AI-Enabled Blood Test Can Detect Brain Tumors With 92% Success Rate
Thu, April 22, 2021

AI-Enabled Blood Test Can Detect Brain Tumors With 92% Success Rate

An artificial intelligence-enabled blood test can now detect signs of brain tumors with 92 percent success / Photo by: westsib via 123RF

 

An artificial intelligence-enabled blood test can now detect signs of brain tumors with 92 percent success, according to the study that was presented at the recent Cancer Conference conducted by UK-wide partnership the National Cancer Research Institute. The researchers believe that it would help improve the survival of patients as their diagnosis will be done more efficiently and quicker than before.

Brain Tumors Diagnosis: Traditional vs. Using AI

The traditional diagnosis of brain tumors has been challenging because the condition tends to have vague symptoms, such as memory problems and headaches. Doctors would have to rely on a brain scan to diagnose them. But with artificial intelligence, the AI will look into chemical clues shed by the tumors into a person’s blood.

University of Edinburgh’s honorary consultant neurosurgeon and senior clinical lecturer Dr. Paul Brennan explained that the life expectancy of patients diagnosed with brain tumors is reduced by an average of 20 years, which is the highest of any type of cancer. He added that nowadays, 62 percent of patients are only diagnosed in the emergency department, although these same people may have already seen their GP many times before the diagnosis. The neurosurgeon believes that the reason for this is that diagnosing brain tumors is “so difficult.” A patient may be suffering from a headache and it is already a sign of a brain tumor, but there is also a possibility that the cause is something else.

It is likewise not practical on the part of the GP to send as many people to undergo a brain scan because they complained of a headache. The challenge to the doctors is to identify who they should prioritize that needs an urgent brain scan.

The traditional diagnosis of brain tumors has been challenging because the condition tends to have vague symptoms, such as memory problems and headaches / Photo by: auremar via 123RF

 

Identifying Patients Who Most Likely Have a Brain Tumor

This led Dr. Brennan to work with the University of Strathclyde’s reader in chemistry and ClinSpec Diagnostics Ltd.’s chief scientific officer Dr. Matthew Baker. The two created a test to help doctors efficiently and quickly identify patients who most likely have a brain tumor. Their test relies on infrared spectroscopy, a technique that examines the chemical makeup in the patient’s blood, coupled with artificial intelligence that can detect chemical clues in the blood.

To test their discovery, the researchers used the blood samples of 400 patients with possible brain tumors. These are patients referred by the UK-based Western General Hospital to undergo a brain scan; 40 of them were subsequently diagnosed with a brain tumor.

“Extremely Promising” Result

The AI-powered blood test correctly detected 84 percent of the people who had a low false positives rate or those who don’t have brain tumors. The test also correctly identified 82 percent of the subjects who have brain tumors. In terms of detecting patients with glioma, the most common form of brain tumor, the technology achieved a 92 percent accuracy rate. Dr. Baker went on to say that the results of the AI-powered blood test were “extremely promising” as it can spot not just the people who have a brain tumor but also those whose results are negative.

It likewise requires a small blood sample, making the technique a potential solution to test a large number of individuals suspected to have brain tumors or to determine those who need an urgent brain scan. Overall, it will lessen the patients’ anxiety of waiting for the test and get them treated the soonest possible time.

The researchers said that they will be trying out their technique with 600 other patients, who either were referred to undergo a brain scan by the hospital’s emergency department or directly by their GPs. The team believes that the AI-powered blood test can be used in other types of cancer that are also not easy to diagnose, such as prostate, bowel, pancreatic, and ovarian cancer.

Although not involved in the study, NCRI’s glioma subgroup and Addenbrooke’s Hospital’s Clinical Director for Cancer Dr. Sarah Jefferies commented that there is an urgent need for better ways to treat and detect brain tumors as the number of people dying because of it has already increased. She believes such a kind of testing provides “potential advantages” and is ideal because it is straightforward.

Brain Tumors, AI Application in Healthcare: Statistics

The American Brain Tumor Association shared that nearly 80,000 new cases of primary brain tumors are expected to be diagnosed this year--32 percent of which are malignant, but about 16,000 people “will lose their battle” with the central nervous system and primary malignant brain tumors.

Accenture analysis showed that the health AI market will register an explosive CAGR of 40 percent through 2021. From $600M market size in 2014, it is set to reach $6.6billion in 2021. The top 10 AI applications in healthcare as well as its value are the following: robot-assisted surgery ($40B), virtual nursing assistants ($20B), administrative workflow assistance ($18B), fraud detection ($17B), dosage error reduction ($16B), connected machines ($14B), clinical trial participant identifier ($13B), preliminary diagnosis ($5B), automated image diagnosis ($3B), and cybersecurity ($2B).

The American Brain Tumor Association shared that nearly 80,000 new cases of primary brain tumors are expected to be diagnosed this year--32 percent of which are malignant / Photo by: agenturfotografin via 123RF