Could AI cure cancer?

AI can improve cancer detection

Spotting cancer is not always straightforward. Radiologists must train for years to learn how to read patient images, to understand the patterns and changes on those images that suggest cancer.

In the UK and many other countries, every mammogram (a special scan of the breast) must be analysed by two different radiologists. This takes up valuable time, and resources.

But new research suggests that with its superpowered ability to spot trends in data, AI could be a very useful tool to help radiologists detect when cancer is present.

Evidence from a study conducted by researchers in Sweden now suggests that one radiologist supported by a specialised AI tool could detect breast cancer from mammogram images as accurately as two radiologists.

If true, this could mean that AI can potentially help radiologists with their work - perhaps by acting as a safety net, helping to relieve the pressure of their workload, or maybe even helping to speed up diagnosis for patients. Improved diagnosis can mean earlier, more effective treatment for the patient.

But more research is needed, and other similar studies around the world are ongoing, investigating just how AI could help with cancer screening and detection.

AI could help us to better diagnose rare cancers

AI’s ability to detect complicated trends and patterns also means that it could help doctors to diagnose and treat more rare and unusual cancers.

Rare cancers can sometimes be harder to diagnose and treat, partly because we know less about how they grow and behave.

In a brand new UK study, researchers used an AI tool to analyse scans of patients with a rare type of cancer, called retroperitoneal sarcoma. This type of cancer grows deep in the abdomen (the tummy), near to major organs like the kidneys. It is difficult to characterise, and invasive biopsies are usually needed to determine how fast growing, and aggressive the cancer is.

The study found that compared to a biopsy, the AI tool was better able to predict which cases of cancer were aggressive, and could also potentially help with diagnosis of more rare types of the disease, which are harder to spot.

This study is only the first stage of research in this area, but it is exciting.

If AI can help to improve diagnosis in this way for rare cancers, it can help doctors better understand the type of cancer they are dealing with, and make sure that patients receive the most effective treatment as soon as possible.

AI can make treatments more accurate

AI looks set to improve the precision of more invasive cancer treatments.

Many cancer treatments involve removing all or part of a cancer, whether by surgery, radiotherapy, and other techniques like ablation (where cancer cells are destroyed using targeted methods, like heat or cold).

With these methods, it is really important to destroy and remove as much cancer as possible, while preserving surrounding healthy tissue.

Accuracy is everything, and research is beginning to show that AI tools could help doctors to improve their aim. For example, researchers have already trained an AI tool to accurately guide doctors carrying out thermal ablation for some cases of liver cancer.

And in another project, AI is being used to help doctors work out where to target radiation beams during radiotherapy, and avoid healthy parts of the body. Doctors and researchers involved in this work hope that this AI approach will help to reduce the time it takes to prepare for radiotherapy, and perhaps also cut waiting times.

AI can help to improve personalised medicine

AI looks set to improve the precision of more invasive cancer treatments.

Many cancer treatments involve removing all or part of a cancer, whether by surgery, radiotherapy, and other techniques like ablation (where cancer cells are destroyed using targeted methods, like heat or cold).

With these methods, it is really important to destroy and remove as much cancer as possible, while preserving surrounding healthy tissue.

Accuracy is everything, and research is beginning to show that AI tools could help doctors to improve their aim. For example, researchers have already trained an AI tool to accurately guide doctors carrying out thermal ablation for some cases of liver cancer.

And in another project, AI is being used to help doctors work out where to target radiation beams during radiotherapy, and avoid healthy parts of the body. Doctors and researchers involved in this work hope that this AI approach will help to reduce the time it takes to prepare for radiotherapy, and perhaps also cut waiting times.

AI will boost discovery cancer research

AI tools will also help cancer researchers make even more groundbreaking discoveries.

Researchers are already developing AI tools that can analyse large datasets containing information about molecular structures, biological pathways, and drug interactions to identify possible new drugs. This has the potential to help accelerate drug discovery and develop new cancer treatments.

And scientists also want to use AI to help them understand more about exactly how cancer starts and develops, and what can be done to stop it.

Professor Peter Carmeliet and his team in Belgium are using their Curestarter funding to work with AI to develop what could one day become a new type of immunotherapy.

The researchers want to target a type of cell called an ‘endothelial cell’. These cells line blood vessels, and research suggests that they can sometimes dampen down the immune system’s ability to target cancer. This can make some immunotherapies less effective.

So the team want to use AI to help them search for the best way to ‘rewire’ these cells, and make them boost the immune system against cancer, instead of repressing it. If successful, this work could ultimately help more patients benefit from immunotherapy.