Developing Objective Functional Metrics for Advanced Upper Limb Prosthetics

By Paul Marasco, PhD; Jacqueline Hebert, MD, FRCPC; and Jon Sensinger, PhD, PEng

Prosthetic limb technology has advanced significantly in recent years, but there is no standardized set of metrics to evaluate these technologies. This lack of objective information leaves insufficient evidence to guide research and medical decision-making. It also hinders the ability to communicate benefits to patients and demonstrate improved outcomes to insurance payers.

A Cleveland Clinic-led research team has been awarded up to $2.5 million through a contract from the Defense Advanced Research Projects Agency (DARPA) to develop a suite of outcome metrics for advanced prosthetic limbs that are clinically relevant and rooted in cutting-edge science.

Our project’s goal is to inform future prosthetics development; help physical medicine and rehabilitation physicians make better clinical care decisions regarding prosthetic selection; justify to payers the need for advanced prosthetic devices; and ultimately improve the quality of life for patients with upper limb amputations. Clinical practice will not change until it can be demonstrated to insurers that technological advances in prosthetics actually improve outcomes.

Figure 1. A neural-machine interface on this advanced prosthetic arm enables hand closure to be controlled by thought. The robotic hand also is equipped to provide sensory feedback to the user. (Photo by Paul Marasco, PhD.)

The contract is through DARPA’s new Hand Proprioception and Touch Interfaces (HAPTIX) program, which aims to deliver naturalistic sensations to amputees and enable better control over their prosthetic limbs.

The research team of neuroscientists, clinicians and engineers intends to develop and validate a battery of functional metrics for advanced prosthetic limb systems that are integrated with the nervous system to provide intuitive motor control and relevant feedback for tactile and proprioceptive sensation.

This Sensory-Motor Prosthetic Evaluation Suite will focus on the assessment of the amputee-prosthetic system in clinical application. Using functional tasks, videography, virtual/augmented reality, motion-capture equipment and analytic software, the various metrics will quantify key aspects of sensory-motor functional performance to provide evidence on effectiveness. The group of tests will be correlated to current standard outcome metrics for upper limb prosthetic users, and will require minimal technical expertise to operate.

Getting a GRIP

The metrics will include the following approaches:

Grasping Relative Index of Performance (GRIP). Principal tasks for prosthesis use involve grasping, gripping or squeezing. Quick and accurate application of desired forces is critical for manual manipulation, from holding hands to heavy lifting, and is necessary for obtaining fluid, natural prosthesis use. Fitts’ law is a widely applicable descriptive model relating the time required to achieve a target to movement size and accuracy. Applying this law to grip forces across the dynamic range of a terminal device is expected to quantify a relative effective accuracy and index of performance, irrespective of the prosthetic control scheme or feedback modality.

Prosthesis Efficiency and Profitability (PEP). Humans use their hands to acquire and manipulate objects for a multitude of activities of daily living. This involves seamless interaction between motor control, touch and proprioception. Upper limb prosthetics replace this lost functionality, and their intrinsic utility is reflected in relative efficiency of use. Accepted methodologies from evolutionary ecology provide a mathematical model-based framework for assessing efficiency and profitability in complex biological systems. Optimal foraging theory, a model that helps predict an animal’s food-searching behavior, will be used as a platform to assess objectively the searching, reaching, grasping, manipulating and decision-making movements involved with prosthesis use.

Gaze and Movement (GAM). The movement of our eyes to specific locations is intimately tied to the demands of a task and is an excellent correlate of our mental focus point. Visual attention is an integral component of motor performance that is expected to change with accurate sensory feedback from the prosthesis and intuitive motor control. The combination of movement and eye tracking during simulated real-world tasks will identify the motion of the prosthesis, compensatory body movements and simultaneous visual gaze behavior. This test will assess the effect of advanced motor control systems and tactile and kinesthetic feedback on movement and visual attention.

3-D Gaze and Movement (3-D GAM). The function of a prosthesis is defined not only by the movement of the device and the body, but also by the movement of grasped objects within their environmental context. Expanding on the gaze and movement tracking described above, goal-directed tasks will be assessed in a real-world environment. The movement of task-critical objects will be joined with 3-D gaze points rendered into a workspace that includes a full breadth of active reaching, grasping and placement tasks. This will use simultaneous tracking of how an object is moved, where the gaze is centered in space, and how the body and prosthesis move during relevant goal-oriented functional tasks to evaluate the benefits of enhanced motor control, including proprioceptive and tactile sensory feedback.

Prosthesis Incorporation (PIC). Measuring how much a prosthesis has been incorporated into the body schema is likely a good indication that control and sensory feedback are intuitive, synchronized and meaningful. We will use the cross-modal congruency effect (which tracks reaction time as the brain “figures out” which different stimuli to combine to feel integrated across vision, touch and movement) to evaluate prosthesis incorporation and assess the ability of a person to ignore one form of feedback in favor of another. We seek to develop an index that can be assessed quickly and accurately, using a simple standardized setup, while being sensitive to control source, tactile feedback fidelity and kinesthetic feedback.

Control Bottleneck Index (CBI). Brain signals can be corrupted by noise from many sources. To counteract this, the brain forms models that help it predict limb control and interpret feedback. Good prosthetic performance can be achieved with many combinations of control and feedback, but performance may be limited by system bottlenecks such as noisy control signals, or errors in feedback with respect to touch or movement. Improvements in feedback and control may not lead to significant improvements in performance until these constriction points are relieved. This work will identify the processing bottlenecks for given tasks, and evaluate the contribution of particular control strategies or sensory feedback modalities independent of the specific impediments.

(Left) The Grasping Relative Index of Performance (GRIP) test uses a Fitts’ law-inspired model to quantitatively describe a person’s ability to quickly and accurately apply grasping forces with their hand or similar device. The GRIP testing device (white cylinder) and software run on any computer through a standard USB port. Here, an able-bodied test subject attempts to accurately produce specific grasping forces while wearing a sensory feedback-enabled bypass prosthesis. (Photo by Zachary Thumser, Louis Stokes Cleveland Veterans Administration Medical Center.) (Right) Motion-capture markers (anatomic and rigid body) are used to test functional task protocols. Each reflective point follows the motion of a key part of the body, most of which are involved in compensatory movements during prosthesis use. The full system combines gaze monitoring to assess visual attention (how often a user looks at their prosthesis to operate it) and tracked compensatory movements to help researchers assess how closely a given technology mimics normal function. Normative values for each task are derived from able-bodied visual attention and movements. (Photo by Jacqueline Hebert, MD, FRCPC, University of Alberta.)

Paving the way for lifelike robotic limbs?

These metrics could conceivably change the way that prosthetic devices are defined. With the Sensory-Motor Prosthetic Evaluation Suite, we want to help usher in a new era of prosthetic limb replacement by having our evaluation approaches be sensitive to testing advanced multifunctional robotic prosthetic limbs with sensory feedback directly connected to the prosthetic user.

Along the way, we want to advance substantially how we demonstrate the benefits of currently available standard-of-care prosthetic limb systems.

Project principal investigator Dr. Marascois an associate staff member of Cleveland Clinic’s Lerner Research Institute and a researcher at the Louis Stokes Cleveland VA Medical Center’s Advanced Platform Technology Center. Project co-investigator Dr. Hebert is an associate professor on the University of Alberta’s Faculty of Rehabilitative Medicine, where she directs the Bionic Limbs for Improved Natural Control (BLINC) Lab. Project co-investigator Dr. Sensinger is an associate professor of electrical and computer engineering at the University of New Brunswick, where he is the Associate Director of the Institute of Biomedical Engineering.

Photo at top of post: Direct sensory feedback is an important component of prosthesis use that has, until recently, been unavailable to users. The fingers and thumb of this advanced system are individually controllable, and the pointer finger is equipped with a touch system that helps the user feel that the hand is part of his or her body.