Muscle Modeler  
Objective  quantification of architectural parameters (e.g., fascicle length, pennation angle, physiological crosssectional area and volume) and 3D reconstruction of surface and volume geometry  
Methods  dissection and digitization of cadaveric specimens (MicroScribe G2X) and ultrasonography  
Outputs  architectural analysis and geometric mesh (triangular and tetrahedral)  
Applications  cadaveric studies : extensor/flexor forearm muscles, supraspinatus/infraspinatus, pectoralis major, vastusmedialis, piriformis, sacral nerves, ACL/PCL ligaments, humerus tendon rupture. invivo studies: supraspinatus/infraspinatus (current)  
Development  C/C++, MFC, OpenGL, BLAS, LAPACK, TAUCS, MS Windows platforms  
Keywords  dissection, digitization, cubic spline, voronoitessellation, levelset method, tetrahedralization, Laplacian surface, linear/nonlinear least squares, finite element method  
Screenshot examples  
Mainframe and examples of dialogbased control (editing digitized points, anatomical frame estimation and fascicular investigation)  
Main features include basic geometric manipulation, architectural analysis, surface/volume extraction, FEMbased muscle deformation, geometric deformation, virtual ultrasound imaging, digitization error reduction, geometric crossmatching of architectures, 3D registration and etc.  
Reconstruction of surface and volume geometry  
Digitized fascicles are wrapped around by the closed surface that is tracked using the levelset method. Associated implicit functions are determined based on the estimated anatomical crosssectional area of each fascicle (i.e., elliptical and piecewise cylinders). Those functions are carefully blended by using the locallyweighted interpolation, which controls smoothness and prevents undesired geometric defects (e.g., hole, gaps, concavity). Brachioradialis is shown in figures.  


Estimation of line of action  
Linear regression is used to determine line of action (intraand extra muscular tendon direction) of pennate muscles. (a. supraspinatus and b. vastusmedialis)  
Assessment of intramuscular variation of pennation angle  
It is observed that muscles may have strong pattern of pennation angle distribution with respect to anatomical axis. Proximodistal, lateromedial and superficial to deep axes are accounted for presented studies.  
Estimation of physiological crosssectional area  
Voronoitessellation is used to approximate crosssectional area of each fascicle. This method iteratively computes all fascicles and their points. Our study uses 3060K resampled points, depending on architectural complexity or size.  
Integration of ultrasonographic assessment for invivo architectural modeling  
Our study stresses that anatomical details are needed to enhance reliability and reproducibility of ultrasonography. Furthermore, it is demonstrated how effectively ultrasonographic noninvasive assessment can be complemented by cadaveric invasive measurement (paper under review). This is currently being implemented into practical application. Details will be given soon.  
Ultrasonographic assessment vs virtual (simulated) ultrasound  


3D Registration for muscle architecture (fitting cadaveric model to invivo ultrasonographic assessment)  

