Using the artificial neural network to discriminate between normal controls with different APOE ε4 genotypes and probable AD cases in PIB-PET studies

Napatkamon Ayutyanont, Kewei Chen, Victor Villemagne, Graeme O'Keefe, Xiaofen Liu, Cole Reschke, Wendy Lee, Justin Venditti, Dan Bandy, Meixiang Yu, Stephanie Reeder, Christopher Rowe, Eric M. Reiman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

The standardized uptake value ratio (SUVR) provides a semi-quantitative index of fibrillar amyloid deposition, a neuropathological feature of Alzheimer's disease (AD), in Pittsburgh Compound B (PIB) PET studies. As accurately identifying individual probable AD patients and subjects at increased risk for AD is of clinical use, we developed the classification model based on SUVR of several ADassociated brain regions to distinguish normal subjects with different APOE genotypes, the major AD genetic risk factor, and probable AD patients. After normalizing PIB PET scans to standard brain template coordinate space, SUVR was computed for 8 brain regions: frontal, posterior cingulateprecuneus, lateral temporal, lateral parietal, and basal ganglia, medial temporal, occipital, and a mean cortical region (consisting of frontal, posterior cingulate-precuneus and lateral temporal region). These regions were defined using anatomical automated labeling toolbox in SPM5. 70% of the subjects were randomly partitioned for training, with remaining 30% for testing. The Artificial Neural Network (ANN) is then applied to the SUVR data to classify the subjects into three groups: low risk (APOE non-carriers), high risk (APOE carriers) and certain (probable AD patients). The process of data partitioning and ANN training/testing was repeated 3 times. ANN was found to classify the subjects into these three groups with the average accuracy of 100% and 95.2% in training and testing respectively. ANN is a promising multivariate-based alternative to discriminate among normal subjects with different APOE genotypes and probable AD patients and is potentially useful in evaluating the change in risk profile of the individual subject.

Original languageEnglish (US)
Title of host publication2009 ICME International Conference on Complex Medical Engineering, CME 2009
DOIs
StatePublished - 2009
Event2009 ICME International Conference on Complex Medical Engineering, CME 2009 - Tempe, AZ, United States
Duration: Apr 9 2009Apr 11 2009

Publication series

Name2009 ICME International Conference on Complex Medical Engineering, CME 2009

Other

Other2009 ICME International Conference on Complex Medical Engineering, CME 2009
CountryUnited States
CityTempe, AZ
Period4/9/094/11/09

ASJC Scopus subject areas

  • Biomedical Engineering
  • Health Informatics

Fingerprint Dive into the research topics of 'Using the artificial neural network to discriminate between normal controls with different APOE ε4 genotypes and probable AD cases in PIB-PET studies'. Together they form a unique fingerprint.

Cite this