利用頻率偏移估計超音波衰減係數之系統考量

A Consideration on System Factors for Spectral Shift Method in Estimating Ultrasonic Attenuation Coefficients

張辰瑋(Chen-Wei Chang)

指導教授 ： 張建成

共同指導教授 ： 崔博翔

國立臺灣大學/工學院/應用力學研究所/碩士(2016年)

https://doi.org/10.6342/NTU201601791

摘要

超音波系統因為具有方便性、即時影像、非侵入性且成本相較於其他如核磁共振造影(MRI)或電腦斷層掃瞄(CT)低廉等優點，成為第一線臨床診斷工具。利用逆散射訊號分析組織參數並量化評估的方式逐漸成為主流，對於臨床診斷以及超音波B-Mode影像分析，估測軟組織的衰減係數為其中一重要課題，基於上述原因，本研究透過短時距傅立葉轉換(STFT)分析訊號在不同深度的頻率成分，搭配頻率偏移法估計衰減係數，並觀察在臨床上隨著脂肪肝病情嚴重程度衰減係數的變化趨勢，更進一步探討多項系統因子對於此方法的影響。
本研究實驗分成兩大部分：第一部分為石墨仿體實驗，第二部分是將仿體實驗所得的最佳系統考量嘗試應用於脂肪肝患者臨床實驗。在前者的實驗裡，會先以標準測定實驗測得標準的仿體衰減係數，接著探討探頭中心頻率、繞射校正、時間增益補償(TGC)等三項系統因子對頻率偏移法的影響。由實驗結果得知，隨脂肪濃度的提升會使衰減係數增加以及頻譜的能量振幅更加劇烈地遞減。而中心頻率越大的探頭其頻譜中心頻率對深度的斜率將大幅增加，驗證高頻率高衰減的特性。為了精準估測衰減係數，需補償因超音波探頭傳遞聲波之聚焦特性造成的繞射效應，結果無論弧形探頭或線性探頭，校正後衰減係數的準確度均大幅提升。臨床醫師為清晰觀察B-Mode影像內較深區域的組織樣貌，會開啟時間增益補償，此部分著重探討Terason t3000超音波掃描系統內的時間增益補償對於頻率偏移法估計衰減係數所造成的差異，發現在大部分的情況下，未開啟TGC (N-TGC)所估測的衰減係數誤差比開啟TGC時小。在臨床實驗裡，TGC所估得的衰減係數較N-TGC高，且在ROC曲線分析的AUC值比N-TGC低，並初步得到臨床數據估測衰減係數與脂肪肝診斷的關聯性。
綜合前述，使用頻率偏移法估測衰減係數選擇低頻探頭、加入繞射校正、N-TGC為較好的系統因子選擇，而此結果在臨床應用上區分輕度、中重度脂肪肝具有優勢。

關鍵字

衰減係數 ； 頻率偏移法 ； 短時距傅立葉轉換 ； 繞射效應 ； 時間增益補償

並列摘要

Ultrasonic system, the widely used in clinical diagnosis, is providing the relatively convenient, instantaneity, non-invasion and low cost than MRI, CT, etc. Besides, the assessment of backscattering signal processing on tissue parameters by quantitative analysis have become increasingly popular recently. Based on the above reasons, the spectral shift method is proposed by using the short-time Fourier transform to estimate the ultrasonic attenuation coefficients which is an important issue for clinical diagnosis. Investigating the influence of system factors by proposed method is the main concept of this study, e.g. evaluate the patient’s fatty liver condition by observing the variation tendency of the ultrasonic attenuation coefficients. There are two parts in this study: 1. the graphite phantom experiment and 2. the clinical experiment based on the results of phantom experiment. In the former experiment, the measured standard attenuation coefficients were compared with the results of the spectral shift method in three system factors adjustments, the center frequency of ultrasonic transducer, diffraction compensation, and Time Gain Compensation (TGC). Measurement results showed that the attenuation coefficient increased with high fatty concentration. The characteristics of high frequency with high attenuation was verified because the failing slope of center frequency were increased with larger center frequency of transducer. To precisely improve the estimated attenuation coefficients, the diffraction calibration of the beam focal effects caused by the transducer in acoustic transfer is involved. Experimental results showed that no matter the convex transducer or the linear transducer, the accuracy of modified attenuation coefficients have significantly improved. To have better image visualization of B-Mode, the TGC will be opened to identify the deeper region in the clinical. It was frequently investigated that the attenuation coefficients of the N-TGC estimated have better accuracy than the TGC. Clinical experiment on fatty liver is the second part of this study. It was discovered that the attenuation coefficients of TGC estimated is higher than N-TGC; however, in the ROC curve analysis, the results of TGC estimated is lower than N-TGC. To sum up, lower center frequency of ultrasonic transducer, diffraction calibration, and N-TGC are the three beset system factors to identify the mild and moderate fatty liver in clinical applications.

並列關鍵字

Attenuation coefficient ； Spectral shift method ； Short-time Fourier transform ； Diffraction effects ； Time gain compensation

參考文獻

Baldeweck, T. L., Herment, A., & Berger, G. (1995). Application of autoregressive spectral analysis for ultrasound attenuation estimation: interest in highly attenuating medium. IEEE transactions on ultrasonics, ferroelectrics, and frequency control, 42(1), pp. 99-110.

Bridal, S. L., Fornès, P., Bruneval, P., & Berger, G. (1997). Parametric (integrated backscatter and attenuation) images constructed using backscattered radio frequency signals (25–56 MHz) from human aortae in vitro. Ultrasound in medicine & biology, 23(2), pp. 215-229.

Clifford, L., Fitzgerald, P., & James, D. (1993). Non-Rayleigh first-order statistics of ultrasonic backscatter from normal myocardium. Ultrasound in medicine & biology, 19(6), pp. 487-495.

d'Astous, F. T., & Foster, F. S. (1986). Frequency dependence of ultrasound attenuation and backscatter in breast tissue. Ultrasound in medicine & biology, 12(10), pp. 795-808.

Fink, M. A., & Cardoso, J. F. (1984). Diffraction effects in pulse-echo measurement. IEEE transactions on sonics and ultrasonics, 31(4), pp. 313-329.