本研究使用自行開發之中場核磁共振系統以及ez-SQUID公司開發的中場核磁共振系統，進行肝臟切片生檢法與細針抽吸之組織重量模擬量測，量測20管的正常組織(Normal Tissue)與20管的腫瘤組織(Tumor Tissue)。測量肝臟切片生檢法檢體重量約0.075 g~0.125 g、細針抽吸檢體重量約0.009 g~0.012 g，不經任何組織染色處理直接測量，量測完訊號使用快速傅立葉轉換法(FFT頻譜)以及本研究提出的強度法(Power)來分析擬合出T1弛豫時間(Relaxation time)，並比較使用這兩種分析方法所測量正常組織與腫瘤組織之T1弛豫時間的差異。利用T1弛豫時間來驗證細針抽吸在微量的狀態下是否也能進行腫瘤分辨的可行性。
最後測量結果發現使用強度法分析，穩定度以及準確度最好，也發現腫瘤組織的T1值大於正常組織，顯示確實能利用T1值來區分腫瘤組織。而在測量肝臟切片生檢法以及細針抽吸的靈敏度與特異度都相同，分別為85 %、100 %。說明了肝臟細針抽吸檢測上，能夠與肝臟切片生檢法有相同的檢測結果，因此證明本系統應用在細針抽吸也能清楚區分出腫瘤組織，可提供醫師們一個參考資訊。並利用AI機器學習邏輯回歸(Logistic Regression)模型將T1值用來分類並預測出正常與腫瘤組織可能的機率，提供一個區分腫瘤組織依據。

關鍵字：中場核磁共振、肝臟切片生檢、細針抽吸、T1弛豫時間、靈敏度、特異度、機器學習、邏輯回歸

In this study, the self-developed Mid-Field NMR system and the Mid-Field NMR system developed by ez-SQUID were used to perform Liver biopsy and Fine-needle aspiration tissue weight analog measurement, measuring 20 tubes of normal tissue and 20 tubes of tumor tissue. Measure the weight of the liver bioassay method about 0.075 g~0.125 g, and the Fine-needle aspiration sample weight of about 0.009 g~0.012 g, directly measurement without any tissue staining. After measurement, fast Fourier transform method and Intensity method (Power) proposed in this study are used to analyze and fit T1 relaxation time, and compare the use of these two methods. The difference in T1 relaxation time between normal tissue and tumor tissue measured by this analytical method. The T1 relaxation time was used to verify the feasibility of Fine-needle aspiration for tumor discrimination even in the state of trace amounts.
In the final measurement results, it was found that the intensity method analysis had the best stability and accuracy. It was also found that the T1 value of tumor tissue was greater than that of normal tissue, indicating that T1 value could indeed be used to distinguish tumor tissue. The sensitivity and specificity of liver biopsy and Fine-needle aspiration were the same, 85 % and 100 %, respectively. It is explained that the detection of liver Fine-needle aspiration can have the same detection results as the biopsy method of liver biopsy. Therefore, it is proved that the system can clearly distinguish tumor tissue when it is applied to Fine-needle aspiration, which can provide doctors with a reference information. And use the AI machine learning logistic regression model to classify the T1 value and predict the possible probability of normal and tumor tissues, providing a basis for distinguishing tumor tissues.

Keyword：Mid-field NMR、Liver biopsy、Fine-needle aspiration、T1 relaxation time、Sensitivity、Specificity、machine learning、logistic regression

[1]衛生福利部 109年國人死因統計結果, 2021年11月20日取自於衛生福利部網頁https://www.mohw.gov.tw/cp-5017-61533-1.html
[2]Michael Müück, “Manual for the NMR Dmonstrator type ez SQUID NMR-1.”, Mess-und Analysegerääte, Germany, 6 March 2018
[3]M .A. Bernstein, K. F. King and X. J. Zhou, “Handbook of MRI Pulse Sequences.＂, Elsevier Academic Press, 960 (2004)
[4]陳彥呈 (2018)。《低場核磁共振系統於乳癌組織檢測應用與特性研究》。 國立台灣師範大學 光電科技研究所碩士論文 台北市 。
[5]廖宇庭(2019)。《變場及低場核磁共振系統於人類肝癌組織檢測應用與特性研究》。國立臺灣師範大學光電科技研究所碩士論文，台北市。
[6]Damadian, R., “Tumor detection by nuclear magnetic resonance.”, Science 171,3976 (1971)
[7]徐嘉敏 (2019)。《 核磁共振系統應用於肝癌檢體之應用研究 》。 國立臺灣師範大學光電科技研究所碩士論文，台北市。
[8]肝病等不起 解析查肝4種醫學影像手段, 2021年12月20日取自大紀元健康電子報https://www.epochtimes.com/b5/17/4/9/n9020326.htm
[9]Rockey, D. C., Caldwell, S. H., Goodman, Z. D., Nelson, R. C. and Smith, A. D, “Liver biopsy”, Hepatology, 49: 1017-1044 (2009)
[10]了解肝臟穿刺,2021年12月20日取自於台大醫院健康電子報https://epaper.ntuh.gov.tw/health/201906/health_2.html
[11]Chow, A. M., Gao, D. S., Fan, S. J., Qiao, Z., Lee, F. Y., Yang, J., Man, K., & Wu, E. X., “Measurement of liver T1 and T2 relaxation times in an experimental mouse model of liver fibrosis”, Journal of magnetic resonance imaging,36(1):152-158. (2012)
[12]Onitilo, Adedayo A et al., “Breast cancer subtypes based on ER/PR and Her2 expression: comparison of clinicopathologic features and survival.”, Clinical medicine & research vol. 7,1-2 (2009): 4-13.
[13]E. S. Pearson, “ 'Student' as Statistician ”, Biometrika vol.30,Issue3-4, Pages 210–250. (1939)
[14]de Winter, J.C.F., “Using the Student's t-test with extremely small sample sizes.”, Practical Assessment, Research, and Evaluation: Vol. 18, Article 10. (2013)
[15]Hanley, J A., “Receiver operating characteristic (ROC) methodology: the state of the art.”, Critical reviews in diagnostic imaging vol. 29,3 (1989): 307-35.
[16]Swets, J.A., Pickett, R.M., “Evaluation of diagnostic systems: Methods from signal detection theory.”, Academic Press, New York. (1982)
[17]Obuchowski, Nancy A, and Jennifer A Bullen.,“Receiver operating characteristic (ROC) curves: review of methods with applications in diagnostic medicine.”, Physics in medicine and biology vol. 63,7 07TR01. 29 Mar. 2018
[18]Mitchell, T., “Machine Learning”, McGraw Hill. ISBN 0-07-042807-7. (1997).
[19]Logistic Regression in Machine Learning, 2022年1月2日取自於javatpoint https://www.javatpoint.com/logistic-regression-in-machine-learning
[20]Sarvagya, A., “Logistic Regression- Supervised Learning Algorithm for Classification”, Analytics Vidhya, May 23, 2021
[21]Joseph P. Hornak, “The Basics of MRI”, Rochester Institute of Technology, (2006)
[22]S. Appelt, A. Ben-Amar Baranga, C.J. Erickson, M.V. Romalis, A.R.Young, W. Happer, “ Theory of spin-exchange optical pumping of 3He and 129Xe ”, Phys. Rev. A 58, 1412 (1998)
[23]周彥廷 (2014)。《 低磁場核磁共振梯度接收線圈應用於肝腫瘤組織檢驗 》。 國立臺灣師範大學光電科技研究所碩士論文，台北市。