Quantitative ultrasound assessment of the influence of roughness and healing time on osseointegration phenomena

Abstract

The evolution of bone tissue quantity and quality in contact with the surface of orthopedic and dental implants is a strong determinant of the surgical outcome but remains difficult to be assessed quantitatively. The aim of this study was to investigate the performance of a quantitative ultrasound (QUS) method to measure bone-implant interface (BII) properties. A dedicated animal model considering coin-shaped titanium implants with two levels of surface roughness (smooth, Sa = 0.49 µm and rough, Sa = 3.5 µm) allowed to work with a reproducible geometry and a planar interface. The implants were inserted in rabbit femurs and tibiae for 7 or 13 weeks. The ultrasonic response of the BII was measured ex vivo, leading to the determination of the 2-D spatial variations of bone in contact with the implant surface. Histological analysis was carried out to determine the bone-implant contact (BIC) ratio. The amplitude of the echo was significantly higher after 7 weeks of healing time compared to 13 weeks, for both smooth (p < 0.01) and rough (p < 0.05) implants. A negative correlation (R = − 0.63) was obtained between the ultrasonic response and the BIC. This QUS technique is more sensitive to changes of BII morphology compared to histological analyses.

Ultrasonic assessment of osseointegration phenomena at the bone-implant interface using convolutional neural network

“…Based on this transformation, we assumed that the soft tissue thickness decreases when osseointegration phenomena progress, which is a similar model than that used in (Raffa et al, 2019;Raffa et al, 2020) to model osseointegration phenomena in the static regime. Moreover, we have shown experimentally that the ultrasound reflection coefficient of the BII is very sensitive to the properties of bone tissue located around the BII (Mathieu et al, 2012;Hériveaux et al, 2019;Fraulob et al, 2020).…”

“…A large-scale annotated dataset with diverse conditions should be considered to employ deep learning methods. However, accumulating a large-scale dataset in biomechanical engineering may be difficult because i) of the difficulty to obtain an important number of samples due to ethical requirements and ii) it is impossible to control all parameters affecting the ultrasound response of the BII since they vary simultaneously (Fraulob et al, 2020) as a function of healing time. Here, we present a method based on the deep convolutional neural network to assess osseointegration phenomena.…”

“…Experimental approaches have already shown that the reflection coefficient of the BII is sensitive to healing time (Mathieu et al, 2012), to the implant surface roughness (Fraulob et al, 2020) as well as to the mechanical loading applied to the BII (Hériveaux et al, 2019b). Based on the aforementioned experimental results, an experimental set-up was developed to assess dental implant primary and secondary stability (Vayron et al, 2014a;Vayron et al, 2014c;Vayron et al, 2018a;Vayron et al, 2018b).…”

“…The two aforementioned studies justify the use of a 2-D sinusoidal description of the implant surface roughness. Second, only normal incidence of the ultrasonic wave is reported, because it corresponds to the experimental situation of interest (Mathieu et al, 2012;Fraulob et al, 2020) and future studies should also consider the normal incidence. Third, this study only considered a center frequency of 10 MHz for the incident wave because it corresponds to the frequency used in the QUS device developed by our group (Vayron et al, 2018a).…”

“…Approaches using quantitative ultrasound (QUS) seem promising to retrieve information on the BII because it is non-invasive, easy to use and nonionizing. Preliminary studies using coin-shaped implant models have evidenced a significant decrease of the reflection coefficient of the BII as a function of healing time (Mathieu et al, 2012;Fraulob et al, 2020), which can be explained by increases of the bone-implant contact (BIC) ratio, of the bone Young's modulus (Vayron et al, 2012) and of bone mass density (Mathieu et al, 2011;Vayron et al, 2014b) as a function of healing time. The sensitivity of the reflection coefficient of the BII to compressive stresses was also investigated (Hériveaux et al, 2019b).…”