The Adelaide MR Web Site MR Spectroscopy Case Study Lactate or Lipid?    A case of  ambiguous phase	click me to mail me
Prepared by Greg Brown    Royal Adelaide Hospital North Terrace Adelaide South Australia 5000 September 15th 2000	Back to MRS Page

This case raises a few questions of post processing technique, but also illustrates that body MRS isn't as difficult as we might think.
Presentation
Male with pre-sacral mass known to be an adenocarcinoma that has been treated with radiotherapy some time earlier.  Symptoms of nerve compression were returning.  No previous imaging available to assess recurrence of morphological criteria.
 

T2 axial (TSE) Ovoid mass with mixed moderate T2 intensity but no obvious fat internal	STIR coronal  some suggestion of fat in the lesion

Figure 1-1	In this case we want to identify unusual choline levels that may indicate recurrence, and any lactate that may indicate cell breakdown in the tumour.   Lipid peaks have the potential to mask information in the 0-2 ppm range.  The voxel was placed to avoid obvious fat, but no other specific fat suppression method was available.  The figure to the left is the spectra as automatically processed.  The large peak around 1.4 ppm dominates the scale, so the obvious question is how to fix this spectrum to make it  useful.  I had two aims:  correct the baseline to make the peak unipolar, rescale to better observe choline
Figure 1-2	Figure 1-3
Fig 1-2.  From the auto processed spectrum (fig. 1-1) I identified the region 1.1~1.9 ppm for exclusion from baseline correction, then made slight adjustments to the phase to display this broad peak.   Initially I accepted this as lipid contamination but the bifid top made me consider if it represented lactate doublet.	Fig 1-3.  Taking the spectrum of fig. 1-2, I adjusted the phase correction and easily made the broad peak appear as an inverted doublet.  Question 1: Which is the correct phase representation of this peak?

The body spectra have no NAA, so there is no obvious positive peak to decide the phasing question or assess the actual shim quality through observing its line width.  Additionally the lack of the NAA peak at 2 ppm confounds the frequency fit step that is used to bring the metabolite peaks to their standard positions.

I decided to use the residual water peak at 4.7 ppm to determine the correct phase solution.  I used the display control to increase the X axis range to 0~6ppm. Unfortunately the water suppression pulse was near perfect and I couldn't see residual water.  Now what ?
By turning off the water referencing step in post processing (click inactive) the residual water peak became visible (fig. 1-4)
 

Figure 1-4	Figure 1-5
Fig.1-4 Spectral scale set at 0 ~ 6 ppm and water referencing switched off.	Fig 1-5 The water peak was identified as a region to eliminate from baseline correction. At this scale setting the suspicious peak is difficult to identify.

.

Figure 1-6	Figure 1-7
Fig 1-6 The phase correction was adjusted to give the best representation of the positive residual water peak. The suspicious peak is clearly inverted relative to water.  Note that the residual water peak is at 4.8 ppm, 0.1 ppm more than its text book location.   This is due to the lack of NAA as a spectral reference.	Fig 1-7  With the water referencing back on the spectrum looses the residual water peak and re-scales the Y axis.  The mystery peak has lost its doublet shape.

Figure 1-8	Figure 1-9
Fig. 1-8 The doublet shape returns with very minor phase correction (5 degrees)	Figure 1-9  When the scale is returned to the standard 0~4.3 ppm the peak appears to centre on 1.4 ppm, placing it too high for lactate, but after correcting for the scale error noted in figure 1-6. it would be well placed as a strong Lactate resonance.

Discussion
Is the peak really lactate?  I could be convinced either way.
 

The evidence against it being Lipid rather than Lactate is: The amplitude and width of the peak is more in keeping with lipids. There is some question about the phasing methods I used here. The images suggest some fat in the voxel.

The evidence for it being Lactate is the doublet shape and chemical shift are correct for Lactate if the method is valid, the peak is inverted compared to water as would be expected at an echo time of 135 msec the lactate could be expected in a necrotic mass.

What do you think?   I am very interested to hear from more experienced MR spectroscopy users.  As comments come in I will include them here and develop a definitive answer to the problem.