Microwave Imaging for Brain Stroke Detection and Monitoring using High Performance Computing in 94 seconds
The device, which measures electrical activity in the brain and then uses an algorithm to decide if a patient is likely to have brain bleeding, can help with clinical decision-making and triage of patients, and could reduce the need for computed tomography (CT) scans.
“Before our study, there were no objective, quantitative measures of mild head injury other than imaging,” said Daniel Hanley Jr., MD, Johns Hopkins University School of Medicine, Baltimore, Maryland. “This work opens up the possibility of diagnosing head injury in a very early and precise way.”
“This technology is not meant to replace the CT scan in patients with mild head injury, but it provides the clinician with additional information to facilitate routine clinical decision-making,” he said. “If someone with a mild head injury was evaluated on the sports or battlefield, then this test could assist in the decision of whether or not he or she needs rapid transport to the hospital. Alternatively, if there is an accident with many people injured, medical personnel could use the device to triage which patients would need to have CT scans and who should go first. Those showing a ‘positive’ for brain injury would go first.”
The study, published in Academic Emergency Medicine, only looked at adults and didn’t assess how well the device could predict traumatic brain injuries in children or teens.
The study was designed to test the accuracy and effectiveness of AHEAD 300, which is now available to a limited audience through a centre of excellence program. The point of the device is to assess the likelihood that a patient has more than 1 mL of bleeding in the brain and needs immediate evaluation by medical personnel.
To begin, the researchers recruited 720 adults aged 18 to 65 years who came to 11 Emergency Departments across the nation between February and December 2015 with a closed head injury. Upon entry to the Emergency Department, each physician performed standard clinical assessments for head injuries used at their site. A trained technician then administered the Standardised Assessment of Concussion and the Concussion Symptom Inventory to characterise the patient’s symptoms, and then used the AHEAD 300 device to measure electroencephalogram (EEG) data from patients while they reclined quietly for 5 to 10 minutes.
The device includes a disposable headset that records the EEG data from 5 regions on the forehead and feeds the signals back to the hand-held AHEAD 300 device in real time. In addition, the technician entered certain clinical/demographic information into the device, including age, the Glasgow Coma Scale score, and if there was a loss of consciousness related to the injury.
The device was programmed to read approximately 30 specific features of brain electrical activity. The accuracy of the device was tested using CT scans from the participants. The presence of any blood within the intracranial cavity was considered a positive finding, indicating brain bleeding. After 72 to 96 hours, the researchers followed up with phone calls to the patients and/or looked at medical records after 30 days to further confirm the accuracy of each participant’s injury status.
Of the 720 patients, 564 turned out not to have traumatic brain injuries, and 156 did have them, as determined by independently measured and judged CT scan assessments.
On the basis of AHEAD 300 classification, the researchers sorted patients into “yes” or “no” categories, indicating likely traumatic brain injury with over 1 millimetre of bleeding or not. Of 564 patients without brain bleeding, as confirmed with CT scans, 291 patients were scored on the AHEAD 300 as likely not having a brain injury. Of the 156 patients with confirmed brain bleeding, 144 (92%) were assessed as likely to have an injury by the AHEAD 300 classification. Of those confirmed with brain bleeding via CT scan, 12 (8%) had some intracranial bleeding and 5 (3%) had more than 1 mL of blood in the brain.
Because many of the incorrect yes/no classifications don’t contain information about how close a patient is to the cut-off, the researchers then created 3 categories (“yes,” “no” and “maybe”) to see if this boosted the accuracy of the device. The maybe category included a small number of patients with greater-than-usual abnormal EEG activity that was not statistically high enough to be definitely positive. When the results were recalculated on the 3-tier system, the sensitivity of detecting someone with a traumatic brain injury increased to 97%, with 152 of 156 traumatic head injuries detected, and 99% of those had more than or equal to 1 mL of bleeding in the brain. None of the 4 false negatives required surgery, returned to the hospital due to their injury, or needed additional brain imaging.
The trial results also show the device predicted the absence of potentially dangerous brain bleeding 52% of the time in the participants tested with the yes/no classification. Using the yes/no/maybe classification, the device classified 281 patients as having a brain injury, correctly predicting whether someone didn’t have a head injury 39% of the time.
Although an exact cost hasn’t been set by the maker of the device, the company says it will be a fraction of the cost of a CT scanner.
SOURCE: Johns Hopkins Medicine