Stars and hearts: the increasing collaboration between astronomers and doctors
In the cramped, noisy environment of an MRI scanner, alien worlds orbiting distant stars are probably the furthest thing from the patient’s mind. And yet techniques developed by astronomers to hunt out other planets are now being used by doctors to take more accurate pictures of brains and bones. In the past, the image often blurred as the patient moved inside the machine, requiring retakes that were expensive for the hospital and extending the time the patient spent within. Estimates suggested rescanning was costing the NHS around £30m a year. Now, software originally developed at the Royal Observatory in Edinburgh can detect movement and correct for it, allowing images to be taken first time.
Such collaboration between those who study space and those who diagnose disease is on the increase. In February, it was revealed that a technique originally used by astronomers to pick out faint, distant galaxies has been adapted to study tumour samples in breast cancer patients. Doctors have traditionally looked down a microscope to search for subtle differences in tumour cells caused by the way they release proteins. This technique—known as immunohistochemistry—allows clinicians to assess how aggressive the cancer is, but it is a laborious and time-draining process. Astronomers have long faced the similarly painstaking task of making out distinct objects in large images of the night sky and so have already developed methods to do it automatically using computer software. Researchers at the Institute of Astronomy at the University of Cambridge shared this technique with their colleagues in the Department of Oncology.
Working alongside the Cancer Research UK Cambridge Institute, they put their astronomical trickery to the test against the conventional microscope method on tumour samples from over 2,000 breast cancer patients. Not only did the astronomy-inspired technology match the accuracy of the human technique, it did so much faster. The venture was such a success that plans are under way to extend the test to 20,000 samples in the hope of refining the technique even more, with the ultimate goal of outstripping the human eye in terms of accuracy as well as time.
Similar knowledge sharing holds great promise in the field of heart disease. In order to investigate the state of a patient’s blood vessels, doctors use CT scans to build up a 3D model of the heart. However, to see every part of the heart, the image needs to be rotated or animated which takes time. Astronomers face a similar problem when they are trying to work out where newly forming stars are nested in giant gas clouds called nebulae. To solve their problem, astronomers turn their 3D map into a 2D tree diagram projection. Harvard researcher Michelle Borkin hit on the idea of doing the same thing in cardiology. She was able to turn a 3D CT scan model of the heart into a 2D projection of its arteries and blood flow patterns. Using that technique, she found doctors and students became more accurate and efficient at heart disease diagnosis. She saw a jump in accuracy from a 39% success rate with the old 3D model to an initial 62% using the 2D projection. She managed to jack this up to 91% just by playing with the colour of the 2D diagrams.
This exchange of knowledge is not a one way street, however. Astronomers in the US have exploited the methods used to build up a three dimensional image of the human brain to construct maps of star-forming regions, supernova explosions and even our own Milky Way galaxy. A radio telescope images a region of the cosmos in “slices” in a similar way to how an MRI scanner takes image slices of the brain to reveal its structure. Using software originally developed in hospitals, astronomers can now analyse these slices in minutes rather than hours.
Whether it is astronomers gazing at the universe with giant radio telescopes, or a medical imager diligently building up a picture of events beneath our skin in a hospital, their increasingly pooled know-how is helping to save lives and understand the cosmos. Next time you’re unlucky enough to visit a hospital, the universe out there will perhaps be closer than you realise.
- H.R. Ali, et al, Astronomical algorithms for automated analysis of tissue protein expression in breast cancer, Br. J. Cancer 108, pp. 602–612, 2013.
- M. Borkin, et al, Evaluation of artery visualizations for heart disease diagnosis, IEEE T. Vis. Comput. Gr. 17, 12", pages: "2479, 2001.
- M. Borkin, et al, Visualization and analysis of synthetic observations of star forming regions, Proc. APS March Meeting 2012, 2012.
- A. Goodman, et al, The bones of the Milky Way, Bull. Am. Astro. Soc., AAS Meeting 221 234.01, 2013.