Mutations in isocitrate dehydrogenase (IDH) 1 and 2 in gliomas are associated with longer survival and better response to therapy than IDH wild-type tumors^1,2. Elevated in IDH-mutated gliomas, 2-hydroxyglutarate (2HG) is well established as a biomarker for diagnosis and prognosis in brain tumors. Recently, single voxel ¹H MRS has been proposed for in vivo detection of 2HG^3,4,5. There is a high interest in high-spatial resolution spectroscopic imaging of 2HG and other important metabolites in brain tumors. ¹H Echo Planar Spectroscopic Imaging (EPSI) is an effective tool for high-resolution imaging of brain metabolites at 3T and 4T^6,7. A recent study showed linear dependence of SNR and linewidth on field strength⁸. Hence, 7T EPSI may be advantageous for mapping low concentration metabolites. However, application of EPSI at 7T is strongly restricted due to small spectral width, which occurs in the conventional odd-even-echo reconstruction method⁹. Since the published data processing methods^10,11 for expanding EPSI spectral width are complicated and time consuming, a simple EPSI method with high spatial resolution may be highly beneficial for clinical application. Here we report a novel EPSI approach, a read-out gradient-alternated scheme, that can reliably measure the distribution of 2HG and other important metabolites at 7T with high spatial resolution, doubled spectral width, and low read-out gradient strength.The new EPSI method was tested in 1 normal adult subject and 2 IDH-mutated glioma patients. The volume prescription for EPSI acquisition was obtained with water-suppressed short-TE STEAM in normal subject, and with a previously-reported 2HG-optimized PRESS (TE1,TE2)=(58,20)ms¹² in glioma patients. To avoid the chemical shift artifact in the re-gridding process⁸, data were acquired only during the plateau of read-out gradient. Other scan parameters were as shown in Fig. 1. Data were acquired with a 32-channel head coil in a 7T whole-body scanner (Philips Medical Systems). A read-out gradient-alternated scheme was used for acquisition (Fig. 2a) and data were re-ordered during the first step of post-processing (Fig. 2b). The k-space data were zero-filled to 40x34 and 66x60, resulting in 0.25mL and 0.13mL voxel sizes in normal and glioma subjects, respectively. Data in the time domain were zero-filled to 4096 points. 1-Hz and 3-Hz exponential functions were applied to the time-domain data from normal and glioma subjects, respectively. Water-unsuppressed data were used for eddy-current compensation and multi-channel combination. Spectral fitting was performed, with LCModel software¹³, using in-house calculated basis spectra of 15 metabolites. Metabolites were quantified with reference to creatine in normal gray-matter region at 8mM. As shown in Fig. 2a, the polarities of echo-planar readout gradients for even- and odd-number acquisitions were opposite to each other. During data reconstruction, even-echo data from the two acquisitions were swapped to maintain the same acquisition gradient polarity in the time-domain data (Fig. 2b). This newly-proposed method resulted in two-fold greater spectral bandwidth than conventional odd-even-echo data processing method⁹, while preserving the SNR (data not shown). In addition, low read-out gradient strength and ramp-up rate (7.5-9.5mT/m and 90mT/m/ms, respectively), were used at 7T, which reduced gradient-caused artifact and acoustic noise¹⁴. Figure 3 shows EPSI spectra (voxel size 0.13mL) from the cortex region of normal subjects together with 6 metabolite maps and representative spectra. Metabolite distributions were in good agreement with prior MRSI studies¹⁵. The EPSI spectra from a glioma patient are shown in Fig. 4. A 2HG signal was clearly discernible at 2.25ppm in the spectra from tumor, while 2HG signal was null in the contralateral data (Fig. 4d). Metabolite maps showed that high 2HG, high total choline (tCho) and low NAA (Fig 4c) correlated with T₂w-FLAIR high-intensity region. Fig. 5 shows the EPSI data from a subject with tumor in the temporal region, which is often challenging for MRS imaging. The overall baseline of spectra was quite flat, and metabolite signals were well defined in all spectra without substantial artifacts. In the tumor region, 2HG distribution was heterogeneous between 0mM and 5mM, while tCho and NAA distributions were quite uniform (~1.4mM and ~2.5mM, respectively). We report a novel gradient-alternated EPSI readout scheme, together with the post-processing at 7T. This method doubles the EPSI spectral width, compared with the conventional method. With SNR benefit at 7T⁸, we demonstrated the capability of this new EPSI method for mapping 2HG and other important metabolites in gliomas with high spatial resolution, reasonable spectral width, and low gradient strength. In conclusion, 2HG and other brain metabolites can be reliably mapped with our proposed EPSI method at 7T. The method may provide an effective tool for studying potential alterations in metabolite levels in brain tumors and neuro-psychiatric diseases.