Could we use an AI to detect lung cancer? The Chinese start-up infervision certainly thinks so

Saul Bowden

May 8, 2017

Lung cancer is a huge killer world-wide. In the United states there are around 225,000 cases every year and it is responsible for almost a quarter of all cancer deaths. The problem is even worse in China, with approximately 600,000 deaths due to lung cancer every year. This is probably why the Chinese startup Infervision has created an AI to detect lung cancer.

Infervision has three separate tools that it uses. The Intelligent CT Assisted Diagnosis (AI-CT) is designed to assist in early stage lung cancer screening. This AI is capable of detecting and highlighting cancer features on a series of CT images. It highlights any potential cancer nodules and lets doctors more easily identify them. It is designed to improve early diagnosis of lung cancer and give patients a better chance of survival.

The Intelligent X-ray Assisted Diagnosis (AI-DR) is designed to help radiologists identify lesions in the lung. The AI can detect over 20 different kinds of lesions and has even spotted ones that radiologists missed. The AI-DR is supposed to take on a lot of the difficult time consuming tasks to free the radiologist up to focus on more complex problems.

There is one final tool in Infervision’s arsenal. The AI scholar. This is an intelligent deep learning research platform that is designed to make it easier for doctors with no computing background to take advantage of complex learning algorithms. The AI scholar can process more than 100 images at once and will give doctors access to new tools.

Air Pollution and heavy smoking in China is expected to raise the number of lung cancer cases to 800,00 per year by 2020 – Credit Latin times

Why is Infervision pushing so hard to design an AI to detect lung cancer? The company hopes to address the growing problem that Lung Cancer presents in China. Massive air pollution is causing more and more cases and doctors are hard pressed to cope. Lung Cancer is particularly difficult to treat unless it is caught very early and in many parts of the world the resources for proper screening simply aren’t there.

Infervision hopes that by using an AI to detect lung cancer they can relieve some of the strain on doctors and hand over the repetitive, time consuming task of identifying cancer to an AI that will help assist radiologists in detection. While lung cancer is an acute problem in China early detection could help free up precious medical resources in other countries and using AIs to detect lung cancer and other forms of cancer is a step in the right direction.

Infervision have demonstrated the value of AI in assisting medical professionals and they have processed roughly 100,000 CT scans and 100,000 x-rays over the last year, which is phenomenal. Infervision’s efforts show that the automation revolution can be positive. The hope is that they will expand deep learning concepts to other areas of the medical profession.