Premium
IC‐P‐027: Tracking brain ventricle expansion in Alzheimer's disease using combined intensity and shape‐based segmentation
Author(s) -
Drozd John,
Hayes Alex,
Moreland Robert,
Khan Amanda,
Borrie Michael,
Bartha Robert
Publication year - 2012
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2012.05.059
Subject(s) - segmentation , ventricle , magnetic resonance imaging , brain size , imaging phantom , medicine , third ventricle , nuclear medicine , artificial intelligence , computer science , radiology
50% (and t value >3) of the peak t-value in the VOI. Average normalized FDG counts (NFC) were extracted. AD and MCI NFC obtained with the 2 approaches were compared to NCo NFC mean values. Cutoff points sensitivity and specificity, post-test likelihood ratios and p-values were assessed. Results: We studied 49 AD, 19 MCI, 9 NCo (no significant demographic difference found between groups). As compared to NCo, ANOVA of anatomical VOIs showed bilaterally reduced CG, pCu and IPC NFC in AD (P<0.001); and bilaterally reduced CG and pCu NFC in MCI (P<0.001). Probabilistic VOIs showed bilaterally reduced PCC, IPC and HIP NFC in AD (P<0.01); and bilaterally reduced PCC and IPC and left HIP NFC inMCI (P<0.001).When comparing patients to NCo, the analysis of HIP VOIs yielded no statistically significant results with the anatomically defined VOIs, but significance was reached with the probabilistically definedVOIs, except in the right HIPVOI ofMCI individuals.Conclusions: Metabolic changes in AD and MCI seem not to follow strict structural boundaries for the PCC, IPC and especially HIP VOIs. Probabilistic VOIs might therefore better detect the metabolic signature of AD pathology at different stages of the disease.