RETURN TO ABSTRACT LISTING


Background
MR Spectroscopy (MRS) should be viewed as a biochemical biopsy, and voxel sizes should be chosen appropriately. We have helped implement and perform clinical MRS at multiple institutions. MRS has provided added value for referring physicians, neurosurgeons, and oncologists.

Educational Goals / Teaching Points
Techniques for MRS: MRI studies are reviewed with the MR technologist prior to MRS acquisitions. Localizing 3D FLAIR series is obtained or a diagnostic MRI study with and without contrast. In areas of clinical concern and potential biopsy sites, short and long echo SV PRESS1 acquisitions (TE/TR 35/1500 ms and 288/1500 ms) are commonly used, with the number of averages scaled to voxel size. For voxels of ~2.6 cc at 3 T, 256 averages (6.4 min) are preferred; voxels of 4 cc at 128 scans (3.2 min) gives similar SNR. Accurate shimming is needed for good metabolite resolution. Spectra are processed using manufacturer software. To improve SNR, mild monoexponential filtering is applied prior to Fourier transform. Most recently, the multivoxel (MV) has been implemented on our 7-T scanner to optimize sampling the margins of a suspicious lesion. Interpretation of MRS: spectra are interpreted on a PACS using literature references. High grade vs low grade tumors, tumefactive MS vs tumor, benign gliosis vs low grade tumor, and radiation necrosis from tumor recurrence are interpretable. MRS for radiation necrosis vs pseudo progression and true progression is a significant portion of our referred cases.

Key Anatomic/Physiologic Issues and Imaging Findings/Techniques
We approach MRS studies as noninvasive biopsies, supervising our technologists and planning the voxel placement of tumor/lesion as clinically appropriate. In discussions with neurosurgeons and pathologists, nonenhancing tumors are best sampled in the core; higher grade enhancing WHO III-IV tumors are better sampled closer to the margins instead of necrotic areas. Just as one uses T1- and T2-weighted imaging for evaluation of CNS pathologies, our experience using short echo (TE 35), a T1/PD-weighted spectrum, and long echo (TE 288) T2-weighted acquisition improves the sensitivity and specificity because a priori pathology of a tumor/lesion is often not available. Short echo acquisitions are better at differentiating MS from tumors, CNS lymphoma, PML, and pediatric tumors; long echo MRS, gliosis vs low grade vs high-grade gliomas. Radiation necrosis vs recurrent tumor is a more difficult diagnosis due to microhemorrhage susceptibility and necrosis. Cases where MRI findings are suspicious but MRS spectra are benign, MRS and MRI permit physicians to follow these lesions safely.

Conclusion
With appropriate interaction with technologists, clinical MRS can be routinely utilized without research infrastructures. Voxel sizes should be selected as performing a biopsy. Voxel sizes,1–5 cc, are attainable on 3-T systems with high-quality interpretable spectra with Lorenzian filter postprocessing. Advances in 7-T technology provides additional resolution. The literature provides excellent references for accurate interpretation with appropriately chosen parameters.