iPad TB diagnosis stacks up well vs. LCD monitor

There is no detectable effect of using an iPad rather than a traditional LCD monitor for the diagnosis of tuberculosis (TB), according to a study published in the January issue of the Journal of the American College of Radiology.

The results offer more support for the use of tablet computers in medical imaging and allowing radiologists to use their skills outside of the reading room environment.

“Mobile radiologic diagnosis holds the promise of expanding radiologists' availability for consultation and reducing dependence to specific fixed locations for review of medical images,” wrote Samir Abboud, MS, and colleagues from the University of Maryland School of Medicine in Baltimore. Mobile displays would allow attending radiologists to provide quick review during tumor board consultations and small-group teaching activities, they added.

"The mainstay of diagnostic radiology is, and should remain, the workstation," Abboud wrote in an email to Health Imaging. "However, interpretation of imaging studies is only half of the equation - communication to our clinical colleagues is equally important. While the tablets are not likely to change how studies are read, they do hold promise as communication and education tools."

The study focused on the use of an iPad 2 to interpret de-identified DICOM images of 240 chest x-rays for suspected TB. Images were reviewed independently by five radiologists of varying experience. Readers graded the images as positive or negative for TB on both the iPad and a LCD monitor, with reading sessions spaced two weeks apart to minimize recall bias.

For readers using the same display, complete agreement was found in 90 percent of all cases, with a higher proportion (0.94) of complete agreement in cases classified as negative for TB, reported the study authors.

Agreement between the consensus of readers on each display was present in all but two cases. One case with a consensus classification of positive on the iPad was classified as negative on the LCD workstation, while a second case classified by consensus as negative on the iPad was classified as positive on the LCD monitor. Overall, multirater generalized Cohen’s K was 0.969, according to Abboud and colleagues.

Individual readers tended to agree with their own interpretations on either display, with K statistics ranging from 0.786 to 0.969. More experienced readers had higher K values, according to the authors. The readers also commented that using the iPad felt slower than the LCD monitor, and also that the screen of the iPad was much smaller than the LCD monitor.

"The limiting factor is a question of real estate," said Abboud. "The typical tablet is smaller than a sheet of notebook paper, while a typical workstation can have two or three monitors with a 21" diagonal or greater."

The authors wrote that, despite being suitable for TB diagnosis as they reported, they would not recommend using the iPad as the primary diagnostic display. “Further studies to determine the accuracy of iPad interpretations for a variety of disease processes and careful phantom studies will be necessary before the iPad (which is itself evolving in image display quality) could be considered for use as a primary diagnostic display.”