The wounds of war: AI leads arms race for burn assessment

H

istory shows that in times of war, medical innovation emerges.

From surgery techniques, blood banks and penicillin, many of healthcare’s everyday treatments were pioneered out of necessity on the front line.

As the war in Ukraine approaches a second year, countries around the world are accelerating their defense spending to fuel the next generation of military technologies. Predictably, the US is one of the largest spenders, with a GlobalData report showing that the country’s defense budget rose by 5% in 2022 to $740.4bn, up from $705bn in 2021. The recently passed 2023 National Defense Authorization Act (NDAA) indicates that the U.S. defense budget will increase by 11.8% to $827.6 billion in 2023 and by 2027, the country’s total defense budget is forecast to reach $861.4 billion, a positive CAGR of 3.9%.

For military medicine, lessons learned from previous conflicts mean innovation in wound care and trauma management will be some of the most heavily invested areas. Indeed, researchers are already applying predictive analytics and artificial intelligence (AI) to advance surgical triaging on the battlefield.

A new approach for burn wound assessment is being led by Dallas, Texas-based company SpectralMD, who are using AI and machine learning techniques to develop an imaging tool, the DeepView system, that can accurately predict the severity of burn injuries.

One of the most critical decisions when treating burn victims is assessing how deep the burn is and whether surgery is required. The technology under development is being adapted for use in military settings and emergency rooms to quickly and clearly assess if patients need treatment from a burn specialist or non-burn specialist.

Using multi-spectral imaging, the system integrates AI algorithms and optical technology that can distinguish between healthy and damaged tissues. The imaging technology extracts clinical data, processes the image, and displays a comparison of the original image next to an image with a highlighted area of the non-healing portions of the wound.

According to SpectralMD, the image acquisition takes 0.2 seconds, and the output takes approximately 20 to 25 seconds, helping to avoid unnecessary surgeries or a reduction in military forces.

The system emerged from the technology transfer department of the University of Texas Southwestern with an initial focus of providing clinicians with a tool to predict pressure ulcers for bed-ridden patients.

From 2013 to 2021, the company engaged in contracts with the US federal agency BARDA (Biomedical Advanced Research and Development Authority) to investigate the use of the technology as a surgical-triage tool for burn victims in a mass casualty event. Overall, BARDA has invested over $100 million in SpectralMD since 2013 to improve accuracy of the technology for wound healing.

“Obviously the naked eye can only see the upper epidermis of the skin, but through the spectral images data we can go several layers deeper underneath the skin and that's what the USP is for the device,” SpectralMD CEO Wensheng Fan tells Medical Technology.

“For example, if surgeons in Ukraine used the technology, they could accurately identify which areas to operate on and reduce unnecessary surgeries. This means that patients can be helped effectively within seconds.”

The company is working with the US Department of Defense (DoD) to miniaturize the device to a fully handheld size which can be used on the battlefield, and by first responders such as police and ambulance staff.

Despite the life-threatening nature of burn management, it is an area of research that is often neglected until emergency situations arise, says Fan. “Investment in this space can be overlooked because it’s not a day-to-day thing. Often investors want to deal with more immediate issues such as diabetes, blood pressure, cancer and sometimes these unfortunate areas like burn care are only remembered about in the case of war.”

“Fortunately, it is an area where the US has led from a policy perspective since the tragic events of 9/11, and I applaud that. Our contract started under the Obama administration and other administrations have followed suit. And with the Biden administration we see bipartisan support for funding and burn care very much continues to be a national priority in America.”

The technology is still in clinical trials but in 2018 it secured a breakthrough designation from the US FDA for burn care. If SpectralMD can advance to a commercial stage, the system could be a major step forward for wound management.

According to the American Burn Association, nearly 450,000 individuals suffer burn injuries in America every year which require medical burn care. Studies show that clinicians can achieve only 70% accuracy using clinical judgment or diagnostic tests to assess non-healing burn regions, while the DeepView system has shown a 92% accuracy in studies.

Other researchers are taking a physics-based approach to burn assessment. A team from New York’s Stony Brook University are developing a new handheld tool which employs a neural network model that uses terahertz time-domain spectroscopy (THz-TDS) data.

Led by M. Hassan Arbab, from Stony Brook’s Department of Biomedical Engineering, the portable handheld spectral reflection (PHASR) scanner is designed for fast hyperspectral imaging of in vivo burn injuries using THz-TDS.

The method was tested by using the PHASR scanner to obtain spectroscopic images of skin burns and measure the permittivity of the burns. Data was then used to create a neural network model based on labeled biopsies. According to new research, the model estimated the severity of the burns with an average accuracy rate of 84.5 percent and predicted the outcome of the wound healing process at 93 percent accuracy rate.

“In 2018, approximately 416,000 patients were treated for burn injuries in emergency departments in the U.S. alone,” said Arbab. “Our research has the potential to significantly improve burn healing outcomes by guiding surgical treatment plans, which could have a major impact on reducing the length of hospital stays and number of surgical procedures for skin grafting while also improving rehabilitation after injury.”

These advances in burn assessment technologies feed into a larger healthcare trend taking place where doctors are increasingly relying on imaging tools to aid their diagnoses. Australian telehealth company Coviu is developing what it calls a ‘one-stop shop’ for physicians treating wounds. Using AI-powered mobile imaging, physicians can analyse and monitor wounds remotely by evaluating vital sign metrics, such as respiratory rate and patient heartbeat, from a video feed.

The company is working in collaboration with multiple research institutions to develop the new toolkit, which is backed by the Australian Federal Government’s Medical Research Future Fund.

Overall, clinician access to wound data will guide treatment flow whether that is on the front line or in emergency centres and chronic disease management, says Fan.

“There’s a massive waste of resources going on right now in healthcare and technologies like this can help direct patients effectively. One area we believe the DeepView technology can be helpful in is diabetic foot ulcers (DFU). Diabetes is a global problem and DFU is unfortunately very common in older people with diabetes. Currently there are no tools available to assess whether a DFU will heal or not.”

But for the clinicians on the front line of war, it hopefully charts a new course in precision medicine that can optimize care and minimize errors in the heat of battle. “The world is transitioning to precision treatment, and this will reduce both civilian and soldier injuries and keeps everybody better taken care,” says Fan. “It's a mission that all governments need to pay attention to so they can always offer the best care to their citizens.”