Robotic Transcranial Doppler in Non-Traumatic Subarachnoid Hemorrhage: A Safety and Efficacy Study


 Introduction: Delayed cerebral ischemia (DCI) secondary to vasospasm is a determinate of outcomes following non-traumatic subarachnoid hemorrhage (SAH). SAH patients are monitored using transcranial doppler (TCD) to measure cerebral blood flow velocities (CBFv). However, the accuracy and precision of manually acquired TCD can be operator dependent. The NovaGuide robotic TCD system attempts to standardize acquisition. This investigation evaluated the safety and efficacy of the NovaGuide system in SAH patients in a Neuro ICU.Methods: We retrospectively identified 48 NovaGuide scans conducted on SAH patients. Mean and maximum middle cerebral artery (MCA) CBFv were obtained from the NovaGuide and the level of agreement between CBFv and computed tomography angiography (CTA) for vasospasm was determined. Safety of NovaGuide acquisition of CBFv was evaluated based on number of complications with central venous lines (CVL) and external ventricular drains (EVD) Results: There was significant agreement between the NovaGuide and CTA (Cohen’s Kappa = 0.74) when maximum MCA CBFv ≥120 cm/s was the threshold for vasospasm. 27/48 scans were carried out with CVLs and EVDs present without negative outcomes. Conclusions: The lack of adverse events associated with EVDs/CVLs and the strong congruence between maximal MCA CBFv and CTA illustrates the diagnostic utility of the NovaGuide.


INTRODUCTION
Each year the United States has 30,000 cases of subarachnoid hemorrhage (SAH), which is consistent with the global incidence rate and represents a significant cause of stroke-related death and disability. 1 Vasospasm is the narrowing of the cerebral arteries and is a common complication of SAH that can lead to delayed cerebral ischemia (DCI), with DCI leading to poor patient outcomes and is associated with morbidity and mortality following SAH. 2,3 Digital subtraction angiography (DSA), and increasingly computer tomography angiography (CTA) are considered the gold standard for determination of cerebral vasospasm as it allows for clear visualization of vasculature. 4 However, repeated imaging from DSA and CTA does not come without risks especially in patients with renal comorbidities. Transcranial doppler (TCD) is a non-invasive bedside modality to rapidly assess vasospasm or stenosis of cerebral arteries following SAH.
TCD offers a safe, relatively inexpensive, and repeatable alternative to CTA to measure cerebral blood flow velocity (CBFv) by insonating proximal cerebral arteries, thus enabling the identification of vasospasm and ischemia. 5 In addition, there are a plethora of potential utilizations of TCD in the neurocritical care setting, such as traumatic brain injury, non-invasive intracranial pressure (ICP) monitoring, large vessel occlusions, sickle cell disease, cerebral malaria, periprocedural monitoring, and even brain stem death. 6 Furthermore, TCD monitoring is also able to predict microemboli burden and as such has shown efficacy for stroke risk stratification after blunt cerebrovascular injury to internal cerebral arteries. 7 Specific thresholds of mean cerebral blood flow have been defined to create criteria for the diagnosis of vasospasm based on TCD readings. 8 However, manual capture of TCD requires a trained sonographer which limits its availability to specific institutions and results in TCD being highly operator dependent. Technicians often have difficulty in finding adequate acoustic windows for ultrasound, leading to vessels not being properly insonated or unsuccessful monitoring attributable to movement of the probe during the procedure. 9,10 Moreover, studies have demonstrated as much as 22.1 cm/s variability between experienced operators. 11 The NovaGuide (NovaGuide Intelligent Ultrasound, NovaSignal Corp., Los Angeles CA USA) aims to expedite, standardize, and enhance TCD blood flow measurements through combining artificial intelligence and autonomous acquisition that minimizes the need for a trained operator. The NovaGuide system uses a headset that automatically moves the transducer, eliminating variability between users. This allows the system to be used without a trained sonographer while also providing consistent results. We sought to evaluate the safety and validity of the NovaGuide in patients with SAH.

This research was conducted at Westchester Medical Center and was approved by New York Medical College Institutional Review Board and the Clinical Research
Institute at Westchester Medical Center. All methods were performed in accordance with the relevant guidelines. The data will be shared by the principal investigator (F.A.M.) upon reasonable request. We retrospectively evaluated our SAH database to identify patients who underwent NovaGuide imaging between March 1 st , 2020 to April 30 th , 2020. We collected demographic data, admission clinical and radiologic features, and whether the patient had an external ventricular drain (EVD) or central venous line (CVL) at the time of their TCD examinations. Informed consent was obtained from all subjects and/or their legal guardians. Our management protocol calls for CTA on SAH days 4, 7 and 10-12 (with day 0 being the calendar day of SAH onset). NovaGuide TCD scans were acquired on the same day as the CTAs. The degree of vasospasm in the proximal M1 segment of the middle cerebral arteries was graded as absent, mild, moderate, or severe by a neuroradiologist in a standard clinical radiology report, who was blinded to the TCD results.
To assess for complications associated with the use of the NovaGuide device, we analyzed electronic medical records for nursing documentation of any EVD or CVL disruptions or removals. From the TCD readouts, we collected the maximal and average mean blood flow velocities and pulsatility indices (PI) from the left and right middle cerebral arteries. Average mean velocity ≥86 cm/s or a maximum mean flow velocity ≥120 cm/second were used as thresholds for defining vasospasm of the middle cerebral artery. After categorizing the velocities, we compared the level of agreement between the NovaGuide TCD measured velocities and CTA reports using a Cohen's Kappa test. All statistics were run using R-studio and were considered significant when P<0.05.
TCD data was collected using the NovaGuide System (NovaGuide Intelligent Ultrasound, NovaSignal Corp., Los Angeles CA USA) which uses 2 MHz bilateral probes contained within a five degree of freedom robotic mechanism (Figure 1). The system is setup on the patient in the supine position (0° to 45° head of bed angle) with the patient's head resting in the system's head cradle. Once the patient is positioned in the NovaGuide head cradle, the healthcare provider gently engages the robotic probes onto the patient's temporal region. The system uses machine vision to co-register the patient with the robot using specially designed fiducials. After the registration, signal acquisition includes two steps, transtemporal window identification, and MCA signal optimization with no operator involvement after setup. NovaGuide uses signal quality assessment (SQA) as a rapid means of evaluating signals in real time with segments of data between 350ms -500ms. 12,13

RESULTS
Our study population consisted of twelve patients with two males and ten females with an average age of 63.5 years. From the identified cohort, 50% had radiologically  (Table 1).     (Figure 2c).

Discussion
The present investigation aimed to evaluate the safety and efficacy of the utilization of the NovaGuide system in the setting of a neurocritical care unit to assess for the presents of vasospasm in patients with acute subarachnoid hemorrhage. One of the prominent safety concerns regarding the NovaGuide device is the autonomous guidance of the insonators on the scalp and the risk this motion may pose to external CVL or EVD. We experience no complications associated with the autonomous motion on the EVDs or CVLs in the 27 scans performed. Additionally, we compared the mean and maximum CBFv for the MCA obtained by NovaGuide against CTA confirmed vasospasm. The diagnostic threshold of a maximum CBFv ≥ 120 cm/s produced substantial inter-observer agreement between the NovaGuide and CTA outcomes as well as displayed high sensitivity, specificity, and positive likelihood ratio.
Based on current literature, the consensus among publications is that a normal mean CBFv is below 85 cm/s with any value above this cut-off indicating abnormal flow and vessel stenosis. Specifically for vasospasms, a velocity between 86 cm/s and 120 cm/s is thought to indicate mild vasospasm, velocities from 120 -200 cm/s as moderate to severe spasm, and greater than 200 cm/s to be critical. 14,15 Thus, we first investigated the mean CBFv as a diagnostic criterion for detecting vasospasm using the NovaGuide acquired flow velocities. Surprisingly, there were no true positives detected when using this criterion. It is highly unlikely that this outcome is due to the NovaGuide system as it

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.  Maximum CBFv ≥ 120 cm/s as a diagnostic metric for determination of vasospasm. c) Parameters quantifying the performance of these metrics.