In vivo Nakagami‐m parametric imaging of microbubble‐enhanced ultrasound regulated by RF and VF processing techniques

Abstract

Purpose Application of the Nakagami statistical model and associated m parameter has the potential to suppress artifacts from adjustable system parameters and operator selections typical in echo amplitude‐coded microbubble‐enhanced ultrasound (MEUS). However, the feasibility of applying m estimation and determination of the associated Nakagami distribution features for in vivo MEUS remain to be investigated. Sensitivity and discriminability of m‐coded MEUS are often limited since raw envelopes are regulated by complex radiofrequency (RF) and video‐frequency (VF) processing. This study aims to develop an improved imaging approach for the m parameter estimation which can overcome the above limitations in in vivo condition. Method The regulation effects of RF processing of pulse‐inversion (PI) harmonic detection techniques and VF processing of logarithmic compression in Nakagami distributions were investigated in MEUS. A window‐modulated compounding moment estimator was developed to estimate the MEUS m values. The sensitivity and discriminability of m‐coded MEUS were quantified with contrast‐to‐tissue ratio (CTR), contrast‐to‐noise ratio (CNR), and axial and lateral resolutions, which were validated through in vivo perfusion experiments on rabbit kidneys. Results Regulated by RF and VF processing, the distributions of MEUS obeyed the Nakagami statistical model. The Nakagami‐fitted correlation coefficient was 0.996 ± 0.003 (P < 0.05 in the t test and P < 0.001 in the Kolmogorov−Smirnov test). Among each of the m‐coded MEUS methods, the logarithmic m‐coded PI‐MEUS scheme effectively characterized the peripheral rim perfusion features and details within the renal cortex. The CTR and CNR in this region reached 7.9 ± 1.5 dB and 34.4 ± 1.7 dB, respectively, which were higher than those of standard amplitude‐coded MEUS; and the axial and lateral resolutions were 1.02 ± 0.02 and 0.91 ± 0.02 mm, respectively, which were slightly longer than those of amplitude‐coded MEUS. Conclusions The Nakagami statistical model could characterize MEUS even when the envelope distributions were regulated by RF and VF processing. The logarithmic m‐coded PI‐MEUS scheme significantly improved the sensitivity, discriminability, and robustness of m estimation in MEUS. The scheme provides an option to remove artifacts in echo amplitude‐coded MEUS and to distinctly characterize the inherent microvasculature enhanced by microbubbles, with potential to improve and expand the role of MEUS in diagnostic ultrasound.

Feasibility investigation of logarithmic Nakagami parametric imaging in recovering underestimated perfusion metrics of DCEUS in the uneven acoustic field

“…Figure 1 shows the flow chart of logarithmic m ‐coding DCEUS scheme in the focused acoustic field, which is based on our previous m ‐coding contrast‐enhanced ultrasound studies 18,21 . Considering the different advantages of ME and WMCME methods, 18 we also investigated the performances of the assumed scheme using the ME and WMCME methods to overcome the limitations of DCEUS in the focused acoustic field.…”

“…Under a focal depth of 2 cm, in vivo perfusion experiments ( n = 4 × 66 – 72 frames before pulse‐inversion summation) in the kidneys of four healthy rabbits were performed on DC‐6 platform (Mindray Inc., Shenzhen, China) equipped with a convex array scanner (emitted pulse frequency of 3.0 MHz). Part of in vivo data was from our previous study 18 . After the rabbit was anesthetized, 0.2 ml bolus of SonoVue dilution with a concentration of 2 × 10 8 MBs/ml was intravenously injected.…”

“…The shape parameter m in the Nakagami model can characterize wide‐ranging scattering distributions of Rayleigh, pre‐, and post‐Rayleigh 16 . m parameter was formerly utilized to analyze radar signals and was first used to gather statistics of ultrasonic linear backscattered‐echoes from tissues by Shankar et al., 17 thereby predicting the in vitro vascular flow marked by nonlinear backscattered‐echoes from MBs by Tsui et al., 7 and characterize in vivo m ‐coding contrast‐enhanced ultrasound by Wang et al 18 …”

“…Moreover, the tissue speckles inevitably affect the m estimation of MB backscattered‐echoes in in vivo experiments because their Nakagami distributions overlap in large proportions 18 . Accordingly, we have previously used PIHI to suppress the tissue artifacts and investigate the regulation mechanism and impact of pulse‐inversion, harmonic filtering, and logarithmic transformation during PIHI implementation on the raw envelopes of MB backscattered‐echoes and their m estimation 18 . Compared with the moment‐estimator (ME) method, the window‐modulated compounding ME (WMCME) strategy could improve the estimation accuracy and ensure the estimation stability and robustness of m , respectively 18 .…”

“…Accordingly, we have previously used PIHI to suppress the tissue artifacts and investigate the regulation mechanism and impact of pulse‐inversion, harmonic filtering, and logarithmic transformation during PIHI implementation on the raw envelopes of MB backscattered‐echoes and their m estimation 18 . Compared with the moment‐estimator (ME) method, the window‐modulated compounding ME (WMCME) strategy could improve the estimation accuracy and ensure the estimation stability and robustness of m , respectively 18 . Nevertheless, the ME method has the best computational efficiency 18 …”