The room I’m standing in is unlike any I’ve been in before.
With 4 large desks facing me, and a panoramic view of what looks like a distant universe out the window, this feels more Star Wars than cancer lab.
I place a card labelled ‘sample 1’ onto a desk that’s linked to a projector. A cloud of different shaped bubbles appear in front of me.
I’m told that the bubbles are in fact different types of cell. And this cloud is the complete picture of a single breast tumour.
But the room I’m in, and everything in it, doesn’t actually exist.
I’m in a virtual lab, projected through a headset. And I’m experiencing the future of how scientists and doctors might one day view, study and understand tumours.
This world is the brainchild of Professor Greg Hannon, from our Cambridge Institute, who’s leading a team of international scientists looking to change the way we see cancer.
Hannon’s team has recently been awarded £20 million through our Grand Challenge. And while their virtual reality vision is not yet fully operational, the demo is an impressive centrepiece to an ambitious research plan that draws in cell biology, astronomy, maths and genetics.
“This does sound a bit like science fiction,” says Hannon. “We want to develop an entirely new way to look at cancer.”
The project builds on the work of the Cambridge Institute’s Professor Carlos Caldas, which 5 years ago redefined breast cancer as 10 distinct subtypes of tumour.
Research has shown that each of these subtypes could help predict how a patient may fare. But the classifications can’t, as yet, offer definitive answers for each patient.
“The importance of this project is to be able to take those broad classifications and make more precise predictions,” says Hannon. “So it’s not just ‘you have a type of tumour where we think there is a 50% chance that this treatment will really help you’, but to know which 50% each person falls in to.”
One of the challenges scientists have faced in refining these predications is that the tissue samples (biopsies) they have access to offer just a single snapshot of a tumour.
Biopsies play a crucial part in a patient’s journey through cancer. They are required to diagnose the disease and make decisions around treatment. And researchers also use this precious tissue to learn more about tumours.
Analysing these samples to search for faulty genes and molecules has been incredibly useful in guiding treatment, particularly for breast cancer. But the techniques that do this only take an average of all the cells, molecules and DNA a sample may hold, meaning some details may be missed. This would be like deciding to give everyone at a dinner party steak, just because the majority of guests are meat-eaters. And, according to Hannon, offers “an imprecise picture of what tumours really look like”.
In recent years, it has become clear that tumours differ greatly between patients. These differences can also appear within each patient’s tumour, raising further challenges for treatment. As this idea has taken hold, the techniques used to characterise tumours in the lab have become focused on each individual cancer cell, rather than information averaged from many.
Hannon says this has been an improvement, because it gives much more detailed information. But he believes this needs to be taken a step further, drawing in the geography of tumours to build a more complete picture and refine predications around treatment.
Tumours aren’t made of just one type of cell, says Hannon. So more needs to be done to find out which cells are inside tumours, where they are, and who their neighbours are. “Tumour cells within a cancer have different properties, different capabilities,” he explains. And the team want to define these differences for every cancer cell in a tumour, and study how this affects the healthy cells that can be corrupted and controlled by tumours, such as immune cells.
Building the atlas
“The goal of our project is to create an interactive 3D atlas of cancers where we know every cell, we know what kind of cell it is, and we know its general features,” says Hannon.
This is an ambitious goal, but not out of reach.
“We think about this almost like saying ‘OK, we’re going to put a man on Mars’.” What they’re trying to do doesn’t violate any particular laws of physics, he adds. This means that the technology exists, but they’ll have to push the boundaries of what can be done with it right now to get there.
As with any journey of exploration, the team needs a map. And they’ll start with a specialised microscope to take a picture of every cell in the tumours they’ll initially be studying in mice, and then in samples from patients.
“The kinds of microscopes that do this, you can’t buy you have to build them,” says Hannon. “You can’t just put an order in. So we have found someone who’s a real expert in this and they’ll come to Cambridge and construct this thing.”
This microscope is unique because it’s attached to what is essentially a tiny bacon slicer. As the microscope captures a picture of a minute layer of cells in a tumour sample, that same layer is sliced off and collected for further analysis.
Once this has been carried out for the entire tumour, the image of each single cell has to be put back together as a complete 3D picture, which means processing a lot of data.
“When I say a lot, I talk about 100 terabytes per sample,” says Dr Dario Bressan, who’s part of Hannon’s team leading on the microscopy arm of the project.
That’s more than twice the amount of data collected by the Hubble Space Telescope in its first 20 years of observation, for each sample the team collects. So, with the hope of analysing thousands of samples during the project, it’s no surprise that they’ve turned to the astronomy department at the University of Cambridge to help them process these data.
“When we told the astronomy guys you have to file through 100 terabytes of data they said: ‘oh, only?’ They’re used to this – they’re topic of discussion is the sky – so they’re adapting their algorithms from finding stars to finding cells,” says Bressan.
Once the 3D image has been complied, the team will then have to overlay data from a suite of techniques that measure the genes and molecules inside each individual cell. This is made possible by applying the techniques to the thin slices that were originally taken from the microscope, and then mapping this to the astronomy team’s 3D image.
At the moment, when used in pathology labs, these techniques can measure the levels of a couple of protein molecules and the activity of a couple of genes. “Our edge here is that we’re collaborating with people who have invented ways to extend this to 1000 or more genes and 53 proteins,” says Bressan.
“Ultimately we want to be able to collect 20,000 plus pieces of information on every cell in a tumour,” adds Hannon. Again, that’s a lot of information.
“We will have to invent an entirely new way for people to interact with this information,” says Hannon.
And that’s where the virtual reality comes in.
A new perspective
The challenge of how to present and communicate these data is one the team has thought about a lot.
“The only way to put more information on a piece of paper is to write smaller,” says Hannon. “And at some point, there’s diminishing returns in this.
“In virtual reality we can present many more dimensions of information than you can on a piece of paper.”
To tackle this the team has drawn its inspiration from the world of video games, which Bressan says are “really good at showing a lot of information in the blink of an eye”.
The team’s idea, he adds, is that with virtual reality “we don’t just use the position of each cell, but we can use the colour of each cell, we can use whether it’s blinking or not, we can use the size. And already by doing this you are looking at 6-8 dimensions of data at the same time.”
Thanks to a new collaboration with virtual reality designer Owen Harris, artist Flaminia Grimaldi, and programmer Robby Becker, that ambition has quickly been developed in to a working demo, shown in the video below.
— Cancer Research UK (@CR_UK) May 2, 2017
“When Greg approached me, I’d been working with some Dutch scientists making a VR experience for people with anxiety and depression,” says Harris. While he thought that Hannon’s plan for a virtual cancer lab sounded amazing, Harris wasn’t originally sure he could help due to the scale of the project.
But a visit to Cambridge, and a chat with his aunt Claire, changed his mind.
“I was overwhelmed by the vision of the people in the lab,” he says. “It was exhilarating. So I was thinking I needed to do a little bit on this project.”
His aunt then encouraged him to go further.
“Claire is a very important person in my life. In my teenage years she really encouraged me at times when others weren’t.
“She has breast cancer. And she said to me: ‘you have to do this’. So it’s a mixture of her demand and how impressed with the team I was that has drawn me in deeper and deeper.”
While Harris and his team has developed an impressive early version of the virtual lab, he admits that it has been “held together with virtual duct tape”.
So now the team is working on new features, and strengthening the foundations of the new technology. “The design challenge is exciting because this is the only one of these that exists in the world,” he says.
“It’s one of the most satisfying things I’ve ever worked on.”
The future for the VR is more rooms to look at new data in different ways. These developments, says Harris, are focused on building something that works for researchers, doctors, and people looking to learn about cancer.
A new type of pathology?
Hannon’s hope is that the information they gather over the next 5 years, and the technology they build, will become part of modern pathology.
This won’t come through the thousands of measurements they have planned for millions of cells. Instead, they will have to steadily focus down on the most important information for doctors looking to make decisions around treatment.
A big part of this, he says, is that the custom, clunky instruments they’re building will need to be turned into something that “fits inside a shoebox”.
Working in virtual reality also opens up the chance to collaborate across the globe, both for the scientists involved in the early development stages and potentially for doctors in the future.
“This is not a single user interface,” says Hannon. “The thing the gaming world is really good at is having these multi-user designs. So what we envision is not only for researchers to be able to meet in these virtual reality spaces, but for physicians and patients to be looking at these models together.”
That ambition, although it remains a distant goal, is one that is shared by the 2 patient representatives that will be supporting Hannon’s team.
At a recent event where all Grand Challenge teams gathered to present their projects, the patients spoke of how they think that seeing the data, particularly in 3D, might help others understand some of the treatment choices they are presented with.
And that’s the true reality behind the sci-fi goggles, a desire to learn everything about a tumour, so that doctors, patients and the public each have the information they need to navigate their journey through cancer.