METHOD FOR THE MEDICAL MONITORING IN REAL TIME OF A PATIENT FROM THE ANALYSIS OF ELECTROENCEPHALOGRAMS TO CHARACTERIZE AND DIFFERENTIATE BETWEEN PHYSIOLOGICAL OR PATHOLOGICAL CONDITIONS, AND A METHOD FOR ANTICIPATING EPILEPTIC SEIZURES

• US 6,442,421 B1
• Filed: 04/27/2000
• Issued: 08/27/2002
• Est. Priority Date: 04/27/2000
• Status: Expired due to Fees

• Current Assignee: Centre National De La Recherche Scientifique
• Original Assignee: Centre National De La Recherche Scientifique
• Inventors: Le Van Quyen, Michel, Martinerie, Jacques, Varela, Francisco, Baulac, Michel
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): DROESCH, KRISTEN L
• Application Number:
  US09/559,334
• Time in Patent Office: 852 Days
• Field of Search: 600/544-545, 128/920, 128/923
• US Class Current: 600/544
• CPC Class Codes: A61B 5/369 Electroencephalography [EEG] devices for psychotechnics A61B5/16
• CPC Class Codes: A61B 5/372 Analysis of electroencephalograms
• CPC Class Codes: A61B 5/375 using biofeedback
• CPC Class Codes: A61B 5/4094 Diagnosing or monitoring seizure diseases, e.g. epilepsy
• CPC Class Codes: Y10S 128/92 Computer assisted medical diagnostics

Abstract

A method for the detection of changes in dynamic properties of brain electrical activity is used to characterize and differentiate between physiological and pathological conditions, and to anticipate epileptic seizures. The detection of changes in dynamic properties of brain electrical activity is accomplished by constructing reference dynamics (S_ref) of a normal state from an EEG, comparing the reference dynamics (S_ref) with dynamics of a test segment (S_t), and computing the similarities over the entire EEG recordings by sliding the test segment (S_t) periodically.

Claims

1. A method for medically monitoring in real time a patient from an analysis of electroencephalograms, comprising the steps of:

constructing reference dynamics of a normal state of the patient by prerecording a long normal EEG segment S_ref;

comparing the reference dynamics with dynamics of a distant test segment S_t; and

computing similarities over the entire prerecorded EEG segment by sliding the test segment S_t periodically over the prerecorded EEG segment so as to provide information about a possible medical onset occurring with the patient,

wherein the comparing step comprises the steps of;

building a skeleton of the reference dynamics by randomly selecting a sub-set of points of the reference dynamics so as to provide an adapted reference dynamics picture X(S_ref) of the reference dynamics; and

estimating dynamic similarities between the adapted reference dynamics picture X(S_ref) and a projection X(S_t) of a 16-dimensional reconstruction of the test segment S_t on the principal axes of the reference dynamics,

wherein the dynamic similarities are estimated using a statistical measure based on the following cross correlation integral;

$C<br><br>\left(S_{\mathrm{ref}},S_t\right)=\frac{1}{N_{\mathrm{ref}}<br><br>N_t}<br><br>\underset{i=1,N_{\mathrm{ref}}}{\sum <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>\underset{j=1,N_t}{\sum <br><br>}<br><br>\mathrm{\Theta <br><br>}\left(<br><br>X_i<br><br>\left(S_{\mathrm{ref}}\right)-X_j<br><br>\left(S_t\right)<br><br>-r\right)$

where Θ

is the Heaviside step function, || || is the euclidian norm, N_ref and N_t denotes the number of elements in each set, and r is a distance, and

wherein the following cross-correlation ratio is used;

2. The method of claim 1, wherein constructing the reference dynamics is derived from sequences of time intervals between positive crossings of the prerecorded EEG segment over a fixed threshold so as to form delay vectors defining an m-dimensional embedding of the reference dynamics.

3. The method of claim 2, further comprising:

applying a single value decomposition to the m-dimensional embedding so as to identify an optimal working space including a trajectory corresponding to the m-dimensional embedding.

4. The method of claim 1, further comprising:

dividing the prerecorded EEG segment into non-overlapping consecutive test segments of 25 seconds each.

5. The method of claim 1, wherein the method differentiates between physiological and pathological conditions.

6. A method for anticipating epileptic seizures in real time, comprising the steps of:

constructing reference dynamics of a non-seizure state in a patient by prerecording a long normal EEG segment S_ref;

comparing the reference dynamics with dynamics of a distant test segment S_t;

computing similarities over the entire prerecorded EEG segment by sliding the test segment S_t periodically over the prerecorded EEG segment so as to provide information about a possible seizure onset occurring with the patient,

wherein the comparing step comprises the steps of;

building a skeleton of the reference dynamics by selecting a sub-set of points of the reference dynamics so as to provide an adapted reference dynamics picture X(S_ref) of the reconstruction; and

estimating dynamic similarities between the adapted reference dynamics picture X(S_ref) and a projection X(S_t) of a 16-dimensional reconstruction of the test segment S_t on the principal axes of the reference dynamics, $C<br><br>\left(S_{\mathrm{ref}},S_t\right)=\frac{1}{N_{\mathrm{ref}}<br><br>N_t}<br><br>\underset{i=1,N_{\mathrm{ref}}}{\sum <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>\underset{j=1,N_t}{\sum <br><br>}<br><br>\mathrm{\Theta <br><br>}\left(<br><br>X_i<br><br>\left(S_{\mathrm{ref}}\right)-X_j<br><br>\left(S_t\right)<br><br>-r\right)$

wherein the dynamic similarities are estimated using a statistical measure based on the following cross correlation integral;

where Θ

is the Heaviside step function, || || is the euclidian norm, N_ref and N_t denotes the number of elements in each set, and r is a distance, and

7. The method of claim 6, wherein constructing the reference dynamics is derived from sequences of time intervals between positive crossings of the prerecorded EEG segment over a fixed threshold so as to form delay vectors defining an m-dimensional embedding of the reference dynamics.

8. The method of claim 7, further comprising:

applying a single value decomposition to the m-dimensional embedding so as to identify an optimal working space including a trajectory corresponding to the m-dimensional embedding.

9. The method of claim 6, further comprising:

dividing the prerecorded EEG segment into non-overlapping consecutive test segments of 25 seconds each.

10. The method of claim 6, wherein the epileptic monitoring is conducted in ambulatory conditions.

11. The method of claim 6, wherein EEG electrical recordings are taken from a scalp surface of the patient.