Smart app assesses wounds in seconds

18 April 2019

With the rise in living standards across much of the world, the burden of human diseases has shifted from communicable illnesses to non-infectious health issues, including chronic wounds associated with diabetes and ageing.

According to the World Health Organisation, the number of people with diabetes quadrupled from 108 million in 1980 to 422 million in 2014. Millions of people are now living with the complications of this disease. These range from blindness and kidney failure to chronic wounds.

Dr Ngai Man Cheung (Croucher Fellowship 2009), Associate Professor at the Singapore University of Technology and Design, has developed KroniKare, a smartphone app that provides an automatic assessment of chronic wounds in diabetic, elderly, and other patients, by using computer vision and artificial intelligence (AI) technologies.

Chronic wounds

Diabetes is one of the major causes of chronic wounds, especially on the feet and lower limbs. These can take months or years to heal and, in some cases, never do. They may eventually require limb amputation and are a source of extensive physical and psychological stress, severely impeding the quality of life of patients.

As the prevalence of diabetes increases globally, including in densely populated South-East Asia, the problem of chronic wounds has intensified, Cheung noted.

Current provision for wound assessment, even in developed countries, is limited and remains reliant on manual procedures. It is also expensive, being dependent on the expertise of wound care specialists.

One of the complications of chronic wound treatment is the undermining of tissue caused by erosion under the wound edge. Care providers are often unable to see what lies below the surface of the wound, particularly when it is large and has a narrow opening. Current practice involves the use of a cotton-tipped probe to pull the boundary of the wound to assess the level of undermining, measure the wound size, and note the tissues affected.

“It is a very uncomfortable experience for the patients, in addition to being time-consuming and expensive,” Cheung said. “Given that it is a fairly common complication among diabetic patients, and we face a shortage of qualified wound nurses, we need a better solution.”

KroniKare solution

Cheung and his team have developed that potential solution: an award-winning AI-driven system that automates the assessment and management of chronic wounds. The KroniKare system, filed for patent protection, is a smartphone application that captures normal and thermal images to analyse the condition of a wound.

This innovation requires the care provider to take an image of the wound – ranging from an ulcer to a wound caused by an accident. The KroniKare app then uses multi-spectral image analysis and machine-learning algorithms running on the smartphone to identify different tissue types, size of the wound, and any complications. This analysis is completed in less than 30 seconds. The more subjective manual procedure takes an average of 30 minutes.

“KroniKare is fast and accurate. It is also non-invasive compared to the current method,” said Cheung. In facilitating detection of complications it can enable early treatment and faster recovery, resulting in reduced suffering and significant cost savings. It can, for instance, be used by junior nurses for wound care management.

Cheung explained how the app works. “We use both visible and infrared light and integrate them into a single smartphone using a software we created,” he said. Unlike data-heavy AI systems, expensive computer equipment is not required. He and his team are now working on making the app available to hospitals and care providers.

While much of Cheung’s work on AI is focused on diabetic research, this is not the only area where he applies his expertise. He is also collaborating with urban scientists to generate data that can inform urban design, including traffic systems. This involves generating and correlating data on the numbers of pedestrians, vehicles, and levels of road usage to understand the capacity of urban traffic systems.

Meanwhile, his doctoral and post-doctoral studies were focused on yet another AI application, involving video coding, compression, and transmission. “In my PhD research I tried to develop compression algorithms to reduce the size of videos,” he said. With video streaming now a major source of internet traffic, reducing video size has become essential to enable bandwidth to be better utilised, particularly in wireless settings.

AI for public health

However, it is the medical and public health applications of AI that intrigue him most. AI, combined with computer vision, has huge potential for improving the early detection and diagnosis of diseases.

“One recent paper on skin cancer demonstrated that AI and computer vision can achieve the same accuracy as dermatologists in terms of diagnosis,” Cheung said. This technology could provide preliminary diagnosis, relieving professionals of more tedious, time-consuming investigative tasks, and allowing them to focus on more complex cases.

Early diagnosis was obviously important for increasing the chances for successful treatment, and reducing financial costs, he said. Furthermore, AI could revisit existing data and identify information that humans might have overlooked. It could also be used to make diagnosis more accessible to the public in settings where there was a shortage of health professionals, he explained.

Data challenges

Cheung acknowledged that there has always been an ethical debate on the use of artificial intelligence, which cannot be overlooked. Instead, open discussion and meaningful dialogue are needed to maximise its benefits.

Researchers must also face the challenge that medical data is often insufficient and far from perfect. Cheung has been trying to address this issue through fundamental research to develop algorithms that can either automatically generate data or perform with limited data.

“Despite the advances in AI, we can still improve the technology so that it functions more like a human, that is by performing with limited information and data available,” he explained.

The good news is that with the world recognising the potential of this field, there are funding opportunities for scientists such as Cheung to expand their research in both academic and applied settings.



Dr Ngai Man Cheung is an Associate Professor at Singapore University of Technology and Design. He was a post-doctoral researcher with the University of Stanford from 2009-2011. He received his PhD in Electrical Engineering from the University of Southern California in 2008. He has previously held research positions at Texas Instruments Research Centre Japan, Nokia Research Centre, IBM T. J. Watson Research Centre, HP Labs Japan, Hong Kong University of Science and Technology and Mitsubishi Electric Research Labs. He was awarded a Croucher Fellowship in 2008. 


To view Dr Cheung’s profile, please click here.