美国陆军训练与作战发展司令部的人体数字孪生研究：过去、现在与未来

Gary P. ZientaraPh.D.U.S. Army Research Institute of Environmental MedicineU.S.A. gary.p.zientara.civ@health.mil ABSTRACT The Human DigitalTwin has been under development atthe U.S. Army Research Institute of EnvironmentalMedicine(USARIEM) for overa decade. Software has enabled the creation ofindividualizedFinite Elementmodelsof humanswith complete internal anatomy.This enables biomechanics duringanimation,theapplication ofa physiology enginefor medical monitoring,anda method to compute metabolicenergyconsumption. 1.0INTRODUCTION &EARLY DEVELOPMENT The Human Digital Twin (HDT) concept pairs a human individual,thePhysical Twin–with all of his/herphysical attributes and medical system uniqueness–to a virtual/digital/computer modelof that individual (oras much of the individual as one wishes to consider). This is the Human Digital Twin. This construct’s purpose is to give scientists, physicians and engineers the ability to digitally simulate thephysical systems (i.e. biomechanics, animation) and or medical systems (e.g. physiology engine, metabolicengine)of the Digital Twin, while also being able to include aspects of the Physical Twin: real-time systemdata updates, interactive variations in those systems (e.g. interventions), and enablingthe generation offorward-looking predictions regarding the Physical Twin. The development of the HDTwith completeanatomy has continued since 2010 atUSARIEM, focused onthe independently controlled individualized avatar, a standalone virtual human.Our objective was to createthe Digital Twins of research subjects, which we could exploit for physiological experimentation. The inception of the project evolved from the2003 DARPA Virtual Soldier Project and the computer andtechnicaladvancements of the late 2000’s that enabled the key components necessary to createindividualized avatars:the segmented complete anatomy of a model male and female, software forconversion of volumetric anatomical representations to Finite Element(FE)representations, laser scanningdevices toacquire the complete ‘water-tight’ surface mesh of an individual, and software to morph standardanatomy into the body surface mesh. The original motivationofourHDT researchwas the creation of virtualhuman subjectsthat could representparticipants in the many human subject experimental protocols performed bythe USARIEMMilitaryPerformance Division, Military Nutrition Division and Thermal and Mountain Division.MyUSARIEMphysiologist colleague, DrReed W. Hoyt(ret.),recognized and voiced the need for detailed individualizedavatars with complete anatomy to advance physiology research intothe 21stcentury,studying circumstancesoutside the limits of ethical safe human subject experimentation.He especiallyemphasizedthe need forphysiology simulations with individual soldier avatarsthat spanned the entire range of male and female bodyshapes, size, and heights which varied greatly in the military. Virtual participantsextend the range ofexperimentalactivitiesto study circumstances thatexceedcost limits,logistics limits andsafety limitsassociated withhuman subject experimentation. Virtual human studiesoffer endless experimental repetition, the elimination of the logistics of working with human subjects, andatremendous reduction in experimental expenses. Our research corresponded withthe completion ofthe anthropometric survey (ANSUR II) (1) of militarypersonnelby the U.S. Army Natick Soldier Research,Development and Engineering Center(nowDEVCOM). The ANSUR II study acquiredthe wholebody surface meshformsfrom 3D laser scansofmaleand femaleU.S. Army personnel. Thewholebody surface meshcomprised one key component of our HDTs.Our scheme was to morph theFErepresentation of ‘standard’ male and femaleanatomy so itfit within the surface mesh of an individual. Computing standard figures is laborious, starting with a volumetric representation of a male and female,likely from cadavers obtained by MRI, x-ray CT, and/or thin sections.One source wasthe U.S.NationalLibrary of Medicine formed the Visible Human (VH) project database (3), as was employed in the earlierDARPA Virtual Soldier Project.Then, HDT creation requires 3D imagesegmentation, that is, thedifferentiation of the volumetric representations of the standard figures into specific labeled tissuetypes,bones, vesselsand nerves. This is accomplished byprogressivelysegmenting sections through the anatomyof the volumetric datawith3D image processing softwaresuch as3D Slicer (4). The segmented VH male FEstructure was kindly provided to us in2011 by the French Institute for Research in Computer Science andAutomation in Sophia Antipolis. In 2012 we segmented the VH female volumetric data and created versionswith alteredlower amounts of subcutaneous fat.We also obtained the male and female segmentedvolumetric data from the ITIS Foundation Virtual Family (5). The segmented volumetric data was convertedto labeled FE structures using mesh generation software from the Computational Geometry