Field map - measure field inhomogeneity, take a minute to acquire, help undistort images taken with Echo planar technique, Distortion correction - shimming coils at acquisition but they leave inhomogeneities so a field map can be recorded , discarded data acquisition - volumes at the first 10 seconds of the run result in a lot of noise which the scanner can delete itself, set the origin of the images - 0 point set should be at the middle of anterior commissure to help with later steps such as normalisation, realignment can help correct for - head motion by aligning two image volumes and subsequent images are registered to a single reference volume, realignment assumes that - size and shape of two objects are identical, as a reference image - the first image can be used or the average of all images, cost function - determines how well one image matches another, you want to minimise it, rigid body transformation means - 3 rotations, 3 translations, even after realignment - 90% of variance is due to movement effects, realignment has 2 steps - registration and re-slicing , this variance after realignment needs to be - incorporates as movement parameters as confounds into the statistical model, specificity - ability to correctly identify voxels which are not active, sensitivity - ability to correctly identify brain activations if activity is correlated with movement , motion correction can lead to - loss of info on top and bottom of the image-> solved with prospective motion correction, Slice timing correction - slices are collected one after the other (almost a full TR away), excitation of 1 can carry over to the next so this step interpolates the data as if all volume was acquired at the same time, slices can be collected - in interleaved manner (ascending or descending) or in odd/even sequence, co-registration step - alignment of anatomical and functional images in the same subject: if images acquired in different sessions or the subject moved,

Realignment

打印
嵌入

排行榜

视觉风格

选项

切换模板

恢复自动保存：？