REVIE\N ARTICLE Introduction to MR Hl spectroscopy P Corr MBChB, FFRad(D)SA, FRCR, MMed(UCT) Single voxel MR spectroscopy (MRS) of hydrogen protons is a use- ful investigation to characterize le- sions of the central nervous system. Imaging findings of mass lesions in the brain are often non-specific. Contrast enhancement indicates breakdown of the blood brain barrier and is often useful in diagnosing certain mass le- sions such as meningiomas. However, many ring enhancing lesions such as abscesses, granulomas and tumours cannot be differentiated using CT or MR imaging. Single voxel spectro- scopy of hydrogen protons can estab- lish the chemical spectrum of masses and assist in their characterization. Introduction Professor and Head: Department of Radi%gy, University of Nata/, Durban Methods MRS is essentially the same process as MR imaging, only the presentation of the data obtained is different.' In MR imaging the signal obtained in the time domain is used to produce a two-dimensional image, in MRS the signal in the time domain is used to produce a frequency domain spectrum of the molecules in the 12 SA JOURNAL OF RADIOLOGY· April 2000 voxel. Hl protons resonate at 63.86 MHz at 1.5 T Chemical bonds in and between molecules exist in the form of electron clouds around protons. These electron clouds have their own magnetic moment thus changing the net resonant frequency of each nu- cleus from its neighbour. This is called chemical shift and is the foundation of MRS. Chemical shifts are measured relative to the peak position of a ref- erence chemical. In the brain the ref- erence is N acetyl group of N acetyl aspartate (NAA), one of the brain's major constituents. This reference is set at 2 ppm. Molecules that can be detected by MRS are:' • Methyl groups CH3 - found in fatty acids, N acetyl aspartate, lactate • Methylene groups CH2 - found in fatty acids, glutamate, creatine and glycerol • Methyne group from alcohols found in glucose and myoinositol • Vinyl or olefinic protons of unsaturated fatty acids The concentration of most mole- cules in cells are very small while the concentration of water is very large by comparison, in the region of 10000 to 100 000 larger. Therefore the MR spectrum is totally dominated by wa- ter and the other molecules cannot be detected. Suppression of the water spectra is necessary to visualise the other molecules by MRS. Water is suppressed by two different methods: chemical shift selective method (CHESS) or water elimination Fourier transform method. To identify the metabolites or mol- ecules in a single voxel, two methods are used. Stimulated echo acquisition mode (STEAM) and point resolver spectroscopy method (PRESS) can be used. Both have their advantages and topage 13 Introduction to MR H I spectroscopy from page 12 disadvantages. STEAM is particularly useful for detecting the spectra of fatty acids and amino acids in small concentrations while PRESS is more accurate for detecting larger metabolite concentrations such as lac- tate, choline and N acetyl aspartate. We use the Probe software from GE Medical Systems which is semi-auto- mated. We find it works best using the PRESS technique where there is a better signal to noise ratio than STEAM. The voxels are Scm' in size (2x2x2 cm). Magnetic field homoge- neity is essential for good MRS. Inho- mogeneity leads to molecules having slightly different spin dephasing and broadening of the spectra. This makes it difficult to differentiate spectra from different metabolites. Care must be taken in placing the voxels on the brain scan. If the voxel is too close to the skull or ventricles, field homoge- neity is affected. The choice of time to echo (TE) is critical as it determines which metabolites are detected. Amino acids,long chain fatty acids and myoinositol are best detected at a short TE around 20 msec while choline and NAA and lactate at a 10ngTE of around 270 msec. Lactate has a double peak at TE of 135 msec and 270 msec due to a phenomenon called J coupling. Normal spectrum (Figure 1) NAA is the main peak at 2 ppm. This is due to the acetyl group and is a measure of neuronal density and viability? Any pathological process where neurones die will decrease the NAA peak. The next largest peak is due to creatine. This is a marker of energy metabolism within neurones and is diminished in malignant tumours. It is found at 3.03 ppm. Choline at 3.2 Figure 1: Normal H' MRS brain spectrum NAA- N acetyl aspartate, CH- choline, CR- creatine ppm is a building block of neuronal cell membranes and is an indicator of cell membrane turnover. It is in- creased in all tumours. Lactate is de- tected at 1.32 ppm. Normally lactate is not found. Lactate is present in anaerobic glycolysis where there is reduced blood oxygen and is seen when lesions outgrow their blood sup- ply or if there is ischaemia. With STEAM, metabolites with short re- laxation times such as myoinositol at 3.56 ppm and alanine can be detected. They are involved in neurotransmitter pro- duction. Multiple volume MRS two and three dimensions and both PRESS and STEAM methods can be used.' App-lications of single voxel MR spectroscopy (MRS) MRS using the single voxel tech- nique is useful in characterising intrac- erebral masses. We have used it suc- cessfully in patients with AIDS to de- termine whether lesions were inflam- matory or neoplastic before consid- ering treatment. Neoplasms will cause an elevated choline peak due to in- creased cell membrane production but decreased N acetyl aspartate and Multiple volume MRS is used to deter- mine spectra from mul- tiple voxels in the brain. This is also called chemical shift imaging (CSI). This allows comparison of spec- tra for multiple regions of a tumour especially on treatment. This is in one, 13 SAJOURNAL OF RADIOLOGY. April 2000 Figure 2a, b: Single voxel H' spectroscopy of non-Hodgkin's lymphoma of the left basal ganglia demonstrates an elevated cholme peak end diminished NAA and creatine to page 14 Introducl ion lo MR H I spectroscopy frompsg",3 creatine peaks from destruction of neu- rones. Occasionally elevated lactate peak in areas of ischaemia are detected within the tumour. Choline peaks are highest in tumours with high cellularity such as lymphoma and primitive neu- roectodermal tumours such as medul- loblastoma (Figure 2). MRS is useful in evaluating tumour response to ra- diotherapy or chemotherapy by moni- toring peaks. MRS in particular can distinguish between recurrent tumour and gliosis." Recurrent tumour will Figure 3a, b: Single voxel H' spectroscopy of e left basal ganglia tuberculoma demonstrates a merkedly elevated lactate peak and dIminIshed NAA and choline peaks using PRESS technique demonstrate elevated choline peaks while gliosiswill show depression of all metabolite peaks. Inflammatory lesions will have decreased NAA and creatine levels and increased lactate from anaero- bic glycolysis. Both tuberculomas, pyogenic and toxoplasma abscesses often have markedly elevated lactate Conclusions MRS is a relatively new tech- nique that provides additional infor- mation that is useful in the treat- ment of many patients with brain tu- mours, inflammatory diseases, meta- bolic disorders and demyelination. Figure 4a, b: Single voxel H' spectroscopy of a 9-year-old child's brain with advanced Canavan's Djsea~e with . marked white matter hyperintensity and cerebral atrophy. MRS shows markedly elevated NAA peak typIcal of thIs disorder peaks (Figure 3).5 Tuberculomas also have elevated levels of lipids and fatty acids, best detected on STEAM se- quences. Recent work shows that MRS is useful for differentiating cerebral ab- scesses from necrotic tumour," MRS has an important role in the diagnosis and monitoring the treatment response in certain metabolic brain dis- orders in children. These include dis- orders of lipid metabolism such as ad- renal leukodystrophy, mitochondrial disorders (Leigh' s Disease), and white matter disorders (Canavan's Disease)? (Figure 4). 14 SAJOURNAL OF RADIOLOGY- April 2000 References I. Kwock L. Localised MR spectroscopy. Neuroimaging CLinof NAmerica 1999;8:713-731. 2. Castillo M, Kwock L. Proton MR spectroscopy of common brain tumours. Neuroimaging CLin of N America 1999;8:733-752. 3. Salibi N, Brown M. In: Clinical MR spectroscopy. 1998. John Wiley, New York. 4. Castillo M, Kwock L. Proton MR spectroscopy of common brain tumours. Neuroimag CLin N America 1998;8:733-752. S. Cecil K, Lenkinski R. Proton MR spectroscopy in inflammatory and infectious brain disorders. Neuroimag Clin N America 1998;863-880. 6. Grand S, Passaro G, Ziegler A, et aL. Necrotic tumour vs brain abscess:importance of amino adds detected at H' spectroscopy- initial results. Radiology 1999;213:785-793. 7. Wang Z, Zimmerman R. Proton MR spectroscopy of paediatric brain disorders. Neuroimag Clin N America 1998;8:781-807.