Future medicine: Man + machine

Machines could replace 80 percent of doctors, Eric Topol, MD, director of the Scripps Translational Science Institute in La Jolla, Calif., and chief academic officer for Scripps Health, shared during a recent webinar hosted by the Health Information Management and Systems Society.

Consider: IBM’s Watson technology has established its diagnostic prowess; ICU physicians tap into rolling robots to deliver remote care; telestroke networks have employed robots to help assess patients. There’s no doubt—these are compelling concepts.

But there is a quieter, and equally impressive revolution occurring in the radiology reading room. As imaging datasets continue their upward climb, computer-aided detection (CAD) can lend a powerful helping hand.

CAD has established its utility in breast cancer and is showing promise in aiding detection and workflow in a number of other malignancies.

MR is demonstrating its power in the detection, localization and characterization of prostate cancer. Diffusion-weighted MR also could provide a marker of tumor aggressiveness. However, reviewing multiparametric MR data are challenging, particularly for less-experienced readers.

Enter CAD. In a retrospective study, Thomas Hambrock, MBChB, of the department of radiology at Radboud University Medical Centre, Nijmegen, the Netherlands, and colleagues demonstrated that CAD could level the playing field between less-experienced and experienced radiologists. Using CAD, less-experienced radiologists improved prostate MR performance to a level on par with experienced radiologists.

Similarly, colon CAD could address important clinical challenges. Detection of 6-9 mm polyps is problematic on CT colonography. Specifically, although dedicated centers have achieved 85 percent sensitivity for lesions in this range, actual performance is lower in multicenter trials, and ranges between 59 and 78 percent.

The U.S. Preventive Services Task Force has cited the issue as a source of uncertainty regarding CT colonography.

However, Daniele Regge, MD, of the Institute for Cancer Research and Treatment, in Candiolo, Italy, and colleagues devised a prospective study and demonstrated that CT colonography CAD could raise sensitivity from 65.4 percent to 76.9 percent for lesions in the challenging 6-9 mm range.

The researchers noted that CAD added 1.6 minutes to the image review time, and they also suggested that improved training might boost readers’ confidence in CAD and thus deliver additional gains in sensitivity.

These studies represent a quieter, but equally important, revolution of man and machine. How is your organization leveraging CAD to revolutionize radiology? Please let us know.

Lisa Fratt, editor