Design Engineering
Showcase 2020

Diabetic foot is a syndrome which results in painless ulceration on the sole of the foot. This happens due to two other conditions: peripheral neuropathy and peripheral arterial disease, which cause loss of sensation and reduced blood flow in the extremities respectively.

Patients with diabetes have a 25% chance of developing diabetic foot in their lifetime. These ulcerations, in turn, often lead to amputation. Some studies have shown that the five-year mortality rate of these ulcerations is as high as 80%, due to ulcer recurrence eventually leading to major or minor amputation.

Better footwear, which addresses the unique nature of diabetics’ feet, could both reduce the chance of ulceration, and increase the rate of recovery, by reducing high pressure points, which trigger the ulcers, and adapting the topology around the ulcers so they can better heal.

Bone Deformities

It is worth noting that these problems are exaggerated by bone deformities commonly found in diabetes, such as arthritis, gout and Charcot’s foot. These deformities contribute to high pressure points which, in turn, lead to developing ulcers. It is important to remember these ulcers are painless, due to the degradation of nerves in the extremities, and can develop for some time before being noticed.

Current Footwear Problems

Current clinical footwear cannot compensate for the needs of diabetic foot. Their generic nature means they can never properly address the unique nature of each patient’s bone structure and ulcer profile. Furthermore, the nature of ulcerations is, as they heal their size and shape can vary over time, meaning no one shoe design is sufficient.

Associated Costs

The problem is highlighted when you consider the costs associated with diabetic foot; currently nearly £1 billion are spent annually by the NHS alone on ulceration, amputation and recovery. By using additive manufacturing and automated mid-sole design, patients could see reduced rates of ulcerations and a better rate of recovery.

Taking the image

The first stage of the process is taking a photo of the foot. A primary challenge here was finding a way to passively remove the background, so only the foot and ulcer are considered. In the end the simplest solution was found to be the best; sliding the foot through a white sheet and then taking the photo using a scanner. The advantage of using a physical scanner is by knowing the distance between the foot and camera a size can be associated with each pixel.

K-Means Clustering

A core component of this project is flattening the image to receive the measurements. This is primarily done using K-Means clustering, an unsupervised machine learning method. This process is used in mathematics to cluster points in a dataset. By applying this to the RGB values of an image you can find specific areas in it, and through that the ulcer and foot shapes.

Originally, RGB thresholding was used but this led to algorithmic bias towards skin colour.

Post-Processing

Once the image is flattened into black and white, two csv files are made containing point coordinates of the edges of the foot and ulcer, so the data can be transferred into CAD packages.

Modular Design

Since many different profiles would be needed over time as the ulcers recover, a wider modular design would be needed. Another project was done considering this where the mid-sole could be easily replaced. Furthermore, this gave the opportunity to implement an IoT sensor. This sensor uses a force sensitive resistor and a humidity and temperature sensor to monitor the situation on the foot, primarily to monitor for high pressure areas if the mid-sole needs replacing. This was developed in the Sensing and IoT module earlier this year.

The final stage of the project was a standalone executable program which analyses all images in its directory and saves the csv point coordinate file for each image. This is then imported into a CAD package (Fusion 360 in this project) to be converted into a printable mid-sole. This program can be downloaded from the linked Github repository, together with the base code and run on any computer.

By using additive manufacturing, a patient can go into a clinic, have their foot scanned and have a new midsole printed that same day, replacing the old one, providing better recovery rates and reducing the chances of amputation.