Detection of cancer so far

In Sunshine Coast, Australia, a policeman saw an ad for free AI testing for cancer. Intrigued and with a history of prostate cancer in his family, he decided to get one. He found out he had cancer. He had suffered from none of the usual symptoms so far despite being aware of them.

In New York, a biopsy confirmed that a young girl’s cancer had come back. The neuropathologist noticed a slight unusual pattern in the biopsy. They turned to AI to classify the tumour and the result showed a new kind of cancer which would require a completely different treatment.

There are many such stories from across the world, all coming down to the same fact; Artificial Intelligence is key in catching cancer early and is on its way to becoming more commonly utilised and accepted. Catching cancer early can lead to higher survival rates.

Nothing artificial about AI’s diagnosis
In 2019, Google announced that it had created technology to pick out cancer symptoms which regular CT scans and radiologists missed. They showed examples of a report where their AI powered software picked out cancer symptoms even when the CT scan showed normal results.

Google isn’t the first nor will it be the last. Multiple companies, universities are researching AI and its effectiveness in helping detect different kinds of cancer.

An AI powered blood test
In March 2020, Annals of Oncology published a paper partly funded by Princess Margaret Cancer Centre where researchers talked about an AI powered blood test which can identify up to 50 different kinds of cancer. The test was able to predict the tissue in which the cancer originated in 96% of samples, and it was accurate in 93%.

How does the technology work?
Tumours shed DNA in the blood and this contributes to cell free DNA, known as cfDNA. The confusion with testing arises because cfDNA can come from any type of cell. This test can predict the tissue in which the cancer originated with an accuracy rate of 93%.

Why is detecting cancer so hard?

The 12 kinds of cancers which account for 63% of cancer deaths every year include lung, anal, oesophageal, stomach, head, lymphoma and cancers of white cells. For these cancers, the AI blood test supported detection rate is 67.3% across all clinical stages. Presently, there is no way of screening for these cancers before the symptoms show.
Many researchers are relying on Artificial Intelligence to help screen for these cancers before they reach a deadly stage.  

AI research on cancer detection
Last week, a University of Canterbury student come up with a new computerized method of reading mammograms that will make it easier for radiologists to detect breast cancer.
In 2018, researchers at University of Central Florida developed an AI system to detect lung cancer in CT scans even when they were tiny specks. The AI system is accurate 95% of the time as compared to 65% when done by a human.

FDA approves marketing AI device to detect colon cancer
And just last month, the U.S. Food and Drug Administration approved the marketing of an Artificial Intelligence device that uses AI based on machine learning to help clinicians in detecting lesions during a colonoscopy in real time. The detection rate went up by 13% with the help of the device. 

Cancer detection remains elusive in many cases due to a multitude of factors. With more funding and research with AI, if cancer can be detected early, then the survival rate will also go up, giving people a chance to live their lives despite being diagnosed with a disease which has no cure yet.