Medical Diagnosing Artificial Intelligence

"We have some of the biggest computers in the world. We started wanting to push the boundaries of basic science to find interesting and cool applications to work on."
"The whole experimentation process is like a student in school."
"We're using a large data set for training, giving it lessons and pop quizzes so it can begin to learn for itself what is cancer, and what will or will not be cancer in the future."
"We gave it a final exam on data it’s never seen after we spent a lot of time training, and the result we saw on final exam — it got an A."
Dr. Daniel Tse, project manager at Google

Kateryna Kon | Shutterstock

In a new study published in the journal Nature Medicine, researchers from Google and several medical centers fed CT scans used to screen for lung cancer into a new algorithm system that has resulted from 'training' computers to recognize cancer. This, through a process titled Deep Learning. It is a process used to teach and enable computers to understand speech, to identify objects to a self-driving car to recognize a stop sign, for example. Pathologists can now read microscope slides through systems created by Google, along with systems to enable ophthalmologists to detect eye disease.

Lung cancer is an especially pernicious disease, responsible for the deaths of 1.7 million people around the world, last year alone. Beside finding cancers, scans are able also to identify spots that may potentially become cancerous at a later date. Scans, unfortunately, can miss tumors or erroneously take benign spots for malignancies. Radiologists viewing the same scan may have differing opinions. Leading the researchers to the thought that computers might improve performance accuracy.

With that end in mind, to produce a 'deep learning' situation in artificial intelligence, the researchers created a neural network, feeding it scans from patients with known diagnoses. Numbering 6,716 cases with known diagnoses, the system was tested for accuracy and it performed well up to initial expectations at a 94 percent accuracy. The deep learning model came up with fewer false positives and false negatives in a contest against six expert radiologists.

"I'm pretty confident that what they've found is going to be useful, but it's got to be proven", pointed out Dr. Eric Topol, not involved in the study, but with a background in artificial intelligence in medicine. He warns that a radiologist who misreads a scan may harm a patient, but a flawed A.I. system has the potential to harm many more. "We are collaborating with institutions around the world to get a sense of how the technology can be implemented into clinical practice in a productive way. We don't want to get ahead of ourselves", explained Dr. Tse.

But by feeding immense amounts of data from medical imagining into artificial neural network systems, computers lend themselves to training to recognize patterns linked to a specific condition. Learning as it goes, the system follows an algorithm; the more data it receives the more skilled it becomes."Radiologists generally examine hundreds of 2D images or 'slices' in a single CT scan, but this new machine learning system views the lungs in a huge, single 3D image", explained study co-author Dr. Mozziyar Etemadi, a research assistant professor of anesthesiology at Northwestern University Feinberg School of Medicine in Chicago.

"In order to build the AI to view the CTs in this way, you require an enormous computer system of Google-scale. The concept is novel, but the actual engineering of it is also novel because of the scale."
"The system can categorize a lesion with more specificity. Not only can we better diagnose someone with cancer, we can also say if someone doesn't have cancer, potentially saving them from an invasive, costly, and risky lung biopsy", Dr. Etemadi explained.

New research suggests that a computer algorithm may be better than radiologists at detecting lung cancer.