研發超音波逆散射統計參數以應用於組織特性識別的相關考量

Some considerations on the development of statistical parameters from ultrasonic backscatter for tissue characterization

崔博翔(Po-Hsiang Tsui)

指導教授 ： 王士豪

中原大學/工學院/醫學工程研究所/博士(2006年)

https://doi.org/10.6840/cycu200500811

摘要

不同性質的散射子會對於來自生物組織的超音波逆散射包封訊號造成不同的統計分布特性。這些具有不同統計分布特性的包封訊號可透過一個能涵蓋所有超音波散射情形的Nakagami統計模型來描述其機率密度函數，其中Nakagami參數又對於區分生物組織散射子濃度的變化具有良好的能力。為了擁有Nakagami參數在定量生物組織特性時更佳的執行效能，我們探討了換能器效應、雜訊效應、以及對數壓縮效應在Nakagami參數估計時所可能造成的影響，同時也評估以Nakagami模型為基礎的參數運用在血液濃度偵測的可行性。
首先進行假體實驗與二維電腦模擬以探討換能器效應對於Nakagami參數估計的影響。使用中心頻率為5 MHz的非聚焦式與聚焦式換能器擷取散射子濃度由2至32 scatterers/mm3的假體超音波逆散射訊號。以相同的電腦模擬方法進一步探討在相同散射子濃度範圍內，雜訊與對數壓縮對於Nakagami參數估計之效應。最後以5 MHz的聚焦式換能器擷取來自血容比範圍由3%至50%的紅血球懸浮液，以此評估使用Nakagami統計模型的參數辨別血容比之可行性。
實驗結果證明，以聚焦式換能器所估計的Nakagami參數相較於使用非聚焦式換能器所估計的參數擁有更高的靈敏度以偵測散射子濃度的變化。Nakagami參數區分不同散射子濃度的靈敏度會隨著逆散射訊號訊雜比的下降而降低。此外，對數壓縮的計算會將大多數散射子濃度所相對應的包封訊號統計移至post-Rayleigh分布，使得透過壓縮的逆散射包封來估計的Nakagami參數具有更佳的靈敏度以區分不同的散射子濃度。另一方面，以未經壓縮的超音波逆散射包封所估計的Nakagami參數並無法區分不同的血容比，原因在於這些不同濃度的包封訊號統計皆為Rayleigh分布。然而，在經過對數壓縮的計算之後，所得到的Nakagami參數可有效的辨別血容比的變化。本研究結論，Nakagami參數在組織特性化中較佳的執行效能可在使用良好聚焦的換能器以及低雜訊干擾的情況下獲得。可用來提昇Nakagami參數靈敏度的對數壓縮技術，亦可幫助超音波逆散射統計參數進一步的應用於未來非侵入式血容比測量技術的發展與推廣。

關鍵字

Nakagami分布 ； 組織特性化 ； 血容比 ； 超音波逆散射

並列摘要

Various properties of scatterers may cause different statistical distributions for the envelopes of ultrasonic backscattered signals received from biological tissues. The Nakagami statistical distribution, which has been demonstrated to be a general model capable of encompassing all conditions of ultrasonic backscattering, was proposed to describe the probability density function (PDF) of ultrasonic envelope signal. In particular, the Nakagami parameter estimated using the backscattered envelopes has a good ability to distinguish the variation of scatterer concentration in tissues. For the better performance of the Nakagami parameter in quantifying the properties of biological tissues, the effects of transducer characteristics, noise, and logarithmic compression on the estimation of the Nakagami parameter were explored. The feasibility for the application of the Nakagami-model-based parameters to the detection of blood concentration was also evaluated in this study.
The measurements of phantoms and two-dimensional computer simulations were carried out to explore the effects of transducer characteristics on the estimation of the Nakagami parameter. The Nakagami parameter as a function of scatterer concentration was calculated using backscattered signals acquired from the scattering medium of different scatterer concentrations ranged from 2 to 32 scatterers/mm3 using both 5 MHz non-focused and focused transducers. The same simulation method was further applied to produce the 5 MHz backscattered echoes in the same range of scatterer concentration to investigate the effects of noise and logarithmic compression on the Nakagami parameter estimation. The ultrasonic signals backscattered from red cell suspensions with hematocrits ranging from 3% to 50% were collected using a 5 MHz focused transducer to further verify the feasibility of the hematocrit differentiation by the parameters of the Nakagami distribution.
The results demonstrated that the Nakagami parameter estimated using a focused transducer tends to be more sensitive than that by a non-focused transducer to detect the variation of the scatterer concentration. Moreover, the sensitivity of the Nakagami parameter to differentiate different scatterer concentrations would decrease gradually corresponding to the decrease of signal-to-noise ratio (SNR) of backscattered signals. In addition, the logarithmic compression would move the statistics of the backscattered envelopes toward post-Rayleigh distributions for most scatterer concentrations, making the Nakagami parameter calculated using compressed backscattered envelopes is more sensitive than that calculated using uncompressed envelopes in differentiating variations in the scatterer concentration. On the other hand, the Nakagami parameter calculated using the uncompressed backscattered envelopes cannot be used to separate various hematocrits due to that the probability distributions of the uncompressed envelopes of different hematocrits all follow Rayleigh statistics. However, different hematocrits can be effectively distinguished using the Nakagami parameter after applying logarithmic compression to the envelope signals. This study concluded that the better performance of the Nakagami parameter in characterizing tissues could be achieved in conditions with both the use of well-focused transducer and the low noise interference. The logarithmic compression that assists in the enhancement of the Nakagami parameter sensitivity has a great potentiality in the future development of a new method to measure the hematocrit based on the Nakagami model.

並列關鍵字

Nakagami ； Hematocrit ； Tissue characterization ； Ultrasonic backscattering

參考文獻

Adler RS. On the relationship between power mode and pressure amplitude decorrelation. Ultrasound Med. Biol. 2001;27:1291-1296.

Ahuja AT, Ying M, Evaluation of cervical lymph node vascularity: a comparison of colour Doppler, power Doppler and 3-D power Doppler sonography. Ultrasound Med. Biol. 2004;30:1557-1564.

Atkinson P, Woodcock JP. Doppler ultrasound and its use in clinical measurement. London: Academic Press, 1982.

Burckhardt CB. Speckle in ultrasound B-mode scans. IEEE Trans. Sonics Ultrason. 1978;SU-25:1-6.

Ballew C, Haas JD. Hematologic evidence of fetal hypoxia among newborn infants at high altitude in Bolivia. Am. J. Obstet. Gynecol. 1986;155:166-169.