Deep Learning
May 1, 2023FDA Updates – May 2023
May 1, 2023NovaSignal
NovaSignal primarily aims to give doctors more access to vital data they need to better their patients' lives through advanced technology.
Overview
Created in 2013, NovaSignal gives doctors more access to vital data to better their patients’ lives. Co-Founder Dr. Robert Hamilton was working on his Ph.D. in the Department of Neurosurgery at UCLA, where he created NovaSignal’s key technology, which is still used by doctors all over the world, revolutionizing how real-time blood flow data is used for brain well-being by integrating artificial intelligence (AI), robotics, and advanced cerebral ultrasound.
AI Aspects
Ultrasound Brain Imaging for Robotics and AI: Ultrasound processing creates images using acoustic energy in the most fundamental form. The transducer, which emits these high-frequency sound waves and tracks the reflected reflections, is a vital part of an ultrasound system. NovaSignal has developed a more advanced version of a transcranial Doppler (TCD) ultrasound, a non-invasive diagnostic instrument used to assess blood pressure in and across the brain’s main arteries.
The Lucid Robotic Device comprises a handheld ultrasound unit (the Lucid TCD 2.0, pictured above) and a robotic headset (NovaGuide).According to NovaSignal, the machine does not need a technician with advanced expertise and is 99.7% reliable, proven by an experienced sonologist. The invention is secure enough to use in both the United States and Europe, and at least 15 patents protect it.More recently, the firm released an app that notifies doctors when patients’ blood flow changes, which may signify a stroke or other emergency.
NovaSignal has tweaked the AI component, which now takes data from every scan and gives clinicians predictions of how a patient will improve over time. (1)
NovaSignal’s IT Aspects
- Modality worklist: Directly upload of physician’s daily worklist for improved efficiency.
- Easy editing: Edit snapshots, add comments, and use sliders to recalculate blood flow velocities and PI.
- Automated emboli counting: power M-mode, which includes enhanced coloring, automated filtering, and improved waveform tracing.
- Automated: five degrees of freedom combined with AI enables autonomous signal capture and extended monitoring.
- Intelligent: AI quickly locates major blood vessels and assesses blood flow velocity.
- Distanced Monitoring: The autonomous nature of the system allows for contactless patient monitoring, enabling the maintenance of infectious disease protocols.
- Compact, light, and easy to transport.
- Integrated stand.
- It can be mounted on a cart-based system for maximum portability.
- Easy carrying handle.
- Battery operation is available.
- Ability to connect to television systems to enable remote monitoring.
NovaSignal’s Financials Aspects
The recent Series C stock sale raised $22 million, bringing the total funding of NovaSignal to approximately $66 million.
This announcement was followed by the enrollment of the initial patients to their new prospective, single-arm, global multi-center study, called CODEX. This study aims to further explore the applicability of this technology to certain neurologic or other diseases where monitoring of brain blood flow data would be of benefit. (3)
COVID-19 Pneumonia Associated With Pulmonary Dilation
COVID-19 pneumonia is thought to have a different mechanism compared to classical ARDS due to evidence of hypoxemia despite normal lung compliance.
Dual-energy CT scan imaging has demonstrated pulmonary vessel dilatation, and autopsies have shown pulmonary capillary deformation in patients with COVID-19 pneumonia.
A study of transcranial Doppler (TCD) ventilated patients with severe COVID-19 pneumonia showed that 83% of patients had detectable microbubbles. Moreover, the PaO2:FiO2 ratio was inversely correlated with the number of microbubbles which were on contrary relationship to lung compliance. (4)
A Model of Longitudinal Hemodynamic Alterations After Mild Traumatic Brain Injury in Adolescents
In adolescents, it is well understood that vasculature is distinctly affected by a mild traumatic brain injury (mTBI). Researchers presented a hypothesis for the hemodynamic progression of adolescents after a concussion as measured by TCD ultrasound.
For this study, 70 subjects between the ages of 14 and 19 were selected. The cases consisted of clinically diagnosed mild traumatic brain injury. TCD data were collected with 2 MHz US probes and end-tidal CO2 registered through a nasal cannula.
Results showed that a three phases model of hemodynamic alterations captured with TCD after mTBI can be described. (5)
Robotically Assisted Transcranial Doppler With AI for Assessment of Cerebral Vasospasm After Subarachnoid Hemorrhage
Transcranial Doppler (TCD) is an important instrument for the diagnosis of cerebral vasospasm after subarachnoid hemorrhage (SAH), but its effectiveness is operator-limited. In this study, researchers evaluated the feasibility and concordance of a robotically assisted TCD system with AI with routinely handled TCD after SAH.
They reviewed the clinical data, TCD results, and imaging reports from two patients with high-grade SAH and intraventricular blood. Once the Manual TCD imaging was completed, the Lucid Robotic System (Neural Analytics) with AI and ML algorithms automatically searched for and detected the bilateral MCAs independent of the operator. The ACA measurement was not fully automated because of required adjustments.
Results showed that robotically measured mean flow velocities were comparable to manual TCS in the MCAs but not in the ACAs..(6)
Detection of Pulse Onsets in Cerebral Blood Flow Velocity Signals
A new algorithm for pulse onset detection in cerebral blood flow velocity (CBFV) waveforms suggested a fundamental improvement compared to existing pulse onset screening algorithms. The system involves a moving difference filter (MDF) and adaptive thresholding to identify windows where changes are predictable and errors can be correctly extracted.
Patients with subarachnoid hemorrhage were scanned, and the algorithm precisely identified all the cases in the data set. More than 99.5% of cases were detected within 30 milliseconds of the onset, with only two false detections.(7)
