Understanding hudspeth connectivity analysis

Copyright © 2008 Gary J. Schummer PhD APC., and Brain Potential Inc. Unauthorized duplication or distribution is illegal.

Overview of the Hudspeth Connectivity Analysis

Gary J. Schummer, Ph.D.

Neuropsychologist

Clinical Director of the A.D.D. Treatment Center

Torrance, California, U.S.A.


Typical Connectivity Report

Page 1 – The patient identification,

referral issue, test date, and an

intelligently chosen sample of EEG

xxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxx



Page 1 – ID, referral issue, date, an


Page 2 – An analysis of the raw data

with the eyes closed





Page 2 – Analysis of data eyes closed

Page 3 – A database comparison in

the eyes closed condition






Page 3 – A database comparison in

the eyes closed condition

Page 4 – An analysis of the raw data

with the eyes open



Page 1 – ID, referral issue, date, an intelligently chosen sample of EEG


Page 3 – A database comparison in the eyes closed condition

Page 4 – Analysis of data eyes open

Page 5 – A database comparison in the eyes open condition


“Normal” Connectivity Report (Raw, Adult)

EigenImage (From EO Norm)

EigenImages

Derived from 3-dimensional

Principle Component’s Analysis

There are 4 principle components that account for the variance; 3 of the 4 sources of variance are able to be illustrated, the 4th is unknown:

1. Horizontal

2. Sagittal

3. Coronal

4. Unknown

.5 - 3.5 3.5 - 7 7 - 13 13 – 22 .5 – 22 Hz Hz Hz Hz Hz

Hudspeth Frequency Range Definitions

Horizontal – view from above (Cz) Sagittal – view from right side (T4) Coronal – view from in front (Fpz)

Views or Factors

New metric – shared variance between the percent distance of the vectors of the frontal areas. A measurement of how well the frontal areas are holding together Percent

The lower the number from the norm (~25-32%) the more the frontal sites cluster together and indicate hypercoherence

As the difference between a percentage approaches 10 (compared to its 4 neighbors) the greater the liklihood that the data for that frequency are invalid “norms” are in development

Midlines – 1 horizontal, 1 vertical per factor and per band

Normal

Abnormal Able to treat

Invalid Unable to treat

r2 =The amount of variance explained by the Horizontal, Sagittal, and Coronal factors expressed as a percentage The higher the percentage, the better

“Total” is typically lowest

r2 – sum of factors

Indication of likely source of artifact or site that

is highly deviant

Note – these do not appear in a “normal” image

These could be valid, but it’s a hint to check the

raw EEG for artifact

Note F7 and F8 are typical areas of muscle tension

Invalid Unable to treat

May indicate the source of the invalidity

Note T5

QUESTIONS ABOUT THE EIGENIMAGE ??

% Shared Variance

What it’s NOT:

– NOT derived from the same computation as the EigenImages

– NOT a 3- dimensional analyses

– NOT derived from Principle Components Analysis

% Shared Variance

What it is:

A multivariate measure of coherence within a 2-dimensional plane

Hudspeth’s “% Shared Variance” measures are more similar to the coherence readings given in NxLink and Neuroguide than they are to the EigenImages

Of course, there will be similarities between the EigenImages and the %Shared Variance

Intra (hemispheric) - each site’s percent of shared variance with 7 sites in the same hemisphere including the 3 midline sites; total 10

Inter (hemispheric) - each site’s percent of shared variance with all 8 sites in the opposite hemisphere

Each Midline site is compared with all other sites; total of 18


Each Midline site

Fz

Cz

Pz

TOTAL Intrahemispheric

% Shared Variance

is compared with the other 18 sites

Fz

Each Midline site

TOTAL Intrahemispheric

% Shared Variance

Fz-Fp2

To give a number which indicates the strength of that site, relative to the other 18

Each Midline sight is compared with all other sites; total of 18














% Shared Variance From EO Normal

Note the outside percentages are typically lower than sites closer to the midline - because they have fewer “neighbors”

Z-Score comparison to the database

Z-Score comparison expressed as a percentile

Intra – each site’s Z-Score or Percentile relative to the other 7 sites in the same hemisphere plus the 3 midline sites (Total of 10)

Inter – each site’s Z-Score or Percentile relative to the 8 sites in the opposite hemisphere

Midline sites are compared with all other sites (Total of 18)

Questions?

Understanding hudspeth connectivity analysis

Health & Medicine

intelligently

schummer phd

typical connectivity

eyes closed

data eyes

frontal areas

raw data

shared variance