Macromolecule Removal

Background

Macromolecule resonances are broad lines that overlap with the sharper metabolite resonances of interest. During acquisition, both a full spectrum containing all metabolites and macromolecules, as well as a macromolecule spectrum is acquired. The macromolecule spectrum must be subtracted from the full spectrum prior to quantification of metabolite signals.

Implementation

You can do macromolecule removal in the fitMAN Suite program.

Load Full spectrum into top window (right click on top window to activate)

• File > Data > and choose the file

Load Macro spectrum into middle window (right click on middle window to activate)

• File > Data > and choose the file

Direct Subtraction

• Arithmetic > Subtract: Manual/No Scale > Specify the correct scaling value (Scale factor for Varian 4T acquisition is 1.2)

The resultant spectrum appears in the bottom window. Save the data:

• Activate the third window by right clicking
• File > Save Active Window

(The macromolecule spectrum may be scaled prior to subtraction to account for the differences in T1 saturation if different repetition times are used to acquire the Full and Macromolecule spectra. At 4T, with TR = 4.2 for Macro and TR = 2.2 for Full acquisition, the scaling factor should be 1.2)

Advanced Option

Prior to subtracting the macromolecule spectrum, it can be fit by SVD. The result can then be subtracted from the Full spectrum. The advantage of this method is that there is no additional random noise added into the resultant spectrum. However, there is uncertainty associated with the SVD fit, resulting in no net benefit to metabolite quantification precision in the final result.

SVD Fit

• Arithmetic > HLSVD Fit

The parameters of the HSVD fit are as follows:

• Number of Points: The number of initial points along the FID to use in forming the Hankel Matrix
• Hankel Matrix size: The number of columns of the Hankel Matrix
• Signal Related Singular Values: The number of Lorentzian peaks to fit
• Tolerance: The number of peaks to retain based on the value of the singular values – the specified value represents a fraction of the largest singular value. In this example, either 20 peaks are retained, fewer peaks if there are less that have singular values > 1% of the maximum singular value.
• Xmin, Xmax: the range of frequencies to use for data reconstruction