Artificial intelligence shows promise in detecting lung cancer
Lung cancer can be there lurking in plain sight, only to be caught too late to be adequately treated.
In fact, it is the sixth most common cause of death worldwide, according to the World Health Organization. Doctors currently have to sift through hundreds of 2-D CT scan images to try to discern the presence of the illness – a process that takes time and isn’t very accurate.
In a new study, Google and medical partners including Northwestern University have unveiled a new AI-based tool that can create a better model of a patient’s lung from the CT scan images. This 3-D image gives better predictions about the malignancy of tumors and incorporates learning from previous scans, enabling the AI to help clinicians in spotting lung cancer in earlier stages when it is vastly more treatable.While the disease can be quashed if found soon enough, Google notes that a miniscule slice of the eligible U.S. population is screened for lung cancer. Along with the tremendous work and opaque outcomes a human-only review can produce, AI assistance in reviewing medical imaging and building a statistical model offers to lift some burden off of clinicians’ shoulders.
What is the trend?
The surge of data across the healthcare industry means all sorts of practitioners are inundated with information they need to review.As providers begin to invest more in population health, having the ability to optimize and improve predictions and diagnoses through the AI-assisted imaging means leveraging technology to free providers up to perform tasks they are better suited to.
Google’s study says AI technology has the possibility to predict patient outcomes and suggest treatment plans with a greater degree of accuracy than practitioners alone.
Why it matters?
Artificial Intelligence has been showing how it can look closer at patient data, be it imaging from a CT scan machine to biometrics from a smartwatch, often picking up on diseases and conditions better than practitioners could on their own.Cancer seems to be a particularly good target for machine learning. With a diagnosis relying so heavily on imaging scans that a computer has been demonstrably better at doing and successful treatment so reliant on early detection, many health researchers have shown where AI can root out cancer more effectively than humans alone.
Benjamin Harris is a Maine-based freelance writer and former new media producer for HIMSS Media.