超音波成像參數對非線性參數估測之影響

Abstract

傳統基頻超音波影像為偵測反射聲波的大小來進行成像，而解析度較高、穿透深度較深的諧波成像，則是藉由超音波在組織間傳遞過程中，因組織非線性特性導致聲波扭曲，經過濾波所得之諧波進行成像。有些人體器官的病變會使特定參數產生變化，如肝臟病變，且除了非線性參數外的其餘聲波參數改變不大。在基頻與諧波成像中，會因這些組織參數不同而帶有不同紋理，但由於不一定有明顯的反射介面，傳統影像無法清楚顯示這些參數不同之區域，只從紋理變化判斷是否產生病變有一定程度的困難。但若能針對非線性參數進行成像，則較能從影像上判斷病變的區域。 本研究模擬聲波在人體組織中行進，並以模擬得到之訊號進行非線性參數之估測。模擬方法為使用pseudo-spectrum對聲波方程式求解，套用PML的邊界條件，並使用二階衰減模型。透過將模擬聲波濾波後所得之基頻與二次諧波訊號，進行衰減係數估測，再用推導出之關係式進行非線性參數之估測並做成像。 影響此種估測方法的聲波參數有頻率、頻寬、估測深度、非線性參數異常幅度與區域大小等。頻率越大則對比度越大，但適用深度越淺。頻寬越大解析度也越大，但考慮亂數雜訊干擾時，雜訊會大幅提高，會使SNR降低。最後，非線性參數異常幅度與區域大小則與解析度有關。 此種估測方法由於估測方法需要對二次諧波能量進行微分，使得雜訊干擾大幅增加，因此最大缺點為對於輸入SNR要求非常高。若將發射訊號改為較低的頻寬，犧牲些許解析度，可以增加SNR。但由於低頻寬發射訊號所得到之解析度較低，因此或許可使用合頻方式增加其解析度，並使SNR提升。

The conventional fundamental ultrasound imaging is by detecting the magnitude of reflected sound, and the harmonic imaging, which has higher resolution and deeper penetration depth, is by detecting the magnitude of reflected sound distortion resulted from tissue nonlinearity and obtained from filtering. However, some organ diseases change specific ultrasound parameters, such as liver disease, and in addition these ultrasound parameters changes little except for nonlinearity. Different textures in ultrasound imaging are cause by different ultrasound parameter in both fundamental and harmonic ultrasound imaging. Though area with different ultrasound parameter do not ensure that there is an interface, it is difficult to detect disease by the fundamental imaging and the harmonic imaging. Nevertheless, it is might be useful to image nonlinearity to detect disease area. In this study, we simulate ultrasound propagating in the tissue and estimate ultrasound nonlinearity by simulation signals. The simulation method is solving acoustics equation by the pseudo-spectrum, the perfectly matched layer boundary condition, and the second relaxation model. To estimate nonlinearity and image, we simulate first, filter the resulted signal, estimate the attenuation, and finally compute by the derived equation. This estimation affect by ultrasound parameter such as frequency, bandwidth, depth, disease area, and the magnitude of changes. The estimating nonlinearity has higher contrast when emit higher frequency, but only for the more shallow depths. The estimating nonlinearity has greater resolution when emit greater bandwidth signals, which increase noise and decrease SNR when considering the random effects. Finally the resolution of estimating nonlinearity is relate to the size and magnitude of disease area. Such estimating method requires differential operation, which is increase the noise effect. Therefore the biggest drawback is the requirement of very high input SNR. It might increase SNR by emit low bandwidth signals and decrease a little resolution. However, due to the low bandwidth of the emit signal obtained with lower resolution, it is possible to increase resolution and SNR by synthetic spectrum.