|Year : 2016 | Volume
| Issue : 3 | Page : 81-87
Computer-aided medicine and surgery
Team INRIA-UPMC-CNRS REO, Laboratoire Jacques-Louis Lions, UMR 7598, University Pierre et Marie Curie, 75005, Paris, France
|Date of Web Publication||24-Nov-2016|
Dr. Marc Thiriet
Team INRIA-UPMC-CNRS REO, Laboratoire Jacques-Louis Lions, UMR 7598, University Pierre et Marie Curie, 75005, Paris
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Thiriet M. Computer-aided medicine and surgery. Digit Med 2016;2:81-7
Digital medicine, or cybermedicine, includes computer-aided procedures and image-guided therapy. The new generation of medical tools is based on experience in sensor fusion, computer vision, robotics, virtual reality, and image and signal processing. They include, in particular, navigation and positioning tools before and during the medical and surgical procedures. Navigation systems enable the determination of optimal patient-specific location and guide the operator to achieve the desired placement.
Precision of medical and surgical gestures is a fundamental requirement. Operating robots receive images of a target organ, analyze its motions, and assist surgical gesture according to these movements.
Telemedicine is based on systems of electronically communicating data from one site to a distant site with data fusion by superimposing patient-specific data. Telepresence operation procedures have two major components: (1) a remote site with a three-dimensional (3D) camera system and responsive manipulators with sensory input and (2) an operating workstation with a 3D monitor and dexterous handles with force feedback.
Telesurgery consists of controlling a remote slave manipulator robot that operates surgical instruments using a master control console. This console measures displacements of fictitious instruments driven by the surgeon and transmits them to the robot. In addition, navigation systems are used to guide the surgeon gesture using imaging. A remotely controlled robot may then be capable of executing the procedure at the site of the operation, in which, nonetheless, are specialists ready to execute tasks. Teletaction sensors react according to the type of material with which the operator is dealing, and imitation tools at the workstation correspond to actual tools on the robotic arms at the site of the operation.
Surgical planning and design gain from modeling implicated in virtual reality tools and mechanical exploration. These preliminary stages enable to select the most appropriate surgical path as well as the most suitable repair technique.
Regenerative and reparative medicine is aimed at restoring the form and function of damaged tissues of the human body or replacing them. It integrates knowledge acquired from biological, bioengineering, and clinical research. It relies on bioreactor design that must incorporate various sources of cell signaling, that is, involved chemical, physical, and mechanical agents. Cells are conditioned in vitro and then administered to patients.
Nanotechnology is used in genome analysis, cellular and tissular engineering, as well as in prevention, early detection, imaging, monitoring, and therapeutics. Nanotechnology is also aimed at improving drug delivery and medical devices. Controlled pharmacokinetics and dynamics enable proper release of drug amounts at optimal periods during the circadian cycle using physical and chemical signals.
Nanomaterials can be used in medicine for their ability to cross biological barriers and target specific cell populations such as cancerous cells. Nanomaterials can thus be used to develop new therapies such as nanoparticle-based ultrasound or magnetic hyperthermia for the treatment of cancers. Nanoparticles coated with aminosilane are taken up faster by tumoral cells than by normal cells and subsequently heated and destroyed by a magnetic field. Similarly, nanoparticles can be used to concentrate the energy of ultrasound beams in cancers. Moreover, the treatment can be repeated as nanoparticles form stable deposits within tumors.
| Precision medicine|| |
Personalized medicine incorporates differences between individuals to optimize screening, diagnostic, and therapeutic decisions as well as checkup. Individualized medicine is based on patient features, integrating risk factors, lifestyle, genetic variants, familial context, medical history, and circulating molecular markers (e.g., secreted proteins, microRNAs, long a priori nonprotein-coding RNAs, and released microvesicles) as well as structural and functional data obtained from clinical examinations, biological measurements, functional exploration, and imaging as well as modeling and simulations.
Medical strategies are aimed at promoting early diagnosis, improving prognosis, and avoiding unwanted drug effects. Pharmacogenomics assesses the response of individuals to drugs and identifies patients at risk for adverse reactions to given medications using genetic markers.
Computer-aided medicine and surgery hence encompass many topics and fields of knowledge. In this multidisciplinary approach, the three basic natural sciences - biology, chemistry, and physics - interact with mathematics to explain the functioning of physiological systems in normal and pathological conditions.
Anatomy, a descriptive science, which most often did not emphasize the huge between-subject variability, is now supported by the subject-specific 3D reconstruction of the body's organs after acquisition of medical images mainly acquired by computed tomography, magnetic resonance imaging, ultrasonography, and nuclear medicine imaging. These techniques can be coupled to obtain both morphological and functional data.
Personalized models require the 3D reconstruction of organs of interest from a set of medical images, from which a proper mesh can be computed for mechanical analysis. In biomechanical problems, a data assimilation strategy can be adopted to reduce uncertainties of the model using partial observations of the state variables.
The Virtual Physiological Human projects from various countries correspond to a framework program that is aimed at modeling the entire human body, using recent advances in medical exploration technologies and high-performance computing. The latter, indeed, relies on specific algorithms and coupling platforms to enable integration of various models and rapid computations. The objective is the optimization of medical decision by achieving a better understanding and description of pathophysiological processes, predicting outcomes, and developing and planning new customized treatment procedures based on patient data, in parallel with computer-aided design of drugs and medical devices.
However, this bottom-up integrative research strategy still relies on proper representative (i.e., developed after completely understanding the complex reality) top-down reductionist approaches. Any complicated physiological system can be analyzed by decomposition into simple parts with identified functions. The combination of these functions enables to deduce system functioning due to linear interactions. Deconstruction into parts of physiological systems is necessary to understand part behavior as well as to determine between-part interactions.
Physiology that targets the macroscopic scale is now coupled to molecular biology to better handle the process arising in the organism at various length scales. Molecular biology is itself assisted by computational biology that tackles biological phenomena using multisample and multivariate data analysis. For example, computational biology aims at pointing out circumstances that trigger a peculiar signaling pathway, among all existing signal transduction axes. Computational biology incorporates techniques from optimization, high-performance computing, medical image processing, and data mining and analysis.
| Bioinformatics|| |
Bioinformatics deals with the production of biological data analysis software. Objectives include deciphering the relationships between genotype and phenotype, and correlations between individual genome, nutrition, environment, and life mode, among other health-related factors as well as diseases.
Analyses of the transcriptional and translational landscape (allele-specific expression, differential gene expression according to the context, alternative splicing, small RNA-mediated regulation, RNA editing, etc.,) are required to investigate disease-associated traits. Bioinformatics not only target sequence analysis, genome annotation, computational evolutionary biology, and analysis of gene and protein expression but also analysis of the structure and molecular dynamics of biological compounds as well as between-molecule interactions. Due to the large amount of DNA and amino acid sequences and related information to be handled, bioinformatics relies on intensive computations. Bioinformatics supports the development of knowledge and data management platforms to identify predisposition to diseases and biomarkers for early diagnosis and therapy adaptation in a framework of individualized medicine. Tasks incorporate data mining and information retrieval, indexing, and representation.
Biomedical informatics aims at integrating the collection of biochemical data, physiological signals, medical images, biological rhythms, epidemiological factors, clinical history events, and simulation results into usable information to foster the creation of new diagnostic and therapeutic methods and improve both life quality and health-care cost. Bioinformatics thus involves algorithm and computation theory, artificial intelligence, discrete mathematics, control and system theory, databases and information systems, and statistics.
| Biomathematics and biostatistics|| |
Medical statistics, or biostatistics, are currently used in epidemiology, public health, forensic medicine, and clinical research. A major goal is the definition of appropriate indices for early diagnosis of chronic diseases and evaluation of prognosis, the determination of risk factors in the development of chronic pathologies, as well as optimization of treatments (administration time, duration, and dose).
Biomathematics focus on modeling and simulations of biological processes at various length and time scales, from any cascade of chemical reactions within the cell, to the fate of a single cell in a given context as well as the collective behavior of cell populations, to the relation between different compartments of a physiological system and interactions between distinct physiological systems, in addition to pharmacodynamics and pharmacokinetics. In addition, direct observation of the dynamics of biological processes being often impossible, theoretical and numerical methods are developed to recover values of control variables and parameters by solving the inverse problem.
In medicine, applied mathematics constitute the basis of many aspects linked to engineering sciences, that is, in: (1) medical image and signal processing; (2) 3D reconstruction of body's organs and virtual endoscopy; (3) navigation tools to assist peroperating gesture as well as manipulation of materials and deployment of devices in regions of interest; (4) medical simulators for training of handling of exploration techniques as well as mini-invasive methods of medical and surgical treatments; (5) design and shape optimization of implantable medical devices; (6) medical robotics; and (7) telemedicine and telesurgery.
| Biomechanics|| |
Biomechanics, that is, continuum mechanics applied to physiology, is aimed at assessing the function of physiological apparatuses with respect to its morphology (shape), geometry (size), structure (tissue composition), and rheology (mechanical properties) for given prestresses (residual tension) in normal and pathological conditions. It deals with the mechanical behavior of biological fluids (air, blood, cerebrospinal fluid, digestive juices and bolus, lymph, urina, etc.,) and solids (hollow and solid organs), the physical properties of which can be defined by continuous functions.
The goal of biomechanical modeling is to predict the fields of involved physical variables by solving the well-posed boundary value problems associated with balance laws of mechanics (e.g., conservation of mass, momentum, and energy) in a given context. Four prerequisites of any problem in biomechanics include: (1) achievement of the computational domain, that is, personalized geometry with its given structure based on anatomical and histological data; (2) determination of the material constants of the body of interest or of each subdomains in the case of a composite material according to available rheological results, if possible obtained properly in vivo; (3) selection of the equation set associated with the problem based on the governing physical laws, depending on assumptions; and (4) definition of the appropriate initial and boundary conditions that incorporate constraints of the neighborhood.
Biomechanics also contributes to the improvement of diagnosis methods and development of new diagnosis techniques, of new measurement techniques, from signal acquisition to processing, elaboration of proper therapeutic strategies, and conception of new surgical procedures and medical implantable devices. Biomechanics also participates in conceiving, designing, implementing, and optimizing health-related nanotechnology, optimal design of nanoparticles being linked to their size, shape, surface charge, and rheology.
| Modeling and simulation for computer-aided medicine and surgery|| |
Modeling aims at better understanding, predicting, optimizing, and controlling natural processes as well as assessing quantities and parameters that cannot be directly measured, testing the effects of involved parameters, creating new hypotheses and paradigms, and redesigning models and theories according to outcomes, thereby driving rather than complementing investigations. Most often, analytical solutions cannot be used. Numerical procedures are then employed, after basic mathematical and numerical analysis, to check stability and when possible, solution unicity.
Any model is developed in the framework of a theory, such as continuum mechanics, but the model is much more specific than the theory. The simplification degree of the model provides its application domain. Certain assumptions that are not strongly justified can be removed in iterative model refinement. Any model is characterized by (1) its potential to improve a system's knowledge as well as the underlying theory and to generate new concepts, and (2) its capacity to define new strategies.
Modeling entails several stages: (1) a definition stage during which the whole set of data related to the actual system is collected, and the problem and its goals are identified (concrete level); (2) a representation stage of the system, in which the number of data is reduced (information filtering), as data relevant to the problem are extracted and kept to describe the model, a simplified version, and falsification of the reality; (3) a validation stage that results from generality loss by comparisons with available observation data that yield a meaning to the model; and (4) possible model improvements.
Two main varieties of models include representation and knowledge models. The former is associated with mathematical relations between input and output variables. The equation coefficients do not necessarily have physical or physiological meaning. The latter is aimed at analyzing the mechanisms that produce the explored phenomena.
Modeling is associated with numerical simulations of physical and chemical processes. Problem formulation and mathematical analysis lead to the construction of solution algorithms. Approximations result from inherent errors of mathematical models of natural processes that arise from a partial understanding of these phenomena and their random nature, as well as measurement uncertainties. Moreover, model equations cannot be commonly solved analytically. Problem equations are converted to a set of algebraic equations (numerical approximations). Numerical procedures introduce truncation errors associated with numerical approximations and round-off errors, as a finite number of digits are used to represent numbers.
Mathematical modeling based on differential equations is used to predict and redesign experiments for a deeper understanding of regulated cellular and tissular processes, such as organogenesis and tissular remodeling, healing, and repair, keeping in mind that mathematical descriptions afford neither quantitative analyses nor complete solutions and explanations.
Multiscale modeling aims at coupling models that describe cell and tissue events at the nano- and micro-scale to standard macroscale simulations of any explored physiological process. On the other hand, a multilevel modeling couples mechanical models at various spatial dimensions (e.g., 0-, 1-, and 3-D models). Multiphysics modeling couples various physical processes (e.g., fluid and solid mechanics, wave propagation, and heat transfer).
Different types of models can be applied to distinct scales: for example, particle methods for tiny structures, partial differential equations for processes at moderate time and space scales, and ordinary differential equations in reduced models at larger scales. Hybrid models incorporate discrete, reactive, moving, and deformable objects in a continuum.
Programs in computational medicine rely on a software-coupling platform aimed at incorporating structural (anatomical) and functional data at various length and time scales to simulate the functioning of the physiological apparatus in both normal and pathological conditions, thereby promoting personalized, predictive, and if possible, preventive health care. Modeling supports computer-aided diagnosis and treatment planning, as well as prediction of therapeutic outcomes and prognosis.
In medicine, modeling is also aimed at developing pedagogical and medical tools as well as assessing physiological quantities inaccessible to measurements using inverse problems.
Medical signal and image processing provide input data for numerical simulations. Image data are characterized by image quality (contrast, edge quality, and artifacts), source with its given temporal and spatial resolution, noise level, and range of time- and frequency-localized features.
3D reconstruction of segmented surfaces of organs of interest is followed by adapted meshing with coarsening and refinement of some regions. A mesh must be adaptive when coping with evolving processes.
Modeling of complex systems relies on a bottom-up approach, starting from the acquired basic knowledge of the system's parts. Models that integrate nano- and micro-scale processes target reacting adaptive systems. Furthermore, they can investigate responses of living tissues to administered substances, especially drugs delivered by nanotechnology-based methods or implanted medical devices. This type of modeling incorporates the dynamics of cell processes, that is spatial and temporal organization of biochemical reactions and major molecular interactions with positive and negative feedback that amplifies or limits the response, respectively.
| Tier architecture of the organism|| |
Organs are made up of composite material constituted of different cell types embedded in an extracellular matrix. Altered cell and matrix mechanics and dysregulated mechanotransduction are observed in hypertension with increased wall stiffness, among other disorders.
Both the intra- and extracellular media, which are composed of water, a ground substance, filaments (cytoskeleton and elastin and collagen fibers), and particles (cellular organelles and cells), can experience fluidization or gelification according to the exerted mechanical stress. These two media are chemically linked by plasmalemmal molecules such as integrins, which are responsible for in-out or out-in signaling.
The cell and tissue organization, homeostasis, adaptation, control, and remodeling are not associated with mathematical theories of optimal structure-function relations, as general quantities measuring the complex cell behavior are lacking. Adaptive and regulated cellular and tissular components cannot be isolated without missing the integrated function of the entire structure. In addition, chemical activities underlying the structure-function relations obey to various rhythms governed by biological clocks. Hence, even in the absence of major changes in cell fate (e.g., growth, division, differentiation, and death), molecules are not only subjected to a constant turnover with given synthesis and degradation rates but also their concentrations fluctuate in time.
The tier architecture of any living system is characterized by its communication means and regulation procedures. It enables integration of environmental changes to adapt. Multiple molecules interact to create the adaptable activity of the cells, tissues, organs, and body. Any integrative model then incorporates a set of models developed at distinct length scales and includes characteristic times to efficiently describe the structure-function relationships of the explored physiological system. One challenge is to couple models that function at different scales, from the order of the nanosecond to the week and from the order of the nanometer to the centimeter.
Biological systems - from the molecular level to the physiological apparatus - are characterized by their complicated structure, variable nature, and complex behavior. The processing of the signals that control the activity of transcription factors and the expression of genes to direct cell decision (differentiation, growth, proliferation, or death), organization of metabolism, and cellular communication for coordinated action in a tissue, relies on nonlinear dynamics that control spatial distribution and clustering of molecular species at a given time.
| Adaptation and remodeling|| |
Living organisms react to environmental changes and maintain the stability of the internal medium by regulation systems (homeostasis). In addition, humans adapt the homeostasis level (heterostasis) to the environment (e.g., sea level vs. mountain and desert vs. icy region).
Biological tissues are capable of short- and long-term adaptation (remodeling) to applied constraints that causes a change in configuration and structure. These constraints include mechanical stresses. The mechanical behavior of biological tissues depends on (1) type of applied forces (tension, compression, shear, and torsion) and their possible combination; (2) loading orientation, magnitude, duration, and rate; (3) eventual periodicity (frequency); and (4) state of the surrounding environment and organism that regulate perfusion (hence, temperature and moisture).
Abnormal situations are also characterized by maladaptive tissue remodeling. Surgical procedures such as grafting and minimally invasive, catheter-based implantation of medical devices such as stents aimed at correcting risky tissue growth can themselves trigger another type of aberrant tissue growth such as intimal hyperplasia.
| Nonlinear dynamics and complexity|| |
Complexity arises from the large number of involved quantities that are related by nonlinear relationships. For example, the cell division cycle is controlled by a huge set of regulators, mainly cyclin-dependent kinases that have many activators and inhibitors. The genome must be precisely replicated and accurately partitioned into its two new progeny. Progression through the cell cycle is thus governed by interactions between proteins as well as between genes and proteins.
The spatiotemporal organization of living systems characterized by feedback and feedforward is associated with self-organization and nonlinear dynamics, distinct types of initial conditions leading to different behaviors. Mutually influencing variables can evolve toward a restricted region of the phase space, the so-called basin of attraction endowed with its single attractor. At bifurcation points, attractors shift and the behavior can change to chaotic motions. Chaotic dynamics enables quick adjustments that are mandatory in physiology.
The cell is a complex system constituted by many components. The features of complex systems are adaptation, self-organization, and emergence. Cells self-organize to operate with optimal performance. The behavior of a complex system is not necessarily predictable from the properties of its elementary constituents, which can nonlinearly interact with feedback loops, contributing to system bulk behavior. The organization and bulk behavior of a complex system not only results from the simultaneous activities of its constituents but also emerges from the sum of the interactions among its constituents. A complex system adapts by changing its organization and possibly its structure in response to environmental stimuli. Yet, a predictive model requires a theory, or at least a framework, that involves relations.
| Physiological and pathological processes - Cell signaling|| |
Cells have the capacity to emit as well as to receive, decode, and transmit information efficiently, and to integrate new signals. Any cell experiences numerous, simultaneous or successive, events that result from permanent, regulated communications between it and adjoining cells and the extracellular matrix as well as remote controller cells of the nervous and endocrine systems, in addition to immunocytes.
Cell fate is controlled by messengers synthesized by and secreted from cells to act on the cell itself (intra- [without release] and autocrine control), its neighbors (juxta- and paracrine control), or distant cells (endocrine control). Once it has reached its cell target, any messenger triggers a signal transduction axis characterized by a cascade of chemical reactions from the cell surface to the cytoplasm, and possibly the nucleoplasm, that allows the cell to adapt. Therefore, cells appropriately respond in a controlled, coherent manner to external stimuli (adaptation robustness). Specific responses with their respective intracellular biochemical reaction cascades can be generated over a wide range of parameter variation. Signals are transduced by information processing networks that are characterized by signal transduction complexity and between-pathway connectivity. Fast protein modifications that result from protein interactions in the cytoplasm propagate signals and can lead to slow transcription and translation. Cell signaling not only governs basic cellular activities but also coordinates actions of cell populations.
Signal transduction starts at the cell surface, where chemical messengers bind and/or stimulate specific sensors and receptors that trigger cellular responses. Plasmalemmal receptors that initiate signaling can be: (1) enzymes such as receptor kinases and phosphatases, membrane-tethered guanylate cyclases, nicotinamide adenine dinucleotide phosphate oxidases, and transfer ATPases; (2) proteins coupled to enzymes such as G-protein-coupled receptors and pseudoreceptor kinases; (3) some types of cytokine receptors; (4) ions channels; (5) adhesion molecules; and (6) mass-transfer receptors, which transport a first chemical messenger; in addition to (7) specialized plasmalemmal nanodomain components that participate in endocytosis, during which signaling can be launched. Members of these various groups of plasmalemmal receptors can cluster and form the so-called transducisome.
In addition, physical agents and mechanical stresses can efficiently deform these sensors and receptors, open their proper activation domains, and deliver messages into the cell. Any plasmalemmal molecule can undergo a strain with a more or less large conformational change when stretched, but most of them do not send any signal. Among plasmalemmal receptors, some are mechanogated and mechanosensitive, that is, transmit signals directly or indirectly. Hence, mechanical stress deforms and activates mechanogated and mechanosensitive molecules that can then initiate signaling by stimulating, directly or not, a second messenger.
Initiated signaling cascades are usually composed of multiple nodes that correspond to major mediators among effectors, scaffolds, and adaptors, which are classified according to their structural and functional features.
Signaling axes prime the release of substances from intracellular stores as well as the gene expression to synthesize messengers for the cell itself and its neighbors, whereas endocrine cells and neurons are responsible for remote control. Intracellular cascades of chemical reactions with signaling nodes, hubs, and modules are represented by mathematical models.
Once information is collected, which describes the huge number of signaling mediators and their main known features, mathematical modeling depicts signaling cascade using a suitable data set with a minimal content in required quantities. Handling of signaling networks and their complex behavior leads to a compulsory preview to distinguish primary from secondary elements. In physics, a similar strategy relies on phenomenological analysis and scaling.
The selection stage is not an obvious task as (1) the number of known molecular participants of any signaling cascade is often quite large; (2) involved effectors possess many names; (3) some effector aliases designate different types of molecules; (4) most mediators interact with many partners; (5) cross talk exists with other signaling axes; (6) the finely tuned intracellular cascade of reactions has a complex functioning; (7) and some regulators and mediator properties as well as values of kinetic coefficients remain unknown.
Signal transduction pathways regulate cell fate and adaptation to its environment. A biochemical reaction network is defined by (1) a set of variables - the state variables - that define the state of the system and (2) rules of temporal changes and possible transport of involved variables. Reaction-transport equations describe the concentration differences that result from production, consumption, and transfer of these chemical species between the various subcellular compartments from the plasma membrane to the cell nucleus, or vice versa, as well as storage and release of molecules.
A given reaction occurs with a certain probability. Stochastic models rely on a stochastic update of system variables. On the other hand, deterministic models are carried out in systems with a large number of molecules, the time and variable states uniquely defining the state at the next time step. In a deterministic continuous formulation based on the mass action formalism, molecular reactions are described by differential equations defining the rate of change of molecular concentrations. A rate constant (or affinity parameter) describes the occurrence rate of a reaction when reactants are close each other.
Among influence agents, mechanical forces and derived stresses can prime molecule release and gene transcription in cells that bear time- and space-dependent mechanical stress fields. These fields participate in the regulation of cell fate (growth, differentiation, and migration) and tissue remodeling.
Mathematical modeling for digital medicine can then be developed at different spatial and temporal scales, from the signal transduction pathway (nanoscopic scale) to the cell reaction (microscopic scale) and tissue adaptation and remodeling (mesoscopic scale).
| Authors|| |
Marc Thiriet' was educated at Medicine Faculty of Lille and University Pierre and Marie Curie ([UPMC] MD), and then at Technology University of Compiègne (3 rd cycle Doctorate in Biomechanics), and Physics College of University Denis Diderot (Accreditation to Supervise Research). He was assistant physician in the lung disease department of Pontoise hospital. He is currently a member of the INRIA-UPMC-CNRS team REO in Laboratory Jacques-Louis Lions (applied math.) of UPMC. He worked in flows in collapsible tubes applied to airways and veins, 3D unsteady developing laminar flows in bend and branchings, both experimentally and numerically, as well as in models derived from 3D reconstruction of human anatomy. He is now involved in mathematical modeling of biological processes. Marc Thiriet is the author of Biology and Mechanics of Blood Flows (2~Vols.) and the book series "Biomathematical and Biomechanical Modeling of the Circulatory and Ventilatory Systems'' (7~published books). He was the associate editor for the medical encyclopedia "Pan Vascular Medicine". He also wrote the chapter 30. Biofluid Flow and Heat Transfer of Handbook of Fluid Dynamics. He is the President of the French Committee for Intensive Computation in Biology and Medicine.