PCADeNoise[data]
removes Rician noise from the data with PCA.
PCADeNoise[data,mask]
removes Rician noise from the data with PCA only withing the mask.
PCADeNoise[data,mask,sig]
removes Rician noise from the data with PCA only withing the mask using sig as prior knowledge or fixed value. Output is de {data denoise, sigma map} by default if PCAOutput is Full then fitted {data denoise , {sigma fit, average sigma}, {number components, number of fitted voxesl, number of max fits}, total fit -time per 500 ittt}. If PCARicianCorrect is True the denoised data is corrected for the Rician noise bias using RicianBiasCorrect with the fitted sigma map. Kernel voxels that are exactly zero (e.g. outside the image or already masked) are not used for the fit. If less than a quarter of the kernel contains data the voxel is not denoised and its sigma is the given sigma (0 if not given).
is based on DOI: 10.1016/j.neuroimage.2016.08.016 and 10.1002/mrm.26059.
PCADeNoise
PCADeNoise[data]
removes Rician noise from the data with PCA.
PCADeNoise[data,mask]
removes Rician noise from the data with PCA only withing the mask.
PCADeNoise[data,mask,sig]
removes Rician noise from the data with PCA only withing the mask using sig as prior knowledge or fixed value. Output is de {data denoise, sigma map} by default if PCAOutput is Full then fitted {data denoise , {sigma fit, average sigma}, {number components, number of fitted voxesl, number of max fits}, total fit -time per 500 ittt}. If PCARicianCorrect is True the denoised data is corrected for the Rician noise bias using RicianBiasCorrect with the fitted sigma map. Kernel voxels that are exactly zero (e.g. outside the image or already masked) are not used for the fit. If less than a quarter of the kernel contains data the voxel is not denoised and its sigma is the given sigma (0 if not given).
is based on DOI: 10.1016/j.neuroimage.2016.08.016 and 10.1002/mrm.26059.
Details and Options
- The following options can be given:
-
PCAKernel 5 PCAKernel is an option of PCADeNoise. It sets the kernel size. PCAOutput False PCAOutput is an option of PCADeNoise. If output is full the output is {datao, {output[[1]], sigmat}, {output[[2]], output[[3]], j}, timetot}. Else the output is {datao, sigmat}. PCATolerance 0 PCATolerance is an option of PCADeNoise and should be an integer >= 0. Default value is 0, which removes exactly the noise components that are found. Each increase of 1 keeps one more component, i.e. the method removes less noise. PCAWeighting False PCAWeighting is an option of PCADeNoise and can be True of False. Default value is False. When True the weights of the per voxel result are calculated based on the number of non noise components. Weighting is not possible with Parallelize True, which is the default, in that case PCAWeighting is ignored. PCAClipping True PCAClipping is an option of PCADeNoise and can be True of False. If True the output is clipped between 0.9 times the minimum and 1.1 times the maximum absolute value of the input data. PCAComplex False PCAComplex is an option of PCADeNoise and can be True of False. If set true the input data is expected to be {real, imag}. PCARicianCorrect False PCARicianCorrect is an option of PCADeNoise and can be True of False. Default value is False. If True the denoised data is corrected for the Rician noise bias with RicianBiasCorrect, using the fitted sigma. Not used if PCAComplex is True. Method "Similarity" Method is an option for various algorithm-intensive functions that specifies what internal methods they should use. MonitorCalc False MonitorCalc is an option for many processing functions. When true the process of the calculation is shown. Parallelize True Parallelize[expr] evaluates expr using automatic parallelization.