Patient-Specific Studies of Pelvic Floor Biomechanics Using Imaging



Fig. 18.1
The pubovisceral muscle was outlined in MR images. A 3D geometric model was reconstructed from these contours which was then used in finite element modeling. (From Saleme et al. [7], with permission)



Melchert et al. [9] utilized finite element analysis to study mechanical birth trauma in 1995. Data input for model geometry of the maternal pelvis and fetal head was obtained from MRI. DeLancey and Ashton-Miller in 2004 applied a geometric model to study the levator ani (LA) muscle stretch induced by simulated vaginal birth [10]. They used MRI of a 34-year-old nulliparous woman as their data for model geometry. The data for LA muscle fiber directions, LA muscle subdivisions, and connective tissue orientations were derived from literature. A simplified geometry using discrete muscle bands was used for this study.

Li et al. in 2008 used MRI data from two women (an athlete and a non-athlete) to examine whether athletes are more likely to have a prolonged second stage of labor than non-athletes (Fig. 18.2) [11]. Constitutive relations and mechanical properties of the tissues were assumed and they compared the outputs for the two geometries as related to the pelvic floor during childbirth. In 2008, Hoyte et al. studied the extent and distribution of levator ani stretch during a simulated vaginal delivery in a 21-year-old nulliparous woman [12]. A pelvic MRI extending from the ischial tuberosity inferiorly up to the upper aspect of acetabulum was used for geometric input data. A sphere of 9 cm diameter was passed through the pelvis, along the path of the vagina to simulate passage of a term fetal head during vaginal childbirth.

A306046_2_En_18_Fig2_HTML.gif


Fig. 18.2
(a) The pelvic floor model consisting of 13 components were reconstructed from MR images of a non-athlete woman. (b) The levator ani muscle was represented by a tri-cubic hermite mesh. (From Li et al. [11], with permission)

In summary, the above studies and others have demonstrated the potential of using MRI to obtain geometric data of the pelvic floor from human subjects. Some of the known limitation of MRI is its relatively high expense and limited availability, which may restrain longitudinal studies, especially those during pregnancy to track maternal changes in the pelvic floor. Ultrasound imaging (US), on the other hand, is safe, cheap, and relatively easy to perform in comparison with other available imaging modalities. Recently, Shobeiri’s group developed approaches to acquire 3D ultrasound image data and build 3D models of the pelvic floor structures. High resolution axial images of the endovaginal ultrasonography (EVUS) volume of 0.2 mm slice thickness were collected. Boundaries of the transverse perinei, puboanalis, puborectalis, iliococcygeus muscle, and the anal sphincter complex, rectum, rectal mucosa, vagina, urethra, inferior pubic rami, and pubic symphysis were manually segmented in the medical imaging viewer OsiriX (Pixmeo). The geometry of each segmented structure was stored in a stereolithography file which was then smoothed and colored in 3D-Coat (Pilgway). Finally, the 3D models were rendered in Meshlab (Visual Computing Lab—ISTI—CNR) (Fig. 18.3). 3D US can potentially be used to create subject specific subject specific 3D geometric model of the pelvic floor with ability to visualize detailed anatomical structures. In order to achieve such goal, advanced image analysis methods are expected to be developed in order to build 3D models accurately and computationally efficiently.

A306046_2_En_18_Fig3_HTML.jpg


Fig. 18.3
3D model reconstruction from ultrasound



Constitutive Material Properties of the Pelvic Floor


Once the 3D anatomy is reconstructed, the material properties need to be assigned to the different tissue elements to simulate the effect of biomechanical forces on these tissues. Finite element modeling (FEM) has been the main mechanical simulation approach on studying the pelvic floor biomechanics. In the following, constitutive models of the pelvic floor structures that have been used in finite element modeling of the pelvic floor will be described. Simulation of FEM is beyond the scope of the book therefore will not be reviewed.

Mechanical properties such as the stiffness and viscosity provide information about the physical behavior of a tissue in response to external forces. To quantitatively examine the mechanical properties of certain material, constitutive models are generally developed that describe the relationship between the applied stress and the resulting strain when the material is under deformation. Neo-Hookean model and Mooney-Rivlin model are the two commonly used hyperelastic continuum models on the pelvic floor tissues.

Martin et al. applied the Neo-Hookean constitutive model in simulating the nonlinear stress-strain response of the connections between pelvic muscles and the coccyx, arcus tendineus, obturator fascia, and the obturator internus [8]. The pelvic floor muscle was assumed to be isotropic. The constitutive model was based on a skeletal muscle model developed previously [8, 13]: σ = σ incomp + σ matrix + σ fiber. The three terms model the Cauchy stress tensor equations of the material incompressibility, hyperelastic matrix, and muscle fibers. The three-element Hill muscle model was used to compute the muscle fiber biomechanics. A first order ordinary differential equation, which models the muscle activation dynamics and has been applied widely in skeletal muscle simulation, was included. A Neo-Hookean constitutive model assuming isotropic, homogeneous, and incompressible material was also used in simulating the mechanical interaction between the levator ani muscle and the fetal head [11, 14]. The strain energy function was defined as ψ = c 1(tr(C) − 3). The material constant of LA was set to 10 kPa and that of the fetal head was set to be 100 KPa. tr(C) was the first principal invariant of C, the right Cauchy-Green deformation tensor. A slightly different Neo-Hookean model capable of modeling the anisotropic hyperelastic response was used in the finite element simulation of the second stage of labor,




$$ W= C\left({I}_1-1\right)+\Big\{\begin{array}{c}\frac{k_1}{2{k}_2}\left[{e}^{k_2{\left({\lambda}^2-1\right)}^2}\right],\lambda \ge 1\\ {}0,\lambda <1.\end{array} $$

The second term on the fiber reinforcement was added to model the anisotropy of the LA muscle and the perineal body [15]. Compared to the higher order Mooney-Rivlin model, the simplicity of the Neo-Hookean model makes the implementation easier and more available [15].

Other researchers used Mooney–Rivlin constitutive model to model the nonlinear hyperelastic material properties of the LA muscle. The strain energy function used in simulation was defined as W = c 10(I 1 − 3) − c 20(I 2 − 3), where I 1 and I 2 are the principal strain invariants of the right Cauchy-Green tensor. The Mooney-Rivlin parameters c 10 and c 20 were set to 4.5 kPa and 2 kPa, respectively, in the subject-specific model developed by Noakes et al. [16], stiffer than the parameter values of 2.5 kPa and 0.625 kPa in an earlier study by Lee et al. [17].

While previous approaches typically used generic values of constitutive material properties, emerging imaging modalities provide the opportunity to directly measure mechanical properties of tissue. Elastography was developed in the 1990s to map tissue stiffness to quantify the palpation performed by clinicians. From a physics point of view, elastography aims to quantitatively image the Young’s modulus, the physical parameter corresponding to the stiffness. Using elastography has two important advantages:



Tags:
Jul 11, 2017 | Posted by in UROLOGY | Comments Off on Patient-Specific Studies of Pelvic Floor Biomechanics Using Imaging

Full access? Get Clinical Tree
Get Clinical Tree app for offline access