Detection of Time Series Signal Variations in Functional Magnetic Resonance Imaging
碩士 === 大葉大學 === 工業工程研究所 === 90 === After functional magnetic resonance technology was discovered in 1991, researchers can acquire functional brain images and analyze those images to map human brain activation, This Thesis applies image smoothing technology and statistical signal processing technolog...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2002
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/11071929849790986137 |
| Summary: | 碩士 === 大葉大學 === 工業工程研究所 === 90 === After functional magnetic resonance technology was discovered in 1991, researchers can acquire functional brain images and analyze those images to map human brain activation, This Thesis applies image smoothing technology and statistical signal processing technology to detect the human brain activation area.
The experimental results indicate that under the criterion of maximum area of ROC curve, combining three dimensional space smoothness and cross-correlation of box-car reference function achieve the best performance.
Under the criterion of minimum False Positive Fraction of ROC curve, the following two combinations both achieve the best performance.
1. one dimensional time smoothness preceded by two dimensional space smoothness and cross-correlation of box-car reference function.
2. one dimensional time smoothness preceded by two dimensional space smoothness and cross-correlation of box-car reference function convoluted with one dimensional Gaussian mask.
|
|---|