Surgery – Diagnostic testing – Detecting brain electric signal
Reexamination Certificate
2001-11-30
2003-07-08
Hindenburg, Max F. (Department: 3736)
Surgery
Diagnostic testing
Detecting brain electric signal
C128S920000, C128S923000
Reexamination Certificate
active
06591132
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to the art of artifact detection, in particular to an apparatus and a method for detecting artifacts in encephalogram (EG) data.
BACKGROUND OF THE INVENTION
In many cases, only a small portion of encephalogram (EG) data is of interest to a user. Typically, EG data comprises a plurality of data channels acquired using respective electrodes located on a human or animal head. The EG data may comprise features of interest, such as spikes, and large portions of data irrelevant for a particular application.
Furthermore, when the features of interest are spikes, not all the spikes observed in the EG data relate to a physiological event of interest. For example, a spike related to a physiological event of interest may be an epileptiform transient in an electroencephalogram (EEG) and an irrelevant spike, referred to as an artifact, may be either related to an eye movement or an electrode-caused artifact. Sometimes, the identification of spikes related to a physiological event of interest can even be difficult for experienced encephalographers.
Against this background, there exists a need in the industry to provide a novel method and apparatus to detect spikes having a high likelihood of being artifacts.
SUMMARY OF THE INVENTION
In a first broad aspect, the invention provides a machine-readable storage medium containing a program element for execution by a computing device for performing spike and artifact detection in EG data. The program element comprises a spike detection module for processing EG data to detect spikes. Each spike is a candidate having a likelihood of being related to a physiological event of interest. The program element further comprises an artifact detection module. The artifact detection module is operative to compute respective models of events manifested by the respective spikes detected by the spike detection module, to use the computed models to determine which spikes among the spikes detected by the spike detection module have a high likelihood of being artifacts, to filter the spikes detected by the spike detection module on the basis of the computed models to produce filtered data and to output the filtered data.
An advantage provided by the invention is the filtering of the EG data through the identification of highly likely artifacts. Consequently, spikes having a high likelihood of being related to a physiological event of interest can be identified more easily. This may be advantageous in reducing the amount of data to be presented for interpretation by a user. Furthermore, the identification of highly likely artifacts may identify spikes that may be hard to identify as unrelated to a physiological event of interest by the user.
An event is a cause of an observable spike in one or more channels of EG data. The EG data may be electroencephalogram (EEG) data or magnetoencephalogram (MEG) data, among others. An event may be physiological or not. A physiological event is an event produced by the physiology of the human or animal from which the EG data is acquired. Very specific non-limiting examples of physiological events include activation corresponding to an eye movement and brain activity producing an epileptiform transient or sleep transient. A non-limiting example of a non-physiological event is an electrode displacement.
An artifact is any spike that does not relate to a physiological event of interest. For the purpose of this specification, a physiological event of interest relates to one or more events that a user is interested in identifying in the EG data. Examples of physiological events of interest include physiological events producing epileptiform transients or sleep transients, such as vertex sharp waves, among others. An artifact may be related to a non-physiological event. Alternatively an artifact may be related to a physiological event, that event being of no interest to the user.
In a specific and non-limiting example of implementation, the computation of the model of the event manifested by the spike that the artifact detection module computes includes computing a dipole equivalent of the event through the computation of one or more characterizing elements. Specific examples of characterizing elements include the location and the eccentricity of the dipole equivalent as well as the residual variance associated with the dipole equivalent. Preferably, the time instant at which the dipole equivalent is to be computed is determined solely from the EG data, without any user intervention. Alternatively, the user may enter rules related to the determination of the time instant or provide interactive inputs to the program element for guiding the computation of the time instant.
The artifact detection module uses criteria based on one or more characterizing elements of The model, which may be a dipole equivalent, to determine if a spike has a high likelihood of being an artifact. In a first variant, the characterizing element is the location of a dipole equivalent in a two-dimensional or three-dimensional space. If the location of the dipole equivalent is unlikely to be related to a physiological event of interest, then the spike can be characterized as being highly likely an artifact. For example, the location of the dipole equivalent may indicate that the spike is related to eye movement, and therefore has a small likelihood of being an epileptiform transient. In other variants, the artifact detection module relies on the residual variance or eccentricity of the dipole equivalent. In a further variant the artifact detection module relies on a combination of characterizing elements, such as those mentioned herein above. The reader skilled in the art will appreciate that other characterizing elements may be used without departing from the spirit of the invention.
The criteria used by the artifact detection module may be a threshold above which or below which a characterizing element of the model indicates that the spike is highly likely an artifact. Alternatively, the criteria may be an interval of values for which the characterizing element of the model indicates that the spike is highly likely an artifact. The criteria may also be a combination of intervals, thresholds or both for one or more characterizing elements.
As indicated previously, the spikes detected by the spike detection module are filtered on the basis of the computed models. The purpose of the filtering step is to separate for the user who will ultimately analyze the data spikes that are highly likely to be related to physiological events of interest from spikes that are highly likely artifacts. There are many ways in which the filtered information can be presented to the user. The extent to which a spike is highly likely and artifact or highly likely related to a physiological event of interest is determined through the criteria used by the artifact detection module. In a non-limiting example of implementation, the filtering performed by the artifact detection module specifically identifies each spike as highly likely being either an artifact or related to a physiological event of interest. Another possibility is to simply remove the spikes that are highly likely artifacts such that the user only sees the spikes determined to be highly likely physiological events of interest.
In a second broad aspect, the invention provides a machine-readable storage medium containing a program element for execution by a computing device for performing artifact detection in EEG data. Tile program element comprises an input for receiving data derived from an EG and representative of spikes, each spike being a candidate having a likelihood of being related to a physiological event of interest. The program element further comprises an artifact detection module for receiving the data representative of spikes and being operative for computing respective models of events manifested by the respective spikes and using the computed models to determine which spikes among the spikes detected by the spike detection module are highly likely to be artifacts. The
Agarwal Rajeev
Flanagan Danny
Gotman Jean
Wang YunHua
Hindenburg Max F.
Natnithithadha Navin
Stellate Systems Inc.
LandOfFree
Artifact detection in encephalogram data using an event model does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Artifact detection in encephalogram data using an event model, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Artifact detection in encephalogram data using an event model will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3027602