Why ROI Region Of Interest

1

• Why ROI ? (Region Of Interest)

• To explore ones data (difficult to discern
  patterns across whole brain)
• To control for Type 1 error and limit the number
  of statistical tests by using pre-defined
  knowledge
• Anatomical landmarks
• separate functional localizer

2
(No Transcript)
3
Individually Vs. Group

• replication across groups and not across
  subjects?
• Different location and/or size of activations
  between subjects?
• Experimenter bias?

Anatomical Vs. Functional

• Using orthogonal contrast
• Including only relevant voxels?
• Relevant in what contrast?
• Experimenter bias?

4

• How ROI ? (Region Of Interest)

• To explore ones data (difficult to discern
  patterns across whole brain)
• peak of activation (sphere) masked with the
  threshold activation map -orthogonal contrast
• To control for Type 1 error and limit the number
  of statistical tests by using pre-defined
  knowledge
• Anatomical landmarks
• Using individual subjects anatomy for detention
  not Talairach
• separate functional localizer
• Orthogonal contrasts in a factorial design

(Poldrack, 2007)
5
Conclusions

• No one good way to extract ROI There has not
  been a systematic study of the power of one
  approach in comparison with the other
• Different brain regions may require different
  methods
• Different protocols may require different
  methods
• Voxels count have been shown to be an unreliable
  measure of activation compared to signal change
  (Cohen Dubois, 1999)
• Tip - using ROI to remove outliers (Wager et al.,
  2005)