Cancer researchers are testing whether artificial intelligence can help doctors decide who should receive chemotherapy. More than 1,000 Norwegian cancer patients will participate in the study.
Computers are trained to be able to estimate who is likely to relapse after cancer surgery. Tarje Svensgerd Hviem and Andreas Klebe from the Department of Cancer Genetics and Informatics are leading a new study.
The short version
- Artificial intelligence (AI) can help identify patients who need chemotherapy after bowel cancer surgery.
The summary is created with the help of artificial intelligence (AI) and quality guaranteed by Aftenposten journalists.
For many people with cancer, chemotherapy is a common treatment. But chemotherapy is also a stressful treatment that can cause a number of side effects. For example, nausea, weak immune system, fatigue, and neuropathy.
Now researchers at Oslo University Hospital are testing a new method that could save many bowel cancer patients from chemotherapy. Today, researchers believe that more people are receiving chemotherapy without it being necessary.
– Many of these patients recover from the operation. They do not need chemotherapy after surgery. But it's difficult to determine who doesn't need it, says Andreas Klepp. He is a researcher in the Department of Cancer Genetics and Informatics.
At the institute, they analyze tissue samples from thousands of patients suffering from, among other diseases, bowel cancer, prostate cancer, lung cancer and uterine cancer.
The goal is to see if it is possible to teach a computer to do something it cannot do today.
Such images are fed into computers. The image shows a tissue sample from a cancer patient. Computers create digital versions, which can then be studied on a computer screen and analyzed using artificial intelligence.
The AI will recommend treatment
About 2,000 patients in Norwegian and British hospitals are participating in the study.
The institute uses a separate method they developed, called histological profiling. The images are scanned and transferred to computers. Artificial intelligence is then unleashed on the images.
There are pathologists who analyze tissue samples for cancer. They are specialized doctors.
Laboratory engineer Ingrid Elise Vidal prepares tissue samples for scanning. Samples must be analyzed by a computer and a pathologist.
Thanks to today's technology, computers are able to distinguish between healthy tissue and cancer.
Klebi points out that today it is difficult for a computer to predict a patient's condition after surgery or a cancer diagnosis.
But with their new method of artificial intelligence, researchers already see that computers can do even more.
– What can patients get from the research?
Computers can be trained, among other things, to select patients who do not need chemotherapy after bowel cancer surgery, says researcher Andreas Klepp.
– We are interested in having our computer system useful in clinical practice, say Tarjei Sveinsgjerd Hveem and Andreas Kleppe.
– For example, artificial intelligence can provide information about the diagnosis of a patient's condition. There may be information that the pathologist does not consider. They're better together than apart.
The new technology primarily assesses the aggressiveness of the cancer. The result will be the same as the patient's condition several years after the operation. The researchers point out that all the factors affecting this are not yet known.
Will pathologists eventually be replaced by artificial intelligence?
– You can train AI to perform pathological assessments. In the future, technology may be used to do some of the pathologist's simpler tasks or help the pathologist do their tasks faster.
Pathologist Manohar Pradhan analyzes the digital tissue sample. Artificial intelligence (AI) can provide support to pathologists on what they are seeing and what type of treatment the patient should receive next. Artificial intelligence is on its way to the Norwegian healthcare system.
In the laboratory, specialist engineer Philipp Nicolassen colors tissue sections before photographing them and entering them into the computer.
They can already be used in hospitals
The computer software used in the study has been in development for several years. The researchers have already published some results in Nature reviews of clinical oncology.
The institute cooperates with researchers at the University of Oxford. They want to use the technology there now.
– The system is CE markCE markThe CE mark is the manufacturer's guarantee that all safety directives have been met. CE is an abbreviation for Conformité Européenne.. Andreas Klepp explains that the system could be used in European hospitals, for example, to personalize treatment after bowel cancer surgery.
The researchers strongly believe that the new technology could enable more patients to avoid heavy chemotherapy.
Today, many patients receive chemotherapy after surgery to be on the safe side. But this could change in the long term.
Computers offer many advantages. They may find unknown features of the cancer. It may be something that has not been discovered or has not been adequately studied.
With enough data, AI will be able to learn how to use such features to estimate forecasts more accurately, says Kleeb.
He points out that the goal is for the largest possible number of cancer patients to receive the treatment that suits their disease.
Our new study will show that artificial intelligence can be used to avoid over-treatment in hospitals in Norway.
Tissue samples are placed on a glass dish and stained before being photographed and entered into a computer. Here in the laboratory of the Department of Cancer Genetics and Informatics, part of the Radium Hospital.
“Web specialist. Lifelong zombie maven. Coffee ninja. Hipster-friendly analyst.”
