Science fiction-like applications for medical imaging where doctors can “see” inside a body in 3D and create personalised treatments from what they find are very close.
Already, doctors can get cross-section images to dig deeper into the body and uncover cancers which had previously been hard to detect.
The next step will be to create scans which reveal specific cancers only, making it easier to create personalised treatment for individuals rather than just the disease.
As a student 30 years ago, Patrick Brennan, now an expert in imaging and a professor at the University of Sydney, was told that medical imaging was proliferating at an unprecedented rate.
And imaging has changed dramatically, the biggest being the switch from bulky film x-rays to digital images.
With film, there was a major storage and retrieval issue. Sometimes it would take days to find and get a patient’s latest x-ray.
“The whole explosion of digitalisation means that if you get x-rayed now in RPA [Royal Prince Alfred Hospital] your image is there on the screen and we can very rapidly compare it with your previous image to see if it’s changed – that’s critically important,” says Professor Brennan.
“We can do things you can do with any camera nowadays: we can enhance it, we can compress it, we can change the grey levels.
“All we are dealing with are numbers and we can do what we like with those numbers to make the image look better. Caution must be applied there because we can make the image look better but be hiding something important.”
The other big advantage to digital images is that they can be sent anywhere in he world. Interpreting what the x-ray shows has always been difficult, an acquired skill, but getting access to experts is now easier.
With the arrival of cloud storage, Brennan can easily send an image across the world for a second opinion. “Suddenly with digitisation we have the ability to get the experts wherever they are located and I think that’s hugely important,” he says.
In Australia, there is Breast Screen Reader Assessment Strategy (BREAST), essentially a database which can be from anywhere, bringing in specialist doctors when needed.
Scanning in 3D
But the big advance is coming with 3D images, looking through layers of the body, pushes aside obstructions to “see” whether a mark on the screen is a cancer or just another shadow from healthy tissue.
“In my own area of specialty which is breast imaging there’s a brand new technology called digital breast tomosynthesis (DBT),” he says.
“Truly this new technique is going to be the future, there’s no question about it. It’s going to transform imaging both in this country and in America and Europe and anywhere else that can afford it.”
With a standard mammogram where the breast is x-rayed and the machine doesn’t move. In digital breast tomosynthesis the machine moves across the breast, enabling almost three dimensional images to be produced.
“If we x-ray a breast and I see a white blob I am thinking: ‘Is that cancer or is it normal tissue?’ It may be white because of a whole load of normal tissue superimposed on itself. I can cut through and see if that’s cancer or not.
“The big advantage of this, however, is that it can be used in current situations with not much more cost. It’s not much more expensive than normal mammography. The rooms can be the same. You just replace the old technology with this new technology for not a lot.”
The technology was developed in the US and brought to Australia three years ago, with the first clinical trials at the Sydney Breast Clinic.
Our eyes have flaws
This technology is exploiting the best parts of the human visual system. Imaging is limited by the range and ability of human eyes which are generally pretty poor at spotting differences.
“If you are looking at a still image, your eyes aren’t that great,” Brennan says.
“However, if there is movement — because we have been have developed natural skills to detect movements because we want to see a prey or a predator — we are really good at spotting anything that moves.
“And when you have slices you are scrolling through in a dynamic way and suddenly the cancer appears. You can see that change from a dynamic perspective rather than just looking at one image and hoping to find a dodgy area.”
Seduced by a sparkling machine
However, Brennan notes some challenges about being seduced by the bright lights of new machines.
“The technology is tremendous but we’ve just got to make sure it’s not the tail wagging the dog,” Brennan says.
“I think one has to temper enthusiasm about technology with a certain caution. Often the manufacturers will come in with fantastic machines which produce beautiful images but the question is: They may be beautiful but do they do any good?”
Machines are often proved within western communities but then start being used in other parts of the world.
“We seem to get lots of data on the US and Europe and Australia that these things work on westernised women but suddenly you see the same machines appearing in different parts of the world without the evidence to support it,” he says.
“For example, the SE Asian and Chinese breast is quite different to the westernised breast. They are more dense and smaller but suddenly they have a technology which was based on an entirely different population.”
The use of cloud as storage and the ease of moving big chunks of data around the world via the internet is creating unique challenges.
Some imaging groups get someone in another country to do the analysis.
“This is open to abuse if it’s not monitored and managed well,” Brennan says. “The images could be going to places where the skills are perhaps not as good as in Australia. They get paid to do the analysis and it’s cheaper than getting it done in a western country. In Australia, we have high standards for what a radiologist should and shouldn’t be and what sort of experience they should and shouldn’t have. How do we know what accreditation colleagues in other places have?”
The promise of technology
This is where collaboration between technologist and medical experts becomes crucial. With input from experts, technologists are training machines to “read” huge amounts of images at very high speeds. In the same way that Google has trained machines to read and organise the content of the internet, machines will eventually be able to analyse medical images to aid with diagnosis by discarding images which present no discernible problems and highlighting those which require attention. Over time, as the technology is refined, this should deliver increased efficiency, freeing more time for doctors to spend with patients.
Not just holiday snaps
Another issue is security. While being able to store in the cloud, instead of a room full of large negatives, is a major benefit, who gets access to those images is important.
“The problem with these new technologies is that we get far more images and because we are taking slices we’re not just taking one images, we’re taking 300,” he says.
“It’s not just about storing a few holiday snaps, we’ve got to make sure you medical information is utterly secure. I am sure there are people out there who have solutions for that.
“I haven’t come across it personally but there is no question that it (tampering with images) can be done. Experimentally we have shown that you can put lesions on chest x-rays. It’s not difficult. I would be very surprised if it’s never been done.”
The next big thing
Imaging is heading toward being more specific to the type of problem being scanned, to reveal the type of cancers being hunted.
“If we could target much more specifically the area we want to look at, looking for molecular biomarkers of the disease so only the disease presents then we are going to be much more able to diagnose, much more efficient,” Brennan says.
“If you look at breast cancer or lung cancer there are a variety of different types. You can see the cancer but it’s there with a whole load of noise. It’s more difficult to know what’s what.
“But with individualised diagnosis and treatment we are going to be much more effective at spotting things.”
Business Insider Emails & Alerts
Site highlights each day to your inbox.