Reliability of smartphone-based radiographic interpretation for evaluating cardiogenic pulmonary oedema in dogs.

Abstract

To compare the accuracy of radiological diagnosis of cardiogenic pulmonary oedema in dogs using two viewing platforms. A retrospective cross-sectional study including cases admitted between January 2012 and December 2016. Thoracic radiographs from 121 dogs with and without cardiogenic pulmonary oedema were included. The study was divided into two phases, and images were interpreted by two observers: a radiologist (R1) and a cardiologist (R2). In the first phase, images were sent in JPEG format for interpretation using a smartphone. In the second phase, DICOM images were interpreted at the standard workstation using a DICOM viewer. Receiver operating characteristic curves were generated to evaluate the accuracy of each method of radiological evaluation, and Spearman correlation was used to evaluate the correlation between the evaluations performed by R1 and R2. The accuracy of R1 in diagnosing pulmonary oedema was 94.1% (95% confidence interval: 88.3 to 97.5) for DICOM images and 90.7% (95% confidence interval: 84.1 to 95.1) for JPEG/smartphone. The accuracy of R2 was 89.5% (95% confidence interval: 82.6 to 94.4) for DICOM images and 90.6% (95% confidence interval: 83.9 to 95.2) for JPEG/smartphone. There was no statistically significant difference between these evaluation methods. The correlation coefficient for R1 and R2 was 0.815 and 0.761, respectively. Cardiogenic pulmonary oedema may be reliably diagnosed on thoracic radiographs viewed on a smartphone by a specialist, which may expedite treatment.