X marks the spot - minimising X-ray exposure using video gaming technology

As the popularity and prevalence of home computers grew in the early 1980s, with offerings from Acorn, Apple and Commodore to name but a few, so too did the possibilities for the booming world video gaming. By the end of the decade, cassette and cartridges were beginning to give way to CD-ROMs: an early taste of the digital future to come.

A similar revolution would take a little longer to arrive in X-ray imaging, with radiologists at that time producing images using film, rather than the computed and direct radiography of today. Dr Steven Don, associate professor of paediatric radiology at the Washington University School of Medicine in St Louis, was a resident in the 1980s, and recalls the importance of perfecting X-ray technique by measuring body part thickness.

“Film, while a very good medium, was not forgiving if the X-ray technique wasn’t right,” he says. “In underexposed radiographs, the image would appear light or white, and you couldn’t see what you wanted, and if it was overexposed, it was too dark and you couldn’t interpret it. So the technologists had to be pretty close to being spot-on.”

Today, it’s a rather different picture. Sophisticated digital methods enable clear images to be achieved within a much broader range of exposures, because the results are reprocessed by computer. In some ways, this has led to speed being prioritised over accuracy of technique, and traditional metal callipers – which need to be cleaned between each use, can only measure one spot at a time, and can frighten children – are largely unused today.

Rather than measuring body thickness, paediatric technologists tend to use a patient’s age to decide the required dose. In adults, a guide of small, medium and large habitus (body type) is applied.

Though speed is an advantage in improving the vital throughput of a department, Don – who has a clinical practice at St Louis Children’s Hospital – emphasises that using age is a problematic approach.

“We know that a skinny 18-year old may be the same thickness as an obese four-year old,” he says. “So by using age, the 18-year-old may be given too much of a radiation dose, and an obese four-year old may be given too little, and neither image looks great.”

What’s more, a gradual increase in radiation dose can result from this less accurate technique. While an overexposed image may go unnoticed by a radiologist because the image is still clear, grainy, underexposed scans are likely to be rejected.

“Radiologists don’t like noisy images, and we tend to complain, and ask that the technologists repeat them,” Don says. “Over time, there’s something called ‘dose creep’, where the technologists know that if they underexpose they may need to repeat the radiography, so they tend to tweak the dial up just a little bit, and the dose keeps increasing.”

This means that, whether through overexposure or a repeat X-ray following an underexposed image, many patients are receiving higher levels of radiation than strictly necessary over time. Clearly, there are advantages to be found in going back to basics to measure body thickness, and using this to inform X-ray technique. Time-consuming metal callipers, however, are an impractical option.

Setting out to address these challenges, Don and collaborators Mr Robert MacDougall, clinical medical physicist at Boston Children’s Hospital, and Dr Benoit Scherrer, instructor in radiology at Harvard Medical School and research fellow at Boston Children’s Hospital, worked together to design a novel way of measuring body part thickness quickly and effectively.

Gently does it

Don and MacDougall met around 2010 while contributing to the ‘Image Gently’ radiation dose reduction initiative in the lead up to the 2012 launch of its ‘Back to Basics’ campaign for managing radiation exposure in children, co-chaired by Don.

With the need to strengthen imaging technique a source of inspiration, the technology at the centre of the team’s system was in fact plucked from an altogether different context: the world of video gaming.

Nintendo launched its popular, motion-sensing Wii console in 2006, and by 2010 companies like Microsoft and Sony were releasing their own next-generation machines. The former produced the Xbox Kinect, which allowed gamers to play without a controller by fusing an ‘RGB’ video camera that could detect people standing in front of it with infrared and CMOS depth sensors able to interpret rooms in 3D.

The Kinect was released just as Don and MacDougall were discussing the possibility of introducing motion and depth-sensing into the X-ray process. The team mounted the Microsoft hardware onto the X-ray tube itself (in their latest model, the newer Kinect 2.0 is used) in order to exploit its sensory capabilities.

Helpfully for the team, the device came with a software development kit that enabled users to capitalise on its function. Dr Scherrer, who has a background in software engineering, came on board to lay out the specification, design and implementation of the software and enable it to interpret the X-ray environment as required.

The result is ‘X-Vue’, a system that uses the data feed from the cameras and sensors to provide technologists with an unprecedented level of real-time information on the patient in the X-ray room, all displayed on a computer screen in the booth.

Starting with its fast measurement of patient body thickness, X-Vue enables technologists to optimise their technique for examinations, using appropriate dosage to a much greater degree of accuracy, and therefore avoiding dose creep. And while callipers can measure thickness with around 1cm precision, Kinect imaging achieves accuracy up to 1mm, without any physical contact.

Motion-detection, meanwhile, can help prevent repeat radiographs by highlighting when a patient has deviated from the required position before an image is taken. It’s a significant improvement on standard procedure: “When the technologist brings you into the room to take an X-ray, they position you in front of the X-ray detector and then walk out of the room,” Don explains. “They’ve turned their back on you, and if you’ve got pneumonia, say, and are coughing, you may have moved from the spot where you were centred. Then they’re viewing you at an angle [from the booth].”

By providing technologists with a screen view taken straight down the line from the X-ray tube, they are already in a better position to note movement, but the X-Vue will flag on screen any departure from the required position. This allows the patient to be repositioned before an image is taken, avoiding repetitions.

The system can also highlight the body part that is meant to be being X-rayed according to the doctor’s instruction, and notify the technologist when this is not centred.

“If the ordered body part, let’s say the left wrist, is not centred on the detector, or on a specific part of the detector – depending on the examination and facility preference – it will show up as red,” says Don. “This will reduce the number of wrong site images and help identify incorrect orders, such as accidently ordering right instead of left side.

Errors on the part of the technologist in centring the wrong side, or on the part of the doctor in mistakenly ordering the wrong side, can then be cleared up prior to the X-ray taking place. The patient is prevented from undergoing unnecessary radiation, and the efficiency of a much-needed department is improved.

In tightening up technique, X-Vue could benefit the gamut of patients, and the researchers from Boston and St Louis are keen to see it applied widely within adult and paediatric radiography. However, it is children that they see as the biggest beneficiaries.

“Children are the most sensitive to radiation, so we think it’s ideal for them,” Don says. “In addition, they are growing rapidly; a four-year-old may be skinny or obese and we have to be able to change our X-ray technique based on that.

“Also, younger children especially can’t necessarily hold still, because they’re in pain, or they’re scared, so the technologist needs that additional information to make sure the patient hasn’t moved, or isn’t moving, before the X-ray.”

While the team is continuing its research within plain X-rays – currently partnering with a children’s hospital to receive feedback on how well the device functions in the clinic – it also has fluoroscopy and CT in its sights.

“In fluoroscopy, maybe we can help with setting up X-ray equipment.” Don says. “Knowing the thickness of the body part that you’re going to image means you can automatically adjust your technique to get an optimum exposure for the type of study that you’re doing. Or maybe in interventional fluoroscopy you’ll be able to measure the dose of the organs that you’re delivering in a skin dose.”

In CT, as in X-ray, body thickness can be used to optimise technique, and the ability to see motion up against the scanner could again be hugely helpful in ensuring that patients are not moving prior to taking the image.

Game on

With funding in hand from Washington University and the Society for Paediatric Radiology, Don, MacDougall and Scherrer are now seeking further funding at the federal level to continue the development of the project. The trio have also established a company, Vantage Medical Imaging, with a view to marketing the device in the longer term.

What’s clear, though, is that the realisation of this system would not have been possible without the passive contribution of the gaming industry in making a sophisticated set-up accessible (an Xbox One with Kinect cost around £300 at the time of writing).

Of course, it’s not the first time that technology designed for gaming has benefitted medical technology. In 2008, researchers at University College London and King’s College London attempting to speed up MRI reconstruction began testing the use of graphics cards as an alternative – at a fraction of the price – to a computing-cluster they had previously trialled. And in September 2015, scientists at California’s University of Irvine completed a proof-of-concept trial in which a paraplegic man used a brain-computer interface (BCI) to control his own legs in walking. Training with an avatar in a virtual reality environment was central to the man’s ability to master the device.

Meanwhile, numerous applications of gamification strategy to healthcare and medical training show that it’s an integration that is set to continue.

For Don (who confesses he isn’t a big gamer), it’s a major facilitator, and one that has allowed nearly 30 years of research and clinical practice within digital imaging to culminate in the X-Vue.

“This mass market of gaming really helps to defer a lot of the development costs for us, and we couldn’t do it cost-effectively if we didn’t have an inexpensive Kinect device with the software development kit that Microsoft developed,” he states. “It would just be prohibitively expensive for us as a clinical team.”