| dc.contributor.advisor | 蔡尚岳 | zh_TW |
| dc.contributor.advisor | Tsai, Shang-Yueh | en_US |
| dc.contributor.author (Authors) | 張順詠 | zh_TW |
| dc.contributor.author (Authors) | Chang, Shuenn-Yeong | en_US |
| dc.creator (作者) | 張順詠 | zh_TW |
| dc.creator (作者) | Chang, Shuenn-Yeong | en_US |
| dc.date (日期) | 2026 | en_US |
| dc.date.accessioned | 2-Mar-2026 12:35:37 (UTC+8) | - |
| dc.date.available | 2-Mar-2026 12:35:37 (UTC+8) | - |
| dc.date.issued (上傳時間) | 2-Mar-2026 12:35:37 (UTC+8) | - |
| dc.identifier (Other Identifiers) | G0112755008 | en_US |
| dc.identifier.uri (URI) | https://nccur.lib.nccu.edu.tw/handle/140.119/161878 | - |
| 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) | 112755008 | zh_TW |
| dc.description.abstract (摘要) | 本研究探討磁共振頻譜(Magnetic Resonance Spectroscopy, MRS)資料處理中,於不同平均次數下頻譜量化的效果,並提出利用卡爾曼濾波(Kalman Filter)改善量化穩定度,結果與傳統窗函數方法(漢明窗與指數窗)進行比較。MRS 本身具低訊號雜訊比的訊號特質,在資料擷取時通常需要透過增加平均來降低雜訊,進而改善量化的穩定度,研究中設計了四種動態測量共變異數 R 模型(Sigmoid、Square root、Log 與 Linear),調節卡爾曼濾波的濾波強度,以期在不增加掃描時間的情況下,改善頻譜品質與代謝物定量穩定性。本研究的頻譜資料是透過 3T MRI 系統收集來自健康成人視丘(Thalamus)區域的頻譜資料,分析時會將資料重組成平均次數分別為 2、4、8、16、32、64 次的頻譜,代表不同雜訊等級。評估頻譜品質的參數為訊號雜訊比(SNR)、半高全寬(FWHM),量化後的化合物濃度會透過濃度變異係數(CV)與 Cramér- Rao Lower Bound(CRLB),評估其穩定度,針對六個的代謝物進行評估。結果顯示,卡爾曼濾波與窗函數在低平均次數時均可提升頻譜品質與定量穩定性,並優於未經處理的原始資料,然而卡爾曼慮波與傳統窗函數相比並無顯著差異,於平均次數較多頻譜 (>32) 三者表現接近,儘管如此,卡爾曼濾波具備可調性,未來研究可進一步優化模型設計,結合其他時頻域技術進行混合式濾波,強化其在低平均條件下的穩定性。另外本研究發現,頻譜在 16 次平均次數,訊號雜訊比到達到一定程度後,在化合物量化時已具一定的穩定性,顯示額外增加訊號雜比對量化的穩定度改善有限,可做為未來在有限時間內取擷取頻譜資料時的參考。 | zh_TW |
| dc.description.abstract (摘要) | This study investigates the effects of spectral quantification under different numbers of signal averages in Magnetic Resonance Spectroscopy (MRS) data processing and proposes the use of Kalman filtering to improve quantification stability. The performance of the proposed method is compared with conventional windowing approaches, including Hamming and exponential windows. Due to the intrinsically low signal-to-noise ratio (SNR) of MRS signals, signal averaging is commonly employed during data acquisition to reduce noise and enhance quantification stability. In this study, four dynamic measurement noise covariance R models (Sigmoid, Square root, Log, and Linear) were designed to regulate the filtering strength of the Kalman filter, aiming to improve spectral quality and metabolite quantification stability without increasing scan time. Spectral data were acquired from the thalamus region of healthy adults using a 3T MRI system. The acquired data were retrospectively reconstructed into spectra with averaging numbers of 2, 4, 8, 16, 32, and 64, representing different noise levels. Spectral quality was evaluated using signal-to-noise ratio (SNR) and full width at half maximum (FWHM). Quantification stability was assessed using the coefficient of variation (CV) of metabolite concentrations and the Cramér–Rao lower bound (CRLB) for six metabolites. The results demonstrate that both Kalman filtering and conventional windowing methods improve spectral quality and quantification stability at low averaging numbers compared with unprocessed raw data. However, no significant differences were observed between Kalman fil-
tering and windowing methods. At higher averaging numbers (>32), all three approaches exhibited comparable performance. Despite this, Kalman filtering offers greater flexibility through parameter adjustability, and future work may further optimize model design by integrating hybrid time–frequency domain techniques to enhance stability under low averaging conditions. In addition, this study found that once the SNR reached a sufficient level at 16 signal averages, metabolite quantification exhibited reasonable stability, indicating that further increases in SNR provided limited additional benefits. This finding may serve as a practical reference for MRS data acquisition under time-constrained conditions. | en_US |
| dc.description.tableofcontents | 第一章 緒論 1
1.1 研究動機 1
第二章 研究方法 3
2.1 研究資料 3
2.2 頻譜品質 3
2.3 LCModel濃度量化 4
2.4 窗函數（Window function） 6
2.5 卡爾曼濾波（Kalman Filter） 7
2.5.1 預測步驟（Prediction Step） 8
2.5.2 更新步驟（Update Step） 9
2.5.3 R 測量共變異數（Measurement Covariance） 10
第三章 研究結果 14
3.1 不同 R 測量共變異數輸出結果 14
3.1.1 頻譜品質 14
3.1.2 濃度量化 16
3.2 卡爾曼濾波, Hamming 與 Exponential 結果 21
3.2.1 頻譜品質 21
3.2.2 濃度量化 23
第四章 結論 28
4.1 卡爾曼濾波測量共變異數 R 的比較與選擇 29
4.2 卡爾曼濾波與窗函數於不同平均次數下的表現 29
4.3 平均次數對頻譜品質與代謝物量化穩定性的影響 30
參考文獻 32 | zh_TW |
| dc.format.extent | 17705526 bytes | - |
| dc.format.mimetype | application/pdf | - |
| dc.source.uri (資料來源) | http://thesis.lib.nccu.edu.tw/record/#G0112755008 | en_US |
| dc.subject (關鍵詞) | 磁共振頻譜 | zh_TW |
| dc.subject (關鍵詞) | 卡爾曼濾波 | zh_TW |
| dc.subject (關鍵詞) | 窗函數 | zh_TW |
| dc.subject (關鍵詞) | Magnetic Resonance Spectroscopy | en_US |
| dc.subject (關鍵詞) | Kalman Filter | en_US |
| dc.subject (關鍵詞) | Window Functions | en_US |
| dc.title (題名) | 卡爾曼濾波自適應降噪對磁共振頻譜量化在不同訊號平均次數下的影響 | zh_TW |
| dc.title (題名) | The effect of Kalman-filter-driven adaptive noise reduction on MRS quantification across varying numbers of averages | en_US |
| dc.type (資料類型) | thesis | en_US |
| dc.relation.reference (參考文獻) | [1] Jamie Near et al. Preprocessing, analysis and quantification in single-voxel magnetic resonance spectroscopy: experts’consensus recommendations. NMR in Biomedicine, 2021.
[2] Roland Kreis. The trouble with quality filtering based on relative cramér–rao lower bounds. Magnetic Resonance in Medicine, 75(1):15–18, 2016.
[3] Arwa Baeshen, Patrik O. Wyss, Anke Henning, Ruth L. O’Gorman, Marco Piccirelli, Spyridon Kollias, and Lars Michels. Test–retest reliability of the brain metabolites gaba and glx with jpress, press, and mega-press mrs sequences in vivo at 3t. Journal of Magnetic Resonance Imaging, 51(4):1181–1191, 2020.
[4] Rudolph Emil Kalman. A new approach to linear filtering and prediction problems. Transactions of the ASME–Journal of Basic Engineering, 82(Series D):35–45, 1960.
[5] Samuel S. Blackman and Robert F. Popoli. Design and Analysis of Modern Tracking Systems. Artech House, 1999.
[6] Honghui Qi and John B. Moore. Direct kalman filtering approach for gps/ins integration. IEEE Transactions on Aerospace and Electronic Systems, 38(2):687–693, 2002.
[7] Özge Sezgin Alp, Levent Özbek, and Bilge Canbaloglu. An analysis of stock market prices by using extended kalman filter: The us and china cases. Investment Analysts Journal, 52(3):203–222, 2023.
[8] Omid Sayadi and Mohammad Bagher Shamsollahi. Ecg denoising and compression using a modified extended kalman filter structure. IEEE Transactions on Biomedical Engineering, 55(9):2240–2248, 2008.
[9] Robert Grover Brown and Patrick Y. C. Hwang. Introduction to Random Signals and Applied Kalman Filtering. Wiley, 4 edition, 2012.
[10] Dan Simon. Optimal State Estimation: Kalman, H∞, and Nonlinear Approaches. Wiley, 2006. | zh_TW |
