Millions of images of cancer
No cancers are alike and different people need different treatment. Studying images of how cancer cells respond to different substances increases our knowledge of how to combat the glioblastoma brain tumour.
‘Different patients respond differently to different treatments and the treatments in themselves often cause a great deal of suffering. By testing different pharmaceuticals on cultured cells, we hope to better understand the differences between different patients,’ says Carolina Wählby.
She is a researcher in image analysis and collaborates with cancer researchers, pharmaceutical researchers, cell biologists and biostatisticians in a newly started project. Together they have examined the glioblastoma brain tumour, which makes up three per cent of all cancers and is a very heterogeneous cancer type.
“We use image analysis to quickly measure how the cells react to a broad range of treatments. For each patient, we test approximately 2500 different pharmaceuticals and doses in parallel and work with about a million images, something that requires a plethora of computing power.´
Researchers use biopsies of tumours and cultivate the cells in a 384-well microarray plate. There are robotic systems that can discharge small molecules, either known pharmaceuticals or potential pharmaceuticals, in the small wells. They are then photographed with an automated microscope.
All these images are then run on the computers at Uppmax.
‘If someone should analyse the images by hand, it would take several lifetimes, while a computer cluster can do it in a few hours. And you can always go back and look at the images where the computer indicates that something interesting has happened,’ says Carolina Wählby.
In addition to the images, the researchers have access to patient information, and how they have responded to different treatments. This, together with the genetic and molecular analysis, is coupled to how the cultured cells react to different pharmaceutical substances.
‘Two people never have exactly the same cancer, but you can still group the variants. The hope is to be able to better understand the different cancer variants, but in the long term, it would be fantastic if we can use this to find new effective treatments and decide which treatment would be most suitable for each patient,’ says Carolina Wählby.
‘Often it is the treatment that is so incredibly stressful, if you can choose the right treatment from the outset you save both time and suffering.’
To process all the material requires mathematical algorithms that can find the cells, measure items and group the cells according to the different properties. The researchers can also pick out some cells that they think shown an interesting change and ask the computer to find similar cells. Are there additional patients that follow the same pattern?
‘It's not as if we only use computer analysis, we constantly try to incorporate knowledge from medical students or cell biologists and utilise this knowledge, so that we maximise the knowledge of all those involved,’ says Carolina Wählby.
There are several advantages in studying the variation between patients on the cultured cells. Partly it is easier from an ethical aspect to test pharmaceuticals on cellular level outside the human body. In part you can test a large number of substances on the same patient, which otherwise would have been an impossibility.
‘But you have to remember that these cultured cells are a model of what is happening inside the body and that the cells are in a completely different environment than the human body. You must always bear in mind that this is a vast simplification of what is really happening, but it can put us on the right track.’
In the background you can see fluorescence microscopic images of cultured neurons in which different proteins light up in different colours with fluorescent labelled antibodies. The same types of images are taken of the cultured cancer cells.