Study: Smartphone system may expedite stroke diagnosis
Treatment options for acute ischemic stroke have advanced, fueling the need for rapid acquisition, visualization and interpretation of images. However, current systems are constrained by three factors, wrote J. Ross Mitchell, PhD, of the imaging informatics laboratory at University of Calgary in Canada, and colleagues. He identified these as:
- Transfer of DICOM images to remote devices may be hampered by speed and security issues.
- Current viewing devices may limit physician mobility.
- Mobile devices may be unable to support advanced visualization.
Mitchell and colleagues developed a teleradiology system based on client-server architecture to address the challenges. With the system, the server loads and renders DICOM files and transmits the frame to the remote device, so images are not stored on the remote device. The process takes a few seconds and enables advanced visualization such as 3D volume rendering prior to transmission to the remote device.
The researchers designed a small retrospective study to determine whether the system provided the speed and accuracy required for diagnosis of acute stroke.
Two neuroradiologists reviewed 112 consecutive non-contrast CT (NCCT) brain scans and 65 CT angiography (CTA) head scans from the Calgary Stroke Program database. The first physician interpreted the images on a workstation with a medical grade display and then on the mobile device, with the workstation interpretations serving as ground truth for the study. The second reader reviewed the images on the mobile device only.
None of the NCCT studies were falsely diagnosed as stroke on the mobile device, reported the researchers. And both readers correctly diagnosed intracranial hemorrhage in seven patients, resulting in 100 percent sensitivity, specificity and accuracy and perfect interrater agreement.
Thirty-four NCCT studies revealed acute parenchymal changes on the workstation. “The sensitivity, specificity and accuracy of detecting acute parenchymal ischemic change were 94.1 percent, 100 percent and 98.09 percent, respectively, for reader 1 and 97.05 percent, 100 percent and 99.04 percent for reader 2 using [the mobile device].” However, among the 18 patients with dense vessel sign, reader 1 had four false positives and reader 2 had five false negatives.
Mean interpretation time for NCCT on the mobile device was 2.7 minutes for reader 1 and 2.3 minutes for reader 2, reported the researchers.
Both readers correctly diagnosed 17 of the 18 vessel occlusions on the CTA studies and missed one occlusion of the V4 segment of the vertebral artery, resulting in sensitivity, specificity and accuracy of 94.4 percent, 100 percent and 98.46 percent, respectively, for both readers on the mobile device.
Mean CTA interpretation time was 3.63 minutes for reader 1 and 3.83 minutes for reader 2.
Mitchell and colleagues reviewed several potential shortcomings of the system. Prior to the development of the system, they had noted that the small display size, 320 x 480 pixel capacity and nonmedical-grade display might limit clinical utility.
However, they reported, “[I]n most NCCT brain scans the patient’s brain does not fill the entire 512 x 512 image array. In addition, readers were able to easily observe all image regions at native resolution through interactive translation, while interactive zoom allowed them to focus on image areas of interest.” Readers compensated for display size by holding the mobile device closer to their eyes. Finally, the mobile display provided 25 percent higher pixel density and luminance than the workstation used in the study.
Mitchell and colleagues acknowledged multiple limitations to the study including its retrospective nature and low incidence of hemorrhagic strokes among the patients, before calling for a larger prospective study of the system under realistic clinical conditions.
The researchers concluded, “[T]he smartphone client-server teleradiology system appears promising and may have the potential to allow urgent management decisions in acute stroke.”
Health Canada has approved the system as a medical device under device class 2 and it is under FDA review.