(November 30, 2017) - It's raining in Belgium, which means it must be the end of November (or a day that end in 'y'). This past month, we have made progress some pretty decent progress and I am eager to tell you all about it. So, let's begin!
Ethics & Intellectual Property: "The number of people whose permission I need before I can do whatever I want..." is apparently quite high. I say that as we begin our project's ethics review. We have submitted our project's ethics application and it is now making its way through the multiple stages of review. This process includes a within-department peer review, and within-institution ethics review, and a review by an nationally-accredited Medical-Ethical Review Board. The project is also under review by a Data Protection Officer to make sure that participants' personal data is properly managed. All told, it is a rather thorough and time-consuming process, but it exists of course for very good reasons. We have - and will continue - to do our due diligence to make sure that this project maintains high ethical standards.
While the ethics review is ongoing, so too are discussions on intellectual property management. I'm not sure how much I can say here other than that the process is underway and that we expect no major roadblocks towards an agreement between the project partners within the new year.
Research Developments: As promised, we began work this month on the actual CAD WALK algorithms and software. The main idea behind CAD WALK is to accurately estimate a healthy person's foot pressure measurement, then compare that estimate to the foot pressure measurement taken from a patient. If the healthy estimate is good enough, then any difference between these two foot pressure measurements would be related to the foot complaint that the patient has. The key part of that idea is of course the "if the estimate is good enough" part. It is on this estimation that we have begun working.
Above, you can see our first result. At this early stage, we are not looking at the full foot pressure video, but instead have averaged all video frames. As a result, the real measurement taken from out "patient" is the leftmost image (and not a video). Our estimated healthy measurement is the middle image, and the error range in that estimate is shown in the rightmost image.
There are a few things to take away from a result like this. First, the size and shape of the estimated foot look ...okay. Our estimated measurement is a little on the large side, but not ridiculously so. Also, the height of the patient's foot arch is a little more than what we predicted, but again, not ridiculously so. Finally, the regions of highest foot pressure also seem to match up, even though the exact pressure values are a little different. All these results put together suggests that our foot pressure estimation step is at least "in the ballpark" of being accurate.
Unfortunately, the "ballpark" in this case is really very big. This is what the rightmost Prediction Error image is saying. The range of potential errors in our healthy foot pressure estimate is really quite high. This is for pretty obvious reasons: we are estimating a foot pressure measurement using very few examples from healthy individuals. When we obtain our ethics approval, we can collect more foot pressure measurements from healthy individuals, and that should help us narrow down the amount of uncertainty in our estimates.
Upcoming Events: In the coming months, we will begin to present CAD WALK in person to research audiences. We have two events already on our schedule. First, CAD WALK will take part in the closing event for the imec FootWork project on December 4th in Antwerp, Belgium. The FootWork project is the precursor to our CAD WALK project, and both projects include many of the same researchers. Unlike CAD WALK however, the FootWork project is interested in the design of custom shoes and insoles using automated techniques.
Second, CAD WALK will also join rs scan to discuss the technical aspects of foot pressure video analysis in front of the InForMed Consortium on January 23rd in Ghent, Belgium. The InForMed consortium is a large-scale EU research project involving 39 partners across 10 countries. Their goal is to develop an environment and network in which medical devices can be more easily seeded, developed, and produced.
We are always keen to discuss our research with interested parties. If you are one of them, feel free to contact us!
Up Next in December, 2017...
