學術產出-Theses

Article View/Open

Publication Export

Google ScholarTM

政大圖書館

Citation Infomation

題名 利用深度學習演算法進行磁共振頻譜重建
Deep learning based MRS reconstruction
作者 江宗諭
YU, JIANG ZONG
貢獻者 蔡尚岳
Shang-Yueh Tsai
江宗諭
JIANG ZONG YU
關鍵詞 磁共振頻譜
深度學習
重建頻譜
日期 2021
上傳時間 3-Jan-2022 16:11:39 (UTC+8)
摘要 最近深度學習技術廣泛的應用在MRS 的研究上,例如使用卷積神經網路CNN
模型來去除雜訊或者移除基線等等,而本研究主要是在探討使用U-NET 模型來進行大腦頻譜的重建,U-Net 是一種卷積神經網絡(CNN)方法,他可以更好的分割生物醫學影像。先將大腦的模擬頻譜傅立葉轉換成FID 之後進行截斷,如果截斷後留下的點數為8 稱為tFID8,若留下16 的點稱為tFID16,以此類推,在進行傅立葉轉換獲得截斷光譜,藉由這些頻譜來訓練模型,一開始訓練了tFID2048、tFID1024、tFID512、tFID256、tFID128、tFID64、tFID32 、tFID16、tFID8,總共8 個模型,通過觀察不同模型的結果和比較,最終挑選tFID128 和tFID32 這兩個模型應用在活體頻譜上,結果因為水頻譜關係得到不好的結果,為了得到最好的結果,將tFID128 和tFID32 模型進行修改嘗試獲取最好的重建活體頻譜,
結果顯示在模擬頻譜的重建算是非常成功,但是應用在活體頻譜的重建上就不盡理想,所以在進一步的分析模擬頻譜與活體頻譜的誤差,並且將進行一些修正,並重新訓練,得知是因為模擬頻譜與活體頻譜不夠相似造成還原結果有些差異。
參考文獻 [1] Gujar, S. K., Maheshwari, S., Björkman-Burtscher, I., & Sundgren, P. C. (2005). Magnetic resonance spectroscopy. Journal of neuro-ophthalmology, 25(3), 217- 226.
[2 ] Dager, S. R., Oskin, N. M., Richards, T. L., & Posse, S. (2008). Research applications of magnetic resonance spectroscopy (MRS) to investigate psychiatric disorders. Topics in magnetic resonance imaging: TMRI, 19(2), 81.
[3] Gu, J., Wang, Z., Kuen, J., Ma, L., Shahroudy, A., Shuai, B., ... & Chen, T. (2018). Recent advances in convolutional neural networks. Pattern Recognition, 77, 354-377.
[4] Hatami, N., Sdika, M., & Ratiney, H. (2018, September). Magnetic resonance spectroscopy quantification using deep learning. In International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 467-475). Springer, Cham.
[5] Kiranyaz, S., Avci, O., Abdeljaber, O., Ince, T., Gabbouj, M., & Inman, D. J. (2021). 1D convolutional neural networks and applications: A survey. Mechanical systems and signal processing, 151, 107398.
[6] Birch, R., Peet, A. C., Dehghani, H., & Wilson, M. (2017). Influence of macromolecule baseline on 1H MR spectroscopic imaging reproducibility. Magnetic resonance in medicine, 77(1), 34-43.
[7] Deep learning based MRS quantification : CNN integrated with water scaling and partial volume correction
[8] Lee, H., Lee, H. H., & Kim, H. (2020). Reconstruction of spectra from truncated free induction decays by deep learning in proton magnetic resonance spectroscopy. Magnetic resonance in medicine, 84(2), 559-568.
[9] Ronneberger, O., Fischer, P., & Brox, T. (2015, October). U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention (pp. 234-241). Springer, Cham. [

[10] Lee, H. H., & Kim, H. (2019). Intact metabolite spectrum mining by deep learning in proton magnetic resonance spectroscopy of the brain. Magnetic resonance in medicine, 82(1), 33-48.

[11] Simpson, R., Devenyi, G. A., Jezzard, P., Hennessy, T. J., & Near, J. (2017). Advanced processing and simulation of MRS data using the FID appliance (FID‐ A)—an open source, MATLAB‐based toolkit. Magnetic resonance in medicine, 77(1), 23-33.
描述 碩士
國立政治大學
應用物理研究所
108755007
資料來源 http://thesis.lib.nccu.edu.tw/record/#G0108755007
資料類型 thesis
dc.contributor.advisor 蔡尚岳zh_TW
dc.contributor.advisor Shang-Yueh Tsaien_US
dc.contributor.author (Authors) 江宗諭zh_TW
dc.contributor.author (Authors) JIANG ZONG YUen_US
dc.creator (作者) 江宗諭zh_TW
dc.creator (作者) YU, JIANG ZONGen_US
dc.date (日期) 2021en_US
dc.date.accessioned 3-Jan-2022 16:11:39 (UTC+8)-
dc.date.available 3-Jan-2022 16:11:39 (UTC+8)-
dc.date.issued (上傳時間) 3-Jan-2022 16:11:39 (UTC+8)-
dc.identifier (Other Identifiers) G0108755007en_US
dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/138383-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 應用物理研究所zh_TW
dc.description (描述) 108755007zh_TW
dc.description.abstract (摘要) 最近深度學習技術廣泛的應用在MRS 的研究上,例如使用卷積神經網路CNN
模型來去除雜訊或者移除基線等等,而本研究主要是在探討使用U-NET 模型來進行大腦頻譜的重建,U-Net 是一種卷積神經網絡(CNN)方法,他可以更好的分割生物醫學影像。先將大腦的模擬頻譜傅立葉轉換成FID 之後進行截斷,如果截斷後留下的點數為8 稱為tFID8,若留下16 的點稱為tFID16,以此類推,在進行傅立葉轉換獲得截斷光譜,藉由這些頻譜來訓練模型,一開始訓練了tFID2048、tFID1024、tFID512、tFID256、tFID128、tFID64、tFID32 、tFID16、tFID8,總共8 個模型,通過觀察不同模型的結果和比較,最終挑選tFID128 和tFID32 這兩個模型應用在活體頻譜上,結果因為水頻譜關係得到不好的結果,為了得到最好的結果,將tFID128 和tFID32 模型進行修改嘗試獲取最好的重建活體頻譜,
結果顯示在模擬頻譜的重建算是非常成功,但是應用在活體頻譜的重建上就不盡理想,所以在進一步的分析模擬頻譜與活體頻譜的誤差,並且將進行一些修正,並重新訓練,得知是因為模擬頻譜與活體頻譜不夠相似造成還原結果有些差異。
zh_TW
dc.description.tableofcontents 致謝....1
摘要....2
目次...3
圖錄...5
表錄...7
第一章 緒論 .......... 8
1.1 磁共振頻譜............8
1.2 卷積神經網路............8
1.3 研究動機................9
第二章 研究方法............ 10
2.1 模擬光譜..............10
2.1.1大分子頻譜.............12
2.1.2 拓寬與噪聲.........14
2.2 截斷FID ..................16
2.3 U-NET……........20
2.4模型設定與評估...........20
2.5 數據預處理.........23
2.5 模型的評估..........23
第三章 研究結果.......... 24
3.1 模擬頻譜結果.......24
3.2 實體頻譜結果........36
3.2.1 略過水頻譜........38
3.3分析結果.................45
3.4修正模型................46
第四章 結論............. 49
第五章 參考資料......... 51
zh_TW
dc.format.extent 5921619 bytes-
dc.format.mimetype application/pdf-
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0108755007en_US
dc.subject (關鍵詞) 磁共振頻譜zh_TW
dc.subject (關鍵詞) 深度學習zh_TW
dc.subject (關鍵詞) 重建頻譜zh_TW
dc.title (題名) 利用深度學習演算法進行磁共振頻譜重建zh_TW
dc.title (題名) Deep learning based MRS reconstructionen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) [1] Gujar, S. K., Maheshwari, S., Björkman-Burtscher, I., & Sundgren, P. C. (2005). Magnetic resonance spectroscopy. Journal of neuro-ophthalmology, 25(3), 217- 226.
[2 ] Dager, S. R., Oskin, N. M., Richards, T. L., & Posse, S. (2008). Research applications of magnetic resonance spectroscopy (MRS) to investigate psychiatric disorders. Topics in magnetic resonance imaging: TMRI, 19(2), 81.
[3] Gu, J., Wang, Z., Kuen, J., Ma, L., Shahroudy, A., Shuai, B., ... & Chen, T. (2018). Recent advances in convolutional neural networks. Pattern Recognition, 77, 354-377.
[4] Hatami, N., Sdika, M., & Ratiney, H. (2018, September). Magnetic resonance spectroscopy quantification using deep learning. In International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 467-475). Springer, Cham.
[5] Kiranyaz, S., Avci, O., Abdeljaber, O., Ince, T., Gabbouj, M., & Inman, D. J. (2021). 1D convolutional neural networks and applications: A survey. Mechanical systems and signal processing, 151, 107398.
[6] Birch, R., Peet, A. C., Dehghani, H., & Wilson, M. (2017). Influence of macromolecule baseline on 1H MR spectroscopic imaging reproducibility. Magnetic resonance in medicine, 77(1), 34-43.
[7] Deep learning based MRS quantification : CNN integrated with water scaling and partial volume correction
[8] Lee, H., Lee, H. H., & Kim, H. (2020). Reconstruction of spectra from truncated free induction decays by deep learning in proton magnetic resonance spectroscopy. Magnetic resonance in medicine, 84(2), 559-568.
[9] Ronneberger, O., Fischer, P., & Brox, T. (2015, October). U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention (pp. 234-241). Springer, Cham. [

[10] Lee, H. H., & Kim, H. (2019). Intact metabolite spectrum mining by deep learning in proton magnetic resonance spectroscopy of the brain. Magnetic resonance in medicine, 82(1), 33-48.

[11] Simpson, R., Devenyi, G. A., Jezzard, P., Hennessy, T. J., & Near, J. (2017). Advanced processing and simulation of MRS data using the FID appliance (FID‐ A)—an open source, MATLAB‐based toolkit. Magnetic resonance in medicine, 77(1), 23-33.
zh_TW
dc.identifier.doi (DOI) 10.6814/NCCU202101736en_US