sponsored

The New Radiology Assistant: Artificial Intelligence

With physicians in short supply, AI can assist physicians in diagnosing disease.

xrays
Dell EMC

The healthcare industry is in the midst of a perfect storm. As our population grows and ages, healthcare costs are rising at unsustainable rates. Against that backdrop, we have too few physicians, a problem that is expected to become more acute over the coming decade. A study conducted for the Association of American Medical Colleges (AAMC) predicts that the U.S. will face a shortage of 42,600 to 121,300 physicians by 2030. This shortage is expected to be particularly large in specialty-care fields.[1]

All of this points to the need to improve the efficiency of our healthcare processes and the productivity of our physicians. One of the keys to meeting these goals is the use of artificial intelligence to assist physicians in the diagnostic process.

As an editorial in the journal Nature notes, “AI diagnostics have the potential to improve the delivery and effectiveness of health care. Many are a triumph for science, representing years of improvements in computing power and the neural networks that underlie deep learning.”[2]

A case in point

In the United States alone, each year about 1 million people seek care in hospitals due to pneumonia, and around 50,000 people die from the disease, according to the U.S. Centers for Disease Control (CDC).[3] Millions of other Americans are living with emphysema and other forms of chronic obstructive pulmonary disease (COPD).[4]

Better diagnostics, assisted by AI, could help stem this deadly tide. That’s a key takeaway point from a study conducted by a team of data scientists from Dell EMC with support from Intel. In this study, the research team demonstrated the potential of deep learning algorithms to assist professional radiologists in the detection of pneumonia and emphysema.[5]

For the study, the research team developed models for using AI to diagnose pneumonia, emphysema and other thoracic pathologies from chest X-rays. Using a Stanford University neural network called CheXNet for inspiration, the team explored ways to develop accurate diagnostic models with fast parallel training on a compute cluster based on Intel® Xeon® processors. The team conducted the training on the Zenith supercomputer at the Dell EMC HPC and AI Innovation Lab.

As summarized in an IDG white paper, sponsored by Dell EMC and Intel, “Tests showed that the model the Dell EMC team built not only performed better than the original CheXNet model, but also outperformed baseline tests in 10 out of 14 different categories, including diagnosis of emphysema, a lung condition that afflicts an estimated 3.5 million Americans.” [6]

A physician’s assistant

AI studies often prompt the question of whether artificial intelligence will replace the work of humans, in this case radiologists. The quick answer is not at all. AI systems will instead serve as assistants that help physicians make better and faster diagnoses, freeing up time for them to help more patients.

“Ultimately, it’s about technology enhancing human performance,” IDG says in the white paper. “While some people debate whether AI-powered machines will eliminate jobs, professionals in the chronically short-handed healthcare field welcome the assistance. Faster and better image diagnosis enables radiologists to work on the problems that machines still can’t tackle, extending care to a greater number of people.”

For the full story, read the IDG white paper “Dell EMC AI Researchers Achieve Diagnostic Accuracy Rates that Match Human Radiologists.”  And if you’re of a mind to dive down into the technical details of the team’s study, read the white paper “Fast and Accurate Training of an AI Radiologist on Intel Xeon-based Dell EMC Supercomputers.”

[1] Association of American Medical Colleges, “GME Funding and Its Role in Addressing the Physician Shortage,” May 29, 2018.

[2] Nature, “AI diagnostics need attention,” March 13, 2018.

[3] U.S. Center for Disease Control and Prevention, “Pneumonia.”

[4] National Emphysema Foundation, “COPD and Emphysema afflict millions of adults and children today.”

[5] IDG Communications white paper, “Dell EMC AI Researchers Achieve Diagnostic Accuracy Rates That Match Human Radiologists.”

[6] IDG Communications white paper, “Dell EMC AI Researchers Achieve Diagnostic Accuracy Rates That Match Human Radiologists.”