Image Data Analysis
Abiant’s data analysis methods have been optimized for reliability, interpretability, and utility in drug
development. We use both Voxel-Based (SPM) and Region of Interest approaches, and a combination of
univariate and multivariate methods, all designed to provide informative insights that can be used for
decision making.
development. We use both Voxel-Based (SPM) and Region of Interest approaches, and a combination of
univariate and multivariate methods, all designed to provide informative insights that can be used for
decision making.
| Complementary methodologies |
| Statistical Parametric Mapping: "Where do significant effects occur?" |
| Region of Interest: "Is there a statistically significant effect in a particular region?" |
Rate calculations
% change
Time activity curves
Kinetic modeling
% change
Time activity curves
Kinetic modeling
T-tests; Regional mapping
Cluster correction; Post-hoc tests
Power calculation
Individual and mean % change
Correlations: pK, cognition, other
Cluster correction; Post-hoc tests
Power calculation
Individual and mean % change
Correlations: pK, cognition, other
Multivariate Analysis
An important multivariate technique is the NPAIRS software developed by Dr. Stephen Strother at the
University of Minnesota and exclusively licensed and adapted for commercial use by Abiant. NPAIRS:
Abiant has been able to use this approach to evaluate dose response, segregate task effects from
drug effects (example shown), differentiate scan acclimation effects from drug effects, distinguish
patients diagnosed with probable Alzheimer’s Disease from controls, and distinguish between inter-
site effects and the AD disease pattern.
An important multivariate technique is the NPAIRS software developed by Dr. Stephen Strother at the
University of Minnesota and exclusively licensed and adapted for commercial use by Abiant. NPAIRS:
- Identifies the component patterns that differentiate scan conditions
- Identifies the degree to which each pattern contributes to the differentiation
- Provides quantitative metrics of reproducibility and predictive power
- Quantifies the placement of an individual within a group response pattern
- Quantifies the influence of each individual upon group results.
Abiant has been able to use this approach to evaluate dose response, segregate task effects from
drug effects (example shown), differentiate scan acclimation effects from drug effects, distinguish
patients diagnosed with probable Alzheimer’s Disease from controls, and distinguish between inter-
site effects and the AD disease pattern.
At right: the two Canonical Variables
(CV, component patterns) that
differentiate 4 conditions: placebo with
a visual monitoring (“neutral”) task;
placebo with an active cognitive task
set; low dose methylphenidate with
neutral task; low dose methylphenidate
with active task. Each graph is
associated with an image pattern. The
task condition dominates CV1, while
the drug effect dominates CV2.
(CV, component patterns) that
differentiate 4 conditions: placebo with
a visual monitoring (“neutral”) task;
placebo with an active cognitive task
set; low dose methylphenidate with
neutral task; low dose methylphenidate
with active task. Each graph is
associated with an image pattern. The
task condition dominates CV1, while
the drug effect dominates CV2.
Technology
Advanced BioImaging to Accelerate New Therapeutics
 | |  |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| Copyright 2006-2007 Abiant, Inc. |