained in 85 stroke cases (46% female, ages 18-96 years, 46% ischemic stroke, 34% intracerebral hemorrhage, 20% subarachnoid hemorrhage). Scans were obtained within 7 days of symptom onset. NIHSS scores ranged from 1 to 29 (median of 7). Of the 85 patients analyzed, 68 underwent T2-weighted imaging, 72 underwent FLAIR imaging, and 39 underwent diffusion weighted imaging (DWI). DWI was only tested in ischemic stroke cases. Patients’ BMI ranged from 20.0 to 46.5 with a median of 26.7. The majority 74 (87%) of patients completed the entire exam. Five patients (6%) were unable to fit in the scanner’s 30 cm opening, while 6 patients (7%) experienced claustrophobia resulting in early termination of the exam. Mean exam time was 28.9 ± 8.4 minutes. The 64 mT static magnetic field, gradient and RF pulses of the POC MRI scanner did not interfere with NICU equipment, and no significant adverse events occurred.

Conclusions: We report the first use of a portable, low-field MRI system to image stroke patients at the bedside. This early work suggests our approach is safe and viable in a complex clinical care environment.