Ultrasonic backscatter signals from cancellous bone are sensitive to the microstructure of trabecular bone,and thus enable the feasibility to extract microstructural information of trabecular bone.The mean trabecular bone spacing(MTBS)is an important parameter for characterizing bone microstructure.This paper proposes an MTBS estimation method based on the combination of Hilbert transform and fundamental frequency estimation(CHF). The CHF was verified with ultrasonic backscatter signals from simulations and in vitro measurements at a central frequency of 5MHz.The CHF method was compared with the simplified inverse filter tracking(SIFT)method,Simons’ Quadratic Transformation(QT)method,Singular Spectrum Analysis(SSA)method,and Spectral Autocorrelation(SAC)method.Monte-Carlo simulations were performed by varying the MTBS,signal-to-noise ratio(SNR),standard deviation of regular spacing(SDRS),amplitude ratio of diffuse scattering to regular scattering(Ad)and frequency dependent attenuation(nBUA).The simulation results showed that the CHF method had a better performance in MTBS estimation under almost all the examination conditions except for SNR.The estimation percentage correct(EPC)was greater than 90% when the MTBS was in the range of 0.4to 1.4mm.In the in vitro measurements,the estimated EPC by the CHF method was91.25±7.81%(mean±standard deviation).A significant correlation was observed for the CHF-estimated MTBS and micro-computed tomography(μ-CT)-measured values(R~2=0.75,p<0.01).These results demonstrate that the CHF method is anti-interference for MTBS estimation and can be used to estimate trabecular bone spacing. Ultrasonic backscatter signals from cancellous bone are sensitive to the microstructure of trabecular bone, and thus enable the feasibility to extract microstructural information of trabecular bone. The mean trabecular bone spacing (MTBS) is an important parameter for characterizing bone microstructure. This paper proposes an MTBS estimation method based on the combination of Hilbert transform and fundamental frequency estimation (CHF). The CHF was verified with ultrasonic backscatter signals from simulations and in vitro measurements at a central frequency of 5MHz. The CHF method was compared with the simplified inverse filter tracking ( SIFT) method, Simons’ Quadratic Transformation (QT) method, Singular Spectrum Analysis (SSA) method, and Spectral Autocorrelation (SAC) method. Monte- Carlo simulations were performed by varying the MTBS, signal-to-noise ratio standard deviation of regular spacing (SDRS), amplitude ratio of diffuse scattering to regular scattering (Ad) and frequency dependent attenuation ( The simulation results showed that the CHF method had a better performance in MTBS estimation under almost all the examination conditions except for SNR. The estimation percentage correct (EPC) was greater than 90% when the MTBS was in the range of 0.4 to 1.4mm.In the in vitro measurements, the estimated EPC by the CHF method was91.25 ± 7.81% (mean ± standard deviation). A significant correlation was observed for the CHF-estimated MTBS and micro-computed tomography (μ-CT) -measured values ​​(R ~ 2 = 0.75, p <0.01). These results demonstrate that the the CHF method is anti-interference for MTBS estimation and can be used to estimate trabecular bone spacing.