RecogniZe User Guide V2 Jan 2008

Recognize User Guide

The BrainZ RecogniZe module incorporates a patented, proprietary automatic periodic event detection algorithm that detects events that may correspond with seizure activity in a patient. The algorithm is based on wave-sequence analysis and operates in near real time, marking sections of the recording containing events that may correspond with seizure activity. This software has been designed to alert bedside clinicians to areas of EEG which require review and should only be used to help obtain the attention of clinical staff. A trained clinician can then review marked sections of the EEG trace and quickly confirm activity that may be interpreted as a seizure. The BrainZ RecogniZe product does not provide any diagnostic indication of the patient’s condition.

Intended use BRM Brain Monitor and RecogniZe detection algorithm is intended for use: • to monitor the state of the brain by acquisition of electroencephalogram (EEG) signals, in clinical environments such as the intensive care unit, operating room, and for clinical research. • only with neonatal patients, defined as from birth to 28 days post-delivery, and corresponding to a post-conceptual age of 24 to 46 weeks. • to identify events in acquired EEG signals that may correspond to seizure activity in the patient. Event detections must be reviewed and interpreted by qualified clinical practitioners. The BRM Brain Monitor does not provide any diagnostic conclusion about the patient’s condition to the user.

RecogniZe algorithm description The main principle of operation of the RecogniZe algorithm (Navakatikyan et al., 2006) is the detection of heightened regularity in EEG wave sequences, using wave intervals, amplitudes and shapes. Heightened regularity is the major

RZe

|

distinguishing feature of seizure discharges, and the most significant indicator used in seizure identification by a trained neurologist. The algorithm comprises filtering of the EEG signal, parallel fragmentation of the signal into wave sequences, wave-feature extraction and averaging, then elementary, preliminary and final detection. The Left and Right channels of the BRM Brain Monitor are used for event detection. Preliminary detections are considered one event if closer together than 30 seconds, and isolated detections shorter than 10 seconds are discarded. Event detection criteria are applied to these stages as identified below.

Detection criteria The RecogniZe algorithm detects areas of regularity in an EEG waveform, using the following criteria: • • • •

at least five similar consecutive waves wavelengths that are equivalent to a frequency of 14 Hz or less peak-to-peak amplitude greater than 5 μV at least 21 seconds of continuous detection, or 26 seconds of discontinuous detection, in one minute of EEG signal.

The algorithm then indicates the most probable location of events that may correspond with seizure activity in the patient.

Limitations The RecogniZe algorithm is subject to a degree of false detection in the presence of certain artifacts (for example, EKG interference, respiratory artifact and electrode movement). These spurious detections can be minimized by good sensor application technique, and by vigilance on the part of care-givers (to avoid creating artifacts by rhythmically patting or moving the baby while RecogniZe monitoring is taking place). The RecogniZe algorithm might not detect a small proportion of events corresponding to genuine seizure activity, particularly those events made up of complex, irregular waves, or of very short duration.

RZe

The RecogniZe algorithm is limited to analysis of the EEG data channels recorded by the BRM Brain Monitor, and may not detect some events corresponding to focal seizures occurring in regions other than the central-parietal regions where the sensors are placed.

Safety warnings In no way, and despite any medical terminology used, do any of the RecogniZe functions represent a diagnostic indication of the patient’s condition. Sections of EEG or aEEG trace marked by RecogniZe must be reviewed by a trained clinician to confirm the presence of seizure activity (from examination of the relevant EEG Waveform trace) both for online and offline use of the device. Do not perform this review unless you are trained to recognize seizure activity from raw EEG Waveform traces. The RecogniZe module is designed for neonatal use only (defined as from birth to 28 days post-delivery, and corresponding to a post-conceptual age of 24 to 46 weeks). Do not use this feature to detect in other patient populations. Do not rely on RecogniZe event detection or alerts for critical alarming or monitoring of patient safety.

Enabling and disabling the RecogniZe module If the BRM Brain Monitor is not running you will need to turn on the Power Switch, log in and navigate to the System Setup screen. If the RecogniZe module is enabled or disabled, it will remain that way for all logged in users until the setting is changed.

RZe

To enable RecogniZe: 1. In the System Setup screen, select the Enable RecogniZe check box to enable the RecogniZe module. 2. Touch OK.

Result: The RecogniZe module is enabled and the Recognize icon is blue. (a tick icon appears) Rze . To disable RecogniZe: 1. In the System Setup screen, de-select the Enable RecogniZe check box to disable the RecogniZe module. 2. Touch OK.

Result: The RecogniZe module is disabled and the Recognize icon is red. (a cross icon appears) Rze .

Using RecogniZe Begin an assessment as normal. Refer to the BRM Reference Manual or BRM Onscreen Help System for detailed instructions on beginning a patient assessment. The RecogniZe module operates by marking sections of the recording that contain events which may correspond with seizure activity. Event detections must always be reviewed and confirmed against the EEG recording by a qualified clinician. The purpose of the RecogniZe module is to focus the attention of the clinician on sections of the trace that may be clinically significant. During an assessment, the RecogniZe alert appears in the Title Bar of all main displays, and in the Signal Status screen, indicating events that may correspond with seizure activity in the patient. The Title Bar remains orange until the event ends, and the Title Bar is touched to reset it to its normal blue color. This latched alert serves as an indication that an event has occurred some time in the past, and may not be visible on the current real-time screen.

RZe

Sections of the recording containing detections are marked with an orange bar that appears at the top of aEEG graphs in real time mode, and above both EEG and aEEG graphs in review mode. The marking is applied to the Left, Right and Cross-cerebral channels jointly, regardless of the origin of the suspect activity.

EEG Waveform Left

EEG Waveform Right

aEEG Left

aEEG Right

Acknowledging a RecogniZe alert To acknowledge a recognize alert: An orange Title Bar indicates that the RecogniZe module has detected an event that may be related to seizure activity. Touch the Title Bar.

Result: The Title Bar remains orange until the indicated event has stopped, then the Title Bar reverts to blue. If the event has already stopped, the Title Bar reverts to blue. Note that there may be a delay before the Title Bar reverts to blue.

RZe

Reviewing a patient assessment To display the Review screen for the current assessment: Touch Review Assessment to display the current assessment in review mode. To display the Review screen for a past assessment: 1. On the Main Menu screen, touch Manage Data to display the Patient Data File Manager screen. 2. Select the file you want to view, in either the Internal Storage or Removable Media file list. 3. Touch the appropriate View button. To review the displayed assessment: 1. Scan the aEEG graphs to identify an area of trace that is marked with an orange bar, using the lower Pan arrows as necessary. 2. Touch one of the aEEG graphs in the area of the orange bar. 3. Scan the area of the EEG Waveform Left and Right graphs marked with an orange bar, to identify areas that may represent seizure activity, using the upper Pan arrows as necessary. Sections of EEG or aEEG trace marked by RecogniZe must be reviewed by a trained clinician to confirm the presence of seizure activity (from examination of the relevant EEG Waveform trace). Do not perform this review unless you are trained to recognize seizure activity from raw EEG traces. The section of the recording marked with an orange bar by RecogniZe represents the most likely location of the suspicious activity the module has detected. Close observation of the unmarked sections of the recording may reveal some additional areas of repetitive EEG activity, particularly where these are short (less than 10 seconds) or made up of complex repeating waveforms. It may also be helpful to touch the 1m button to view the trace on a more compressed time scale while searching for areas of regular activity.

RZe

Changing the low amplitude detection threshold The RecogniZe algorithm has been factory-set to a default low amplitude detection threshold of 5 μVp-p. This setting provides the best balance of sensitivity and artifact rejection over a wide range of environments and circumstances. In individual cases, however, clinicians may choose to raise the detection threshold in order to make the algorithm less sensitive to low-amplitude artifacts (for example, EKG interference and high-frequency ventilator artifact). Raising the detection threshold may cause an increase in the amount of seizure activity that is not identified by the software. To change the low amplitude detection threshold: 1. 2. 3. 4.

Touch Tools to display the Tools menu. Touch Change RecogniZe settings. Select 5.0, 7.5, or 10.0 μVp-p depending on the sensitivity you require. Touch OK to confirm the setting.

If the low amplitude detection threshold is modified during an assessment, this is automatically marked as an event in the recorded trace. Each time a new assessment is started, the low amplitude detection threshold reverts to the default value (5 μVp-p).

Hiding the RecogniZe marking The RecogniZe marking may be hidden, causing the RecogniZe indication to disappear from the current screen in both “review” and “real-time” modes. In “real-time” mode, events continue to be detected and recorded in the patient data file.

RZe

To hide the RecogniZe marking in the current screen: 1. Touch Tools to display the Tools menu. 2. Touch Hide RecogniZe marking.

Result: The RecogniZe marking disappears in the current assessment and the RecogniZe icon is greyed out

Rze .

To show the RecogniZe marking in the current screen: 1. Touch Tools to display the Tools menu. 2. Touch Show RecogniZe marking.

Result: The RecogniZe marking is displayed in the current assessment and the Recognize icon is blue (a tick icon appears) Rze .

Helpful hints Periodic artifacts may trigger a RecogniZe detection, particularly those within the 2 – 14 Hz range. Avoidance of these artifacts is not always possible, particularly when using a high frequency oscillating ventilator. Other artifact sources that could cause a RecogniZe detection include patting the infant, chest percussion therapy and cardiac (EKG) artifact. In these circumstances, comprehensive event marking can help the clinician when reviewing the recording. It is advised to mark events whenever they occur, particularly the presence of: • •

clinical seizures potential artifact sources (as described above).

Maintaining good contact quality of all sensors will ensure a good quality signal, by minimizing artifact, and help the clinician in reviewing the recording. Sensor impedances should be kept below 5 k7 to ensure a good quality signal. Following an assessment, a trained clinician can mark events retrospectively, to confirm the presence of seizure activity (from examination of the relevant EEG Waveform trace) or explain detections or artifact sources.

RZe