Letters to the Editor
Recent eLetters to the Editor are available at radiology.rsnajnls.org. eLetters that are no longer posted under “Recent eLetters” can be found as a link in the related article or by browsing through past Tables of Contents.
Insufficient Spectral Quality to Draw Conclusions
From Chris Boesch, MD, PhD, Peter Vermathen, PhD, and Roland Kreis, PhD
Department of Clinical Research, University of Bern, Pavilion 52A Inselspital, PO Box 35, CH-3010 Bern, Switzerland e-mail: chris.boesch@insel.ch
Editor:
In the December 2007 issue of Radiol- ogy, Dr Faria and colleagues (1) re- ported results of proton magnetic reso- nance (MR) spectroscopy in 60 patients with adrenal masses.
Although we do not doubt the au- thors’ sincere effort to perform the spectroscopic investigations properly, we believe the studies in this article to be illustrative examples of problems that are frequently noted on clinical MR spectroscopic studies. There are fea- tures of the reported spectra that make it obvious that they should not have been used for any kind of diagnosis or study:
1. Figures 4-6 show strong arti- facts, which were likely caused by water filtering in postprocessing (peaks >3.5 ppm) and by lipid fold-in (peaks <2 ppm).
2. Almost all spectra show severe baseline problems.
3. Some of the interpreted reso- nances are barely above the noise level.
4. Most alarming is the dramatically incorrect phasing of the resonances. With proper phasing, all resonances should be in phase; however, in this study, they are all out of phase relative to each other. Obviously, automatic phasing failed owing to the strong arti- facts and baseline problems evident in figures 3-6.
5. Acquisition of spectra from the upper abdomen must always include
methods to cope with breathing motion, which admittedly is difficult, but which was not sufficiently addressed here. Spatial blurring, introduction of arti- facts, and point spread function-related lipid contamination are to be expected from chemical shift imaging data ac- quired during free breathing.
6. Peak assignment must be veri- fied. The creatine signal cannot be seen at 3.08 ppm when the choline signal is at 3.2 ppm.
7. Newly assigned or detected reso- nances, in particular their chemical shift positions, must be reproducible. This does not seem to be the case for the reported resonances, which are gener- ally referred to as peaks at 4.0-4.3 ppm.
We recognize the great effort of the authors in recruiting and examining 60 patients with an interesting pathologic condition. However, because Radiology is widely read and is an important plat- form for research on the diagnostic ap- plication of MR spectroscopy, it is es- sential that only high-quality spectral data are presented and that the review- ers of these types of manuscripts are highly experienced and technically so- phisticated. More rigorous review pro- cesses are needed in order to guarantee adequate quality standards for clinical spectroscopic studies.
Reference
1. Faria JF, Goldman SM, Szejnfeld J, et al. Ad- renal masses: characterization with in vivo proton MR spectroscopy-initial experience. Radiology 2007;245(3):788-797.
Response
From
Suzan M. Goldman, MD, PhD,* Juliano F. Faria, MD,* Jacob Szejnfeld, MD, PhD,* Cláudio E. Kater, MD, PhD,* and Homero Melo, MSc*
Department of Diagnostic Imaging* and Division of Endocrinology, Department of Medicine,+ Federal University of São Paulo, Napoleão de Barros, 800, Vila Clementino, São Paulo, SP, Brazil 04024-002 e-mail: smgold@terra.com.br
Since our article reported preliminary data (1), we acknowledge that we may have failed to avoid some pitfalls. Never- theless, these do not seem to have af- fected the overall results. The spectral data reported were from a larger data set and may not have been the best choice because the abdomen presents unique challenges for spectral acquisition.
We agree that water saturation is im- paired in figures 4-6, but the presence of the metabolite in the 4.0-4.3-ppm region is unambiguous, since both the metabo- lite and water peaks coexist in the same graph. Identification of this metabolite is currently under investigation with in vitro spectroscopy. In addition, to lessen the measurement error obtained from abso- lute values, we chose to employ metabo- lite ratios.
Several of the pheochromocytomas had cystic areas, and if water was a contaminant, we would possibly have found major peaks in the respective voxels. However, substantial 4.0-4.3- ppm peaks were seen in solid areas, but were absent in cystic ones.
Although creatine was measured at 3.03 ppm, both the low resolution of the 1.5-T equipment and the chemical-shift effect promote considerable modifica- tion of the curve.
We cannot agree with the comment regarding the curves’ phase because only lactate appears out of phase on spectra obtained by using a 135-msec echo time. In adrenal pathologic studies, lipid peaks are critical, but those of lactate do not have pathologic correlation. The continu- ous development of the postprocessing software allowed better adjustments of both the base line and the signal-to-noise ratio, which facilitated both acquisition and correction. It must be emphasized that we were required to adjust the spec- troscopy sequence and the postprocess- ing protocol from available brain and prostate data.
No doubt breathing also affected the quality of the results. The approach of acquiring spectra at maximum inspiration and expiration, which we have subse- quently implemented, has proved effec- tive in controlling respiratory artifacts.
Finally, the decision to submit our material to Radiology relied on this journal’s recognition of the potential value of material obtained from initial experiences, which may prove to be clinically useful in the future.
Reference
1. Faria JF, Goldman SM, Szejnfeld J, et al. Ad- renal masses: characterization with in vivo proton MR spectroscopy-initial experience. Radiology 2007;245(3):788-797.
Editor’s Note
From Herbert Y. Kressel, MD, Editor, Radiology
I wish to thank Drs Boesch and Gold- man and their colleagues for their thoughtful and constructive comments. Over the past year, I have received sev- eral other comments about the need to improve the quality of our peer review process in technically challenging areas such as MR spectroscopy. In response, we have added 15 MR spectroscopists to our reviewer panel in the past year to increase the number of MR spectrosco- pists available to us as reviewers. We have also added a number of other new reviewers in more technically challeng- ing areas of MR development. In re- viewing manuscripts on clinical applica- tions of emerging techniques we now commonly include both basic scientists and clinical investigators as reviewers.
We hope that these additions will provide authors with better feedback on their manuscripts and further improve the quality of our peer review process.