Healthtech: Facebook developing AI to make MRI scans 10 times faster

Evidence that Facebook has ambitions beyond its core business of social networking emerged this week, with technology development that stands to benefit society as a whole.

Facebook has collaborated with NYU School of Medicine’s Department of Radiology to launch fastMRI, a new research project that will investigate the use of artificial intelligence (AI) to make magnetic resonance imaging (MRI) scans up to 10 times faster.

If the project is successful, it will make MRI technology available to more people, expanding access to this key diagnostic tool, according to an announcement on Code, Facebook’s development blog.

MRI scanners provide doctors and patients with images that typically show a greater level of detail in soft tissues — such as organs and blood vessels — than is captured by other forms of medical imaging. But they are relatively slow, taking anywhere from 15 minutes to over an hour, compared with less than a second or up to a minute, respectively, for X-ray and CT scans.

The need for speed

These long scan times can make MRI challenging for many patients, such as young children, elderly patients, or people with physical disabilities or limited movement. Additionally, there are MRI shortages in many areas and in countries with limited access to the technology, resulting in long backlogs of patients.

The longer the wait, the more lives could be lost. So by boosting the speed of MRI scanners, the devices can be made more accessible to a more patients, said the announcement.

“Sufficiently accelerated MRI devices could also reduce the amount of time patients must hold their breath during imaging of the heart, liver, or other organs in the abdomen and torso,” it continues.

“Increased speed could let MRI machines fill the role of X-ray and CT machines for some applications, allowing patients to avoid the ionising radiation associated with those scans.”

A new approach

The new project will initially focus on changing how MRI machines operate. Currently, scanners work by gathering raw numerical data in a series of sequential views and turning the data into cross-sectional images, which doctors then use to evaluate a patient’s health. The larger the data set gathered, the longer the scan takes.

Using AI, it may be possible to capture less data and therefore scan faster, while preserving or even enhancing the rich information content of magnetic resonance images, says Facebook.

The key will be to train artificial neural networks to recognise the underlying structure of the images in order to fill in detail omitted from an accelerated scan.

The risk of innovation

The team has already made good progress. Early work performed at the NYU School of Medicine shows that artificial neural networks can generate high-quality images from far less data than was previously thought to be necessary.

The research is not without controversy, however. “In practice, reconstructing images from partial information poses an exceedingly hard problem,” admits Facebook. “Neural networks must be able to effectively bridge the gaps in scanning data without sacrificing accuracy.

“A few missing or incorrectly modelled pixels could mean the difference between an all-clear scan and one in which radiologists find a torn ligament or a possible tumour. Conversely, capturing previously inaccessible information in an image can quite literally save lives.”

Facebook says that, unlike other AI-related projects that use medical images as a starting point and then attempt to derive anatomical or diagnostic information from them, this collaboration focuses on applying the strengths of machine learning to reconstruct the most high-value images in new ways.

“Our aim is not simply enhanced data mining with AI, but rather the generation of fundamentally new capabilities for medical visualisation to benefit human health,” says Facebook.

Internet of Business says

The team plans to open-source the work to allow the wider research community to build on its developments. As the project progresses, Facebook will share the AI models, baselines, and evaluation metrics, and NYU School of Medicine will open-source the image data set.

The work is yet more evidence of how AI is already transforming diagnostic medicine. AI’s ability to find hidden patterns in data that may indicate a medical problem – or signal a propensity to develop one – will be invaluable, and not only improve patient care, but also help people to avoid some diseases or conditions.

Meanwhile, innovations in AI that are essential in other fields – such as autonomous transport – could one day be miniaturised in order to benefit healthcare.

One example is the AI-powered camera/optical computer being developed at Stanford University in the US. Although intended for driverless cars and autonomous drones, miniaturised versions could be incorporated into a new generation of intelligent handheld medical scanners.

AI-enabled wearable devices have already proven themselves in the field of medical diagnosis, as numerous Internet of Business reports have revealed:-