The last time astronomy and medicine were combined was a millennium or two ago in the Middle East when Persian doctors used astrology to diagnose and treat diseases (to uncertain effect, of course…). Now, however, these seemingly most disparate of disciplines have reunited under the image analysis.
Big data, small cells
The truth is that both astronomy and cytology (the science of cells) involve huge numbers, even though the average galaxy is many orders of magnitude bigger than a biopsy slide.
For decades cancer has been diagnosed in patients with the use of a small sample of cells, a microscope and human eyes. This has served us well for a long time, but there are many drawbacks – examining a slide of cells is time-consuming and also involves judgement and even intuition. Many times, diagnoses differ between different operatives, leading to second opinions, while a worried patient waits for the results.
Thankfully, big data has migrated from astronomy and economics to healthcare, with the same principles that astronomers use to examine and analyse galaxies transferring to a smaller, but no less data-rich, scale.
It makes perfect sense
Think about it, the huge optical and infra-red telescopes we use to explore the universe gather billions of pixels every day. These images have to be analysed by software developed by companies like bitplane.com. The heat signature of the telescope, background light and cosmic ray interference all has to be adjusted for and then each star has its location and brightness tagged.
The image analysis software uses algorithms to identify various astronomical patterns and features, as well as classify them. Astronomers can look at the way light is bent by gravity, as well as the way stars seem to be distributed around dark matter in streams.
The same algorithms can help pathologists
Pathologists and oncologist can use the same algorithms at the opposite end of the scale – looking down a microscope. The average biopsy for breast cancer, for example, is less than a millimetre in diameter, but nevertheless, each cell and its nucleus needs to have its own measurements and co-ordinates mapped.
Image analysis software used by astronomers is “taught” to recognise three sorts of cell – immune cells, stromal cells and cancer cells and the information gleaned from the analysis can help pathologists to spot cancers, even in very early stages. Cell slices can be stained to highlight specific proteins involved in the development of cancers, identifying the presence, type and even the aggressiveness before it even properly begins.
Data for the future
After diagnosis, cell samples are often destroyed, taking with them valuable information. However, with digital storage, researchers can pull up and revisit images of any one of the 15,000-plus types of tumour that occur to find out more about them as research uncovers more or new questions need to be asked.
One of the problems with oncology is that not everyone has ready access to it at present. It won’t be long until prepared biopsy samples are photographed and relayed from remote areas of the world to expert oncologists and their algorithms for diagnosis. As we all know, time is of the essence with cancer, so this cosmic tech could improve survival rates here on earth.