I TUOI DATI
I TUOI ORDINI
N. prodotti: 0
Totale ordine: € 0,00
I TUOI ABBONAMENTI
I TUOI ARTICOLI
Rivista di Medicina Interna
Indexed/Abstracted in: Current Contents/Clinical Medicine, EMBASE, PubMed/MEDLINE, Science Citation Index Expanded (SciSearch), Scopus
Impact Factor 1,236
Minerva Medica 2014 April;105(2):157-65
Automated 3D segmentation of hippocampus based on active appearance model of brain MR images for the early diagnosis of Alzheimer’s disease
Luo Z.-R. 1, Zhuang X.-J. 1, Zhang R.-Z. 1, Wang J.-Q. 1, Yue C. 2, Huang X. 1 ✉
1 Department of Radiology, The First Affiliated Hospital, Xiamen University, Xiamen, China;
2 Department of Neurology, The First Affiliated Hospital, Xiamen University, Xiamen, China
AIM: To investigate the hippocampal regional deformation modes by means of a novel method of automatic segmentation for discriminating between Alzheimer’s disease (AD) and normal aging; and to further provide the effective evidence for the early diagnosis of AD.
METHODS: Twenty AD patients and sixty healthy volunteers were included in this retrospective study. High-resolution structural volumetric images were obtained on a 3.0 T MR imaging system. Data were processed to create three-dimensional (3D) active appearance model (AAM) of hippocampus. Automatic recognition and 3D segmentation were carried out on both sides of the hippocampus in brain MR images of individuals with this model, and the hippocampal statistical shape model was established for AD group and control group. Student’s t test was used to identify whether there was difference between AD group and control group in the hippocampal regional deformation detected by automatic segmentation, and to compare whether there was difference between the automated segmentation and the manual tracing for quantifying hippocampal volumes on left/right side of the same sex group of healthy volunteers and if there was genderwise difference. Pearson’s Correlation test was employed to determine whether there was a correlation between automated segmentation and manual tracing for quantifying hippocampal volumes.
RESULTS: No significant difference was detected between automated segmentation and manual tracing for quantifying hippocampal volumes on left/right side of the same sex group of healthy volunteers (P>0.05). Further there was no significant genderwise difference (P>0.05). A very strong positive correlation existed between both methods for quantifying hippocampal volumes (denoted R2 near 1.0, P<0.001). Noticeable atrophy of bilateral hippocampal head was found among twenty patients with AD through statistical shape model compared with control group (P<0.05), especially on the left where inward-deformation was significantly found.
CONCLUSION: This novel method of automated segmentation of the hippocampus based on AAM has been found to be reliable and accurate in our study, which may be an alternative to manual segmentation. The featured atrophy of hippocampal head can be regarded as an important biomarker for the early diagnosis of AD.