Scanning Enables Unique Robotic Aid for Stroke Victims

A 1.844Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE

Several attempts have been made over the years by different organisations to develop a practical and fully functional bionic aid to people who have lost mobility in one of their hands due to a stroke. And while such bionic aids that exist today as a result of these attempts do provide a degree of hand mobility to stroke victims, they all suffer from a number of weaknesses. These include limited capabilities in terms of the amount and flexibility of movement they provide and the fact that they tend to be cumbersome devices that leave a lot to be desired from an aesthetic standpoint.

Now, however, Dr. Kazem Alemzadeh and his team from the Robotics Research Group in the Department of Mechanical Engineering at the University of Bristol, is convinced that, thanks to the latest in 3D scanning technology, he is well on the way to providing the answer to the shortcomings of existing robotic aids for stroke victims. He also sees the possibility of the technology being further developed for use by people with other conditions that restrict the use of their limbs–or who have lost a limb.

The key to Dr. Alemzadeh’s work is the ability that today’s advanced 3D scanning technology has given him to create, for the first time, a highly accurate 3D geometric digital model of the skeleton of a human hand that is clinically correct, with each individual bone identified according to accepted hand surgery standards and procedures. This 3D digital model now forms the basis upon which the work of his team to design and develop an intelligent, natural-looking bionic glove, dubbed the `Intro-Gluv’ is now proceeding.

Glove versus exoskeleton
As Dr. Alemzadeh explains, there are fundamentally two possible approaches to the development of a robotic hand rehabilitation device for stroke victims: the dorsal and the palmar, or glove approaches.

The majority of aids available today use the dorsal approach, which involves an exoskeleton incorporating the mechanics of the device. This sits on the back of the hand and is activated by sensors on the dorsal muscles. As well as being unattractive and cumbersome, this approach provides only limited movement due to the nature of the dorsal muscles’ actions.

So Dr. Alemzadeh and his team have opted for the glove approach. This approach, as the name suggests, does away with the need for an exoskeleton and instead, incorporates sensors, actuators and artificial tendons, etc. into different layers of a silicon glove which the stroke victim wears like a normal glove. This approach has the advantage of using the palmar muscles of the hand as well as the dorsal muscles to provide greater flexibility of movement and a much more natural type of movement. The glove approach also has the considerable advantage of enabling a far more realistic-looking and aesthetically pleasing device to be produced.

The essential first step in Dr. Alemzadeh’s research and development project was to acquire an accurate 3D digital model of the skeleton of a human hand.

However as he says, "As far as we are aware there are no digital models of the hand openly available. The only answer for us therefore was to develop our own by scanning a physical model, such as those used in medical schools. So this is what we have done. It has enabled us to produce what we believe to be the first clinically correct, fully annotated and deconstructable digital model of the human hand’s skeleton. Without today’s advanced 3D scanning hardware and software technologies we wouldn’t have been able to do this", he states.

Physical to digital
For this essential first phase he turned to Bristol-based 3D scanning technology specialists, 3D Scan Alliance, who agreed to carry out the scanning and digital model creation using a Solutionix Rexcan CS+ system.

The Rexcan CS+ provides a complete, integrated hardware and software 3D scanning solution that can sit on a desk. Using Solutionix’ Blue LED technology to provide high resolution, high accuracy structured light scanning, rather than the laser technology found in most other scanners, the scanner in the Rexcan CS+ system comes with a dedicated, portable stand, a fully integrated and automated 2-axis tilt turntable for the object to be scanned and Solutionix’ ezScan software for data capture and processing.

To start, the complete hand skeleton was scanned with the scanner fitted with a 400mm field-of-view (FOV) dual lens-set in order to provide a reference scan of the hand with all bones attached. This set-up provided a scanning resolution (point spacing) of 0.23mm, with better than 30 micron accuracy.

The physical model was positioned on the turntable and the scanner was set to look down on it at 45 degrees. With Solutionix ezScan software controlling and coordinating the turntable’s motion and the scanner’s data capture processes, scans of the upper surface of the hand were then taken automatically at increments of 10 degrees as the turntable rotated through 360 degrees. The hand was then turned over and the process was repeated in order to capture the opposite surface.

Next, the skeleton was separated into its individual bones. These were then scanned individually, or in groups for the smaller bones, using the same coordinate system as the first scans but this time with the scanner fitted with a 100mm FOV dual lens-set. This enabled the fine detail of the individual bones and the joint interfaces to be captured to a resolution of 0.05mm with better than 10-15 microns accuracy.

The whole process was completed in about two hours. This included converting the raw scan data, on the fly as it was captured, to a polygon mesh using Solutionix’ ezScan software and merging the individual mesh models together. Comprising nearly 100 million polygons, the resultant mesh model was extremely high resolution but too big for most postprocessing software to handle efficiently. So curvature-based sampling of the data was performed with the ezScan software to reduce the polygon count in areas of low curvature while retaining it in high curvature areas. This reduced the polygon count to approximately 1 million (50 MB file size) while retaining all the extremely fine detail.

This final mesh model was saved as a .stl (stereo lithography) format file, which was transferred into Geomagic Wrap polygon modelling software where any hole filling and necessary editing of the model could be carried out and where the individual bones were aligned to create the final polygon model of the complete hand.

The final step was to convert this polygon model data into a 3D solid model, which was transferred into the Siemens PLM Software NX CAD environment.

Moving to physical prototype
With this first phase of the project completed, the Bristol team was then able to move to the first of the actual development phases. This first phase involves the biomimetic modelling of the skeleton model by modifying it using NX 3D solid modelling CAD software to improve the geometry of the joints between the individual bones while retaining their natural movement.

Once this phase has been completed, NX will be used to simulate movement of the fingers and joints using as the input data information that the research team is gathering on the way the different muscles and tendons in the hand work in the real world. In parallel with this phase, the positions of the necessary sensors, actuators and `tendons’ on the glove will also be determined.

The fourth phase will then involve mapping the tendon-based technology onto an actual silicon glove to create a working prototype. This phase will require a complete hand to be scanned so that a mould can be designed in order to produce the glove using the injection moulding process.

But as Dr. Alemzadeh states, "Without the ability to create an accurate 3D digital model of the skeleton of a human hand in the first place using the Solutionix Rexcan CS+, we wouldn’t have even been able to start on this project, let alone get to the physical prototype stage. We expect to reach that stage within the next 18 months to two years. I was amazed at the speed at which the scanning process was completed and in the accuracy and detail of the final digital model that was produced," he adds.

The final step, once the technology has been proven with the physical prototype, is for Dr. Alemzadeh to find a partner in the business community in order to commercialise his invention and bring it to market. At that point, he will have realised his dream of providing stroke victims with a robotic aid which not only gives them back the natural hand movement they have lost but is also relatively unobtrusive and aesthetically pleasing.

UK-based Neil McLeod writes about the ways in which the latest 3D scanning and digital modeling technologies benefit the wide variety of industries in which they are used increasingly today.

A 1.844Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE