Results 1  10
of
14
Decisionmaking, errors, and confidence in the brain
 Journal of Neurophysiology
"... a fundamental basis for understanding decisionmaking and decision confidence, we analyze a neuronal spiking attractorbased model of decisionmaking. The model predicts probabilistic decisionmaking with larger neuronal responses and larger functional magnetic resonance imaging (fMRI) bloodoxygen ..."
Abstract

Cited by 6 (1 self)
 Add to MetaCart
a fundamental basis for understanding decisionmaking and decision confidence, we analyze a neuronal spiking attractorbased model of decisionmaking. The model predicts probabilistic decisionmaking with larger neuronal responses and larger functional magnetic resonance imaging (fMRI) bloodoxygenleveldependent (BOLD) responses on correct than on error trials because the spiking noiseinfluenced decision attractor state of the network is consistent with the external evidence. Moreover, the model predicts that the neuronal activity and the BOLD response will become larger on correct trials as the discriminability �I increases and confidence increases and will become smaller as confidence decreases on error trials as �I increases. Confidence is thus an emergent property of the model. In an fMRI study of an olfactory decisionmaking task, we confirm these predictions for cortical areas including medial prefrontal cortex and the cingulate cortex implicated in choice decisionmaking, showing a linear increase in the BOLD signal with �I on correct trials, and a linear decrease on error trials. These effects were not found in a control area, the orbitofrontal cortex, where reward value useful for the choice is represented on a continuous scale but that is not implicated in the choice itself. This provides a unifying approach to decisionmaking and decision confidence and to how spikingrelated noise affects choice, confidence, synaptic and neuronal activity, and fMRI signals.
Teletactile system based on mechanical properties estimation
, 2011
"... Abstract. Tactile feedback is a major missing feature in minimally invasive procedures; it is an essential means of diagnosis and orientation during surgical procedures. Previous works have presented a remote palpation feedback system based on the coupling between a pressure sensor and a general ha ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract. Tactile feedback is a major missing feature in minimally invasive procedures; it is an essential means of diagnosis and orientation during surgical procedures. Previous works have presented a remote palpation feedback system based on the coupling between a pressure sensor and a general haptic interface. Here a new approach is presented based on the direct estimation of the tissue mechanical properties and finally their presentation to the operator by means of a haptic interface. The approach presents different technical difficulties and some solutions are proposed: the implementation of a fast Young's modulus estimation algorithm, the implementation of a real time finite element model, and finally the implementation of a stiffness estimation approach in order to guarantee the system's stability. The work is concluded with an experimental evaluation of the whole system. Keywords: Tactile feedback, haptic, virtual environment, Kalman filter Tactile feedback in surgery In common clinical and surgical practice, tactile feedback is considered as a real diagnostic tool. Many diseases cause hardening of organ tissues, allowing the doctor to identify them using only the sense of touch. Commonly the tactile sense is used for locating tumours, identifying gallstones as a harder ball embedded in a soft background, or localizing a vessel's path, which is identified as pulsing bodies. In the last years many surgical procedures have been converted from open to minimally invasive. Open surgery is performed by accessing internal organs via a wide incision in the patient's skin. These wide openings allow the doctor to reach anatomical structures with his hands and thus perform accurate palpation inspections. During min * Corresponding author. Email: Mauro.Sette@mech.kuleuven.be. imally invasive procedures the access to the body is achieved with small incisions in the patient's skin. The surgeon inserts longshaft surgical instruments through these openings, while the sight is guaranteed by a laparoscope. If on one side minimally invasive surgery allows trauma reduction with subsequent reduction of pain and recovery time after the surgery, on the other the sense of touch is strongly reduced when this last technique is used. Tactile feedback is even completely absent when the direct manipulation of a minimally invasive instrument is substituted with telemanipulation. A telemanipulation system is a device that enables a person to perform manual operations while separated from the site of work (an example of a commercial system is the da Vinci Robotic Surgical System (Intuitive Surgical, Mountain View, CA, USA)). It is then extremely important to restore the tactile sense in the telemanipulation procedures, developing a device that 1176 2322 /11/$27.50 © 2011 enables a person to touch an object or surface while separated from the site where the object/surface is, using the analogy with the telemanipulation system it can be called teletactile system. A teletactile feedback system for minimally invasive surgery This work presents the study of a new system for the restoration of tactile feedback in minimally invasive surgery. The tactile sense is a complex sense: it provides information about shapes, stiffness, and texture. This is the reason why the design of a complete tactile feedback system is not an easy task Nicholls and Lee [15] define a tactile sensor as a device or system that can measure a given property of an object or contact event through physical contact between the sensor and the object. Such a tactile sensor is based on the interaction force between an object and the sensor. If we transfer this definition to the case of searching for liver metastases, we can make a mental scheme in which we have a certain volume with nonhomogeneous structure and we measure the distribution of reaction forces observable on the surface when the volume is subjected to a compression. From a certain point of view, we measure not the real material stiffness but only how this stiffness distribution affects the reaction force. This measurement gives information on the surface stiffness resulting from the interaction of different deep layers with different stiffnesses. It could be useful to have the information on the stiffness distribution in the whole volume because it can happen that important information has a too weak effect on the surface stiffness, and identifying the hard inclusion becomes difficult. This is the idea driving this work: imagine a system where, instead of the surface reaction force, we could measure the stiffness distribution directly by means of a quantitative elastography approach This work is going to propose an implementation of such a teletactile system, and is going to analyze its properties and performances. System design and implementation At a concept level a teletactile system could be designed as a tool for a robotic system, (e.g. the da M.M. Sette et al. / Teletactile system based on mechanical properties estimation 239 Vinci System) composed of four different modules running in three different loops, as shown in The following sections are going to describe in details the four blocks composing the system. Module a Control of the ultrasound probe position through the robotic arm A laparoscopic ultrasound probe that is manipulated by a robotic arm, for example the da Vinci CANVAS described in (1) While the system is engaged in the loop 2 at the time T, calculate an adjacent portion of volume that will be explored at the time period T+1 Both solutions could be implemented efficiently. Each solution is strictly dependent on the haptic interface used. For a kinaesthetic feedback device, the first solution would be preferable because in this case it is necessary to explore the surface along one direction. The second may be more suitable when the system feeds back the information through a tactile device and bidimensional information is needed. In this work, the application for a kinaesthetic device will be developed, and thus a bidimensional model will be implemented, allowing exploration along only one direction. 240 M.M. Sette et al. / Teletactile system based on mechanical properties estimation Module b Young's modulus estimator The time requirements for the Young modulus estimation depend on exploration speed at which normally the fingers explore a surface. The block b calculates the Young modulus distribution of a finite volume portion, and as long as the finger stays in this portion, no new calculation is required. The best control algorithm that is needed to realize this synergy between loop 1 and loop 2 is still the object of investigation. If we assume that at each period T we are able to calculate a surface with width w (which is the width of the LaparoscopicUSprobe's footprint under the condition of using a linear probe) and considering that the maximum exploration speed v ex for the finger is 120 mm/s [24] the update frequency (f u ) of loop 3 is given by: If we consider that footprint of the probe in M.M. Sette et al. / Teletactile system based on mechanical properties estimation 241 The estimation of mechanical properties can be implemented using different strategies, the state of the art lists two different approaches, iterative and direct. Both approaches have advantages and drawbacks that must be analysed in order to make the correct choice for an eventual development. Iterative methods have been studied mainly; they have the big advantage of being less sensitive to noise as well as stable; in fact they are based on the solution of a nonlinear least squares problem that can be solved with optimization algorithms like the GaussNewton or LevenbergMarquardt method There exists a third possibility for the stiffness estimation, which is the use of the shear wave elastography; this technique is very promising, and eventually allows a fast estimation of the mechanical properties. However it has not been studied by the authors of this work and will not be considered in the discussion of possible systems. Nevertheless, it remains an interesting approach that needs to be further analysed. For more detail on this approach refer to Bercoff et al. Module c Real time finite element model Let's suppose that the information on the stiffness distribution is efficiently estimated from the intra operative ultrasound images, as example we report the stiffness distribution calculated with a direct method from a set of ultrasound images: where u is a 2nelement, nodal displacement vector; u andü, the respective velocity and acceleration vectors; F , the external force vector; M, the mass matrix of size (2n, 2n); D; the damping matrix; and K the stiffness matrix, while n is the number of nodes in the FEM. During surgical manipulation, dynamic effects are negligible (1) The problem to solve was identified. The numerical solution which approximates the solution of the elasticity problem, or the NavierLamé problem [1], was found. (2) The FEM was implemented and the boundary conditions were integrated. The numerical integration of the constitutive equation leads to the definition of a linear system. The coefficient's matrix K is built considering the model composed of quadrilateral meshes and linear hat functions (2n, 2n). The vector of the unknowns x (2n) is the vector of the nodal displacements. Finally the vector of the external forces f (2n) is applied on each node (n is the number of nodes). The multiplication factor 2 is the number of degrees of freedom (d.o.f.). The formal description of how to build the FEM is given in The Dirichlet boundary conditions (DBC), which are the displacement imposed at the M.M. Sette et al. / Teletactile system based on mechanical properties estimation 243 nodes, will be applied by means of the Lagrange multipliers. The displacement is fixed in one specified direction and is possibly free in others. This can be translated in defining the system: (3) The linear system was solved with a classic algorithm like Gaussian elimination Module d Haptic interface, hardware and control The haptic interface Control of the haptic feedback In past years researchers have been working on finding a solution able to give a realistic feeling and at the same maintaining the stability of the controlled system. The aim of this section is not to give an exhaustive explanation of the complex theory behind the haptic feedback and telemanipulation, but instead to treat one of the main problems that could be present when interacting with a virtual (or semivirtual) haptic environment: the instability generated by the low update rate in the simulation engine. The virtual environment updating frequency is dependent on the implementation of the FEM and the type and size of FEM used. Colgate et al. M.M. Sette et al. / Teletactile system based on mechanical properties estimation continuous time domain. Colgate's major finding is a formulation of a general rule, based on the passivity theory where ω N = π/T is the Nyquist frequency. From Colgate's rule it can be concluded that, in order to maintain stability, the parameter T should be reduced. A common approach is to linearly interpolate between two values instead of holding the value of f FEM between two periods. A better approach could be to use a stiffness estimation scheme. This scheme was implemented using the formalism of the extended Kalman filter as previously presented by De Gersem in A new concept for contact control The avatar, a software representation of the device's tip, represents the haptic device in the virtual environment; usually the model to be touched and the avatar are incompenetrable; that is, they can only deform, and this property reflects the real object's behaviour. In the experimental phase, this constraint will be removed, giving the user the possibility of touching under the surface; that is, touching a series of virtual surfaces generated below the surface representing the natural organ's surface. The expected advantage is the augmentation of the tactile sense, leading to a more powerful tool for the identification of metastases. In order to implement the capability of touching under the surface it is necessary to formulate a novel scheme for the interaction between the avatar and the virtual surfaces inside the volume Materials and methods The first section of this paragraph explains which platform has been used for the physical implementation of the system. Then the system has been tested with respect to several aspects. The first experiment sets are aimed at characterizing the system under the performance aspects. The speed of calculation of the virtual environment has been tested, varying the model's size; the stiffness estimation has been tested against the normal stiffness; and finally the contact control strategy has been tested evidencing how different the explored stiffness appears when the boundary conditions are changed. The experimental session concludes with a series of experiments evaluating the system's performances using psychophysical tests. M.M. Sette et al. / Teletactile system based on mechanical properties estimation 245 Hardware and software platform The haptic rendering block has been implemented in the PXI Real Time System (CPU1) (National Instruments, Austin, TX, USA) with a dedicated program written in Labview (National Instruments, Austin, TX, USA). The simulation engine block implements the FEM for the elastic interaction between bodies and was implemented using a pseudocode software: Octave System's performances Effect of the Kalman filter As explained in Sec. 3.4.1. the stability of the system is improved by the introduction of a stiffness estimation scheme. This block uses a Kalman filter to decouple the virtual environment from the haptic interface controller. It is necessary to measure which is the difference between the stiffness estimated with the Kalman filter and the stiffness estimated in a ideal case. We expect that the Kalman filter introduces some phase lag in the estimation due to the filtering process. In order to estimate the phase lag we compared a "real stiffness" that is the stiffness used as input to the filter with the stiffness estimated with the Kalman filter. The experiment is conducted, recording the stiffness estimated during the surface exploration, the exploration was simulated shifting the node on which the force was applied from one side of the model and back at the speed of 120 mm/s, and at the same time the degrees of freedom on which the indentation displacement is applied are recorded as a function of the time. Those last data are then used to calculate the real stiffness offline. Calculation speed over the distributed system A major limiting factor for the system's performance is the FE calculation time and the speed of communication between different systems. A test of the time required for the FE calculation and the communication time required to transmit the data is performed. Different FE models where tested with degrees of freedom ranging from 408 to 1071 (the FEM number of columns was kept fixed to 50 while the amount of lines was variated from 8 to 20). New concepts for contact control and stiffness augmentation The user perceived stiffness is measured as the ratio between the indentation force and the nodal displacement on which the force is applied. This quantity is Fig. 10. The virtual environment is divided in two section each working on an independent processing unit (CPU), in the first block in calculated the interaction point of the haptic avatar with the virtual environment and there is block for the control of the haptic device; while the on the second processing unit are solved the calculation of the virtual environment. The two blocks are physically on two different units and are connected with a TCP/IP protocol. considered as index of the stiffness perceived by the subject. A set of experiments was conducted with the aim of measuring this quantity over different touching surfaces of a FE model. Aim of this experiments is to highlight the advantages brought by the semivirtual environment. A linear elastic FE model was created, the plain strain approximation was implemented. The mesh was built with 50 by 10 quadratic elements. In the FE model was simulated a hard inclusion embedded in a softer background. The YM of the inclusion was five times harder than the background. The model is presented in Psychophysics experiments, subjects and protocol The aim of the psychophysical experiments was the testing of the system in terms of the subject's ability to discriminate a hard inclusion embedded in a soft envi M.M. Sette et al. / Teletactile system based on mechanical properties estimation 247
NeuroImage 53 (2010) 694–706 Contents lists available at ScienceDirect
"... journal homepage: www.elsevier.com/locate/ynimg ..."
(Show Context)
Neural encoding Attractor network
, 2011
"... Available online at www.sciencedirect.com ..."
(Show Context)
Contents lists available at SciVerse ScienceDirect Progress in Neurobiology
"... jo u rn al ho m epag e: ww w.els evier.c om /lo cat e/pn eu ro b io Brain mechanisms for perceptual and rewardrelated decisionmaking ..."
Abstract
 Add to MetaCart
(Show Context)
jo u rn al ho m epag e: ww w.els evier.c om /lo cat e/pn eu ro b io Brain mechanisms for perceptual and rewardrelated decisionmaking
Contents lists available at ScienceDirect Progress in Neurobiology
"... journal homepage: www.elsevier.com/locate/pneurobio Stochastic dynamics as a principle of brain function ..."
Abstract
 Add to MetaCart
(Show Context)
journal homepage: www.elsevier.com/locate/pneurobio Stochastic dynamics as a principle of brain function
Neuron Review Decision Making in Recurrent Neuronal Circuits
"... Decision making has recently emerged as a central theme in neurophysiological studies of cognition, and experimental and computational work has led to the proposal of a cortical circuit mechanism of elemental decision computations. This mechanism depends on slow recurrent synaptic excitation balance ..."
Abstract
 Add to MetaCart
(Show Context)
Decision making has recently emerged as a central theme in neurophysiological studies of cognition, and experimental and computational work has led to the proposal of a cortical circuit mechanism of elemental decision computations. This mechanism depends on slow recurrent synaptic excitation balanced by fast feedback inhibition, which not only instantiates attractor states for forming categorical choices but also long transients for gradually accumulating evidence in favor of or against alternative options. Such a circuit endowed with rewarddependent synaptic plasticity is able to produce adaptive choice behavior. While decision threshold is a core concept for reaction time tasks, it can be dissociated from a general decision rule. Moreover, perceptual decisions and valuebased economic choices are described within a unified framework in which probabilistic choices result from irregular neuronal activity as well as iterative interactions of a decision maker with an uncertain environment or other unpredictable decision makers in a social group.