SNR 2000 LORETA Stat Maps

8th Annual SNR Meeting St. Paul, 24 Sept. 2000: 

Presenters: Joel Lubar, Marco Congedo 8th Annual SNR Meeting St. Paul, 24 Sept. 2000 The Authors wish to thanks Lexicor Inc. for support during a visit to the Key Institute of brain-mind research, Zurich, Switzerland, and Dr. Pascual-Marqui for his wonderful hospitality.

LORETA Maps: 

LORETA Maps Basically LORETA Maps are maps of Statistics, i.e., the current density at each Voxel is summarized across time. Mean Current Density (across 15360 Time Frames) of a 25 y.o. female engaged in a eyes open baseline. EEG SR=256. Recording length=60 secs. Source: LORETA Key, R.D. Pascual-Marqui, The key Institute for brain-mind research, Zurich, Switzerland.

Kind of Statistics: 

Kind of Statistics Statistics can be either Descriptive (mean, variance..) or Inferential (z, t, p-values) and can refer to one or several Subjects as compared while performing one or more tasks Standardized Mean c.d. (across 15360 Time Frames) of a 25 y.o. female engaged in a eyes open baseline. EEG SR=256. Recording length=60 secs. Source: LORETA Key, R.D. Pascual-Marqui, The key Institute for brain-mind research, Zurich, Switzerland.

Statistical Analysis in LORETA: 

Statistical Analysis in LORETA Inferential Statistical Analysis in Neuroimaging techniques such as PET, fMRI and LORETA has to face the problem of Multiple Testing and the problem of data non-normality. Here we briefly present two inferential analysis method which are adaptable for LORETA data.

Statistical non-Parametric Maps (SnPM): 

Statistical non-Parametric Maps (SnPM) SnPM are Maps of t-statistics suitable to compare two groups of subjects either in a paired or in a unpaired design. T-values (paired t-test) computed on six subjects comparing a READING to a Eyes Open BASELINE. A red spot indicates that the group mean during “reading” was greater then during baseline Results: 38 Significant voxels located in Left BA 18 and 19 Peak activity: -31, -81, 22, Left Middle Occipital Gyrus, BA 19 -38, -88, 22, Left Middle Occipital Gyrus, BA 19 -31, -88, 22, Left Superior Occipital Gyrus, BA 19 -31, -95, 8, Left Middle Occipital Gyrus, BA 18 -31, -81, -13 ,Left Fusiform Gyrus, BA 19 Threshold of Significance: 3.839

SnPM characteristics: 

SnPM characteristics The test is in the same extent valid when applied to a few subjects or to many subjects. The test requires minimal assumptions (random assignment of Subjects to conditions and independence of observations) The test exercises strong control over the Experiment-wise error (Holmes et al., 1996), i.e., despite the high number of tests performed the probability to commit a type I error is always less then alpha (0.05).

Poisson Tests of Maxima (PTM): 

Poisson Tests of Maxima (PTM) The Poisson Tests of Maxima are valuable in that they allow to test hypotheses in single subject cases. The Poisson Test of Maxima Frequency (PTMF) is applied to a single EEG recording (single task). The Poisson Test of Maxima Frequency Comparison (PTMFC) is applied to Activation Studies, i.e., when a “activation” task is compared with its appropriate baseline (subtraction paradigm).

PTMF rationale: 

PTMF rationale For each Time Frame, the Poisson Test of Maxima Frequency considers the Maximum Current Density Magnitude value across the entire Volume (the voxel with maximal activation). At each Voxel, the number of times the Maximum falls in that Voxel (Maxima occurrence) is compared with the occurrence expected under a by-chance-alone null-hypothesis (all voxels have the same probability of registering the maximum). P-values are corrected by means of a Bonferroni method as 2394 tests are performed.

PTMF Map: 

PTMF Map PTMF Map of a 25 y.o. female watching to a red screen (8192 TF, 64 sec). A red spot indicates that Maxima tends to fall in those Voxels with regularity Main Peak Activities: X= 4 Y= -67 Z= 15 Posterior Cingulum X= -3 Y= 52 Z= 1 Medial Frontal Gyrus X= -52 Y= -67 Z= 8 Middle Occipital Gyrus X= 53 Y= -53 Z= 15 Superior Temporal Gyrus X= -3 Y= -46 Z= 43 Precuneus (PCu) Source: LORETA Key, R.D. Pascual-Marqui, The key Institute for brain-mind research, Zurich, Switzerland.

PTMFC rationale: 

PTMFC rationale The Poisson Test of Maxima Frequency Comparison do not need of a by-chance-alone null-hypothesis. In fact it makes use of the occurrence of Maxima as observed in the “baseline” task. These occurrences form the null-hypothesis with which the observed occurrence for the “activation” task are compared. P-values are corrected by means of a Bonferroni method as 2394 tests are performed.

PTMFC Map: 

PTMFC Map PTMFC Map of a 25 y.o. female comparing watching to a red screen (A) with watching to a white screen (B) (both EEG file 8192 TF, 64 sec). A Red Spot indicates that Maxima tends to fall in that Voxels more in A then in B. A Blue spot indicates the opposite. The two main Red Spots (together f=66% of total for condition A) form a cluster covering a gray matter area at the intersection of the Cuneus, Lingual Gyrus and Posterior Cingulate. Source: LORETA Key, R.D. Pascual-Marqui, The key Institute for brain-mind research, Zurich, Switzerland.