為了產生更多樣性的腦內頻譜，除了各種參數的變化還有產生大量的資料集，所以我們利用模擬的方式以最少的時間產生出非常大量的資料。本研究方法是利用FID-A此工具模擬出人體腦中的純代謝物頻譜，再將所有代謝物的資訊整理成一個basis set，利用此basis set，乘上代謝物的濃度我們可以產生出完整代謝物的頻譜，利用此頻譜，再加上真實腦內頻譜掃描時所可能產生的影響，譬如線寬大小、大分子對基線的影響還有雜訊的成分，模擬人體腦中得到的頻譜。模擬出來的資料，可以應用在深度學習所需的大量訓練集上。最後我們利用相關係數和均方誤差去比較模擬的頻譜與in vivo頻譜的差異，在模擬的5萬筆資料當中，我們找出相關係數最大以及均方誤差最小的頻譜來展示最接近 in vivo頻譜的結果。

In order to generate a more diverse in vivo human brain metabolite spectra, in addition to the variation of various parameters, a large number of data sets needs to be generated, so we use the simulation method to generate a very large amount of data in the least amount of time. The method of this research is to use the FID-A toolkit to simulate the pure metabolite spectrum in the human brain and then organize the information of all metabolites into a basis set. Using this basis set, we can multiply the concentration of the metabolites to generate an intact spectrum of metabolites. Using this spectrum, add the possible effects of spectral scanning in the real brain, such as the size of the line width, the influence of macromolecules on the baseline, and the components of noise, simulate the spectrum obtained in the human brain. The simulated data can be applied to a large number of training sets required for deep learning. Finally, we use the correlation coefficient and mean square error to compare the difference between the simulated spectra and the in vivo spectra. Among the 50,000 simulated data, we find the spectrum with the largest correlation coefficient and the smallest mean square error to show the results closest to the in vivo spectrum.

References
[1] Cecil KM. Proton magnetic resonance spectroscopy: technique for the neuroradiologist. Neuroimaging Clin N Am. 2013;23(3):381-392.
[2] Tkác I, Oz G, Adriany G, Uğurbil K, Gruetter R. In vivo 1H NMR spectroscopy of the human brain at high magnetic fields: metabolite quantification at 4T vs. 7T. Magn Reson Med. 2009 Oct;62(4):868-79.
[3] Kreis R, Hofmann L, Kuhlmann B, Boesch C, Bossi E, Hüppi PS. Brain metabolite composition during early human brain development as measured by quantitative in vivo 1H magnetic resonance spectroscopy. Magn Reson Med. 2002 Dec;48(6):949-58.
[4] Haga KK, Khor YP, Farrall A, Wardlaw JM. A systematic review of brain metabolite changes, measured with 1H magnetic resonance spectroscopy, in healthy aging. Neurobiol Aging. 2009 Mar;30(3):353-63.
[5] Keevil SF. Spatial localization in nuclear magnetic resonance spectroscopy. Phys Med Biol. 2006 Aug 21;51(16):R579-636.
[6] Hamilton G, Middleton MS, Bydder M, Yokoo T, Schwimmer JB, Kono Y, Patton HM, Lavine JE, Sirlin CB. Effect of PRESS and STEAM sequences on magnetic resonance spectroscopic liver fat quantification. J Magn Reson Imaging. 2009 Jul;30(1):145-52.
[7] Klose U. Measurement sequences for single voxel proton MR spectroscopy. Eur J Radiol. 2008 Aug;67(2):194-201.
[8] Simpson R, Devenyi GA, Jezzard P, Hennessy TJ, Near J. Advanced processing and simulation of MRS data using the FID appliance (FID-A)-An open source, MATLAB-based toolkit. Magn Reson Med. 2017 Jan;77(1):23-33.
[9] Mullins PG, McGonigle DJ, O'Gorman RL, Puts NA, Vidyasagar R, Evans CJ; Cardiff Symposium on MRS of GABA, Edden RA. Current practice in the use of MEGA-PRESS spectroscopy for the detection of GABA. Neuroimage. 2014 Feb 1;86:43-52.
[10] Harris AD, Saleh MG, Edden RA. Edited 1 H magnetic resonance spectroscopy in vivo: Methods and metabolites. Magn Reson Med. 2017 Apr;77(4):1377-1389.
[11] Lee HH, Kim H. Intact metabolite spectrum mining by deep learning in proton magnetic resonance spectroscopy of the brain. Magn Reson Med. 2019 Jul;82(1):33-48.
[12] Govindaraju V, Young K, Maudsley AA. Proton NMR chemical shifts and coupling constants for brain metabolites. NMR Biomed. 2000 May;13(3):129-53.
[13] Huang yu long. Deep learning based MRS quantification : CNN integrated with water scaling and partial volume correction.
[14] Rebecca Birch, Andrew C. Peet, Hamid Dehghani, Martin Wilson. Influence of macromolecule baseline on 1H MR spectroscopic imaging reproducibility. Magn Reson Med. 2017 Jan; 77(1): 34–43.