DocumentCode
2522014
Title
DISCRIMINATIVE ANALYSIS OF NEUROMYELITIS OPTICA USING TWODIMENSIONAL HISTOGRAM FROM DIFFUSION TENSOR IMAGING
Author
Lin, Fuchun ; Yu, Chunshui ; Jiang, Tianzi ; Li, Kuncheng ; Chan, Piu ; Lei, Hao
Author_Institution
Nat. Lab. of Pattern Recognition, Chinese Acad. of Sci., Beijing
fYear
2007
fDate
12-15 April 2007
Firstpage
752
Lastpage
755
Abstract
Neuromyelitis optica (NMO) generally has no visible brain lesions. However, occult damage of brain tissue has been investigated through magnetization transfer imaging and diffusion tensor imaging (DTI). In this paper, we discriminate patients with NMO from healthy subjects based on the two-dimensional histogram of apparent diffusion coefficient (ADC) and fractional anisotropy (FA) of the brain derived from DTI. The correct recognition rate is 89.5%, validated by the leave-one-out method. This result was much higher than that using one feature separately (68.4% for FA, 73.7% for ADC). Eigenimages, which reflect the discriminative pattern of NMO patients from healthy subjects, are also generated. Diffuse damage in brain tissue of NMO is detected from these eigenimages. In conclusion, the information from two-dimensional histogram can discriminate NMO patients from healthy subjects. Moreover, eigenimages can to some extent reflect discriminative regions between NMO patients and normal controls, and these discriminative regions will be useful in understanding the mechanism of NMO.
Keywords
biodiffusion; biomedical MRI; brain; diseases; apparent diffusion coefficient; diffusion tensor imaging; eigenimages; leave-one-out method; neuromyelitis optica; recognition; tissue has; two-dimensional histogram; Anisotropic magnetoresistance; Brain; Diffusion tensor imaging; Geometrical optics; Histograms; Image analysis; Lesions; Magnetic analysis; Magnetization; Optical imaging;
fLanguage
English
Publisher
ieee
Conference_Titel
Biomedical Imaging: From Nano to Macro, 2007. ISBI 2007. 4th IEEE International Symposium on
Conference_Location
Arlington, VA
Print_ISBN
1-4244-0672-2
Electronic_ISBN
1-4244-0672-2
Type
conf
DOI
10.1109/ISBI.2007.356961
Filename
4193395
Link To Document