Towards a low-resource digital infrastructure to streamline clinicians’ viewing and interpretation of radiographic images: a case study of Nsambya hospital

Abstract

Background: Picture Achieving and Communication Systems (PACS) are said to improve quality of care through timely access to radiological images. However, Anecdotal reports suggest that it takes about 30-40 minutes to access patient radiographic images by clinicians. In this study, the diagnostic accuracy and image quality of core i3 ordinary PC based display systems were assessed as cheaper alternative solutions for PACS workstations. Materials and methods: The study adopted a comparative study to collect quantitative data. 40 images were randomly acquired and viewed from PACS workstation and PC display. A sample of 20 patients was used to measure patient turnaround time (PTAT). Sensitivity and specificity values, Diagnostic odds ratio (DOR) and receiver operating characteristic (ROC) curve analyses were used to measure accuracy of clinical outcomes. Image quality was assessed using Image Criteria and a Pre- and post- evaluations were observed for PTAT. Results: There was moderate agreement among raters for both display systems (kappa 0.644 for PACS and 0.5164 PC). Accuracy rates were 80% for ordinary PC display and 97.3% for PACS workstation with AUC values of 0.9706 for PACS and 0.7990 for PC. 97.5% of the images viewed on PC were considered fully visible and sharply reproduced. The PTAT before switching to the PC display gave a mean of 22.65 minutes and a mean PTAT of 4.8 minutes after the switch. Clinician's diagnosis showed close similarity between PACS workstation and PC display. The results showed that the PC display equally yielded good detection of abnormalities on the chest x-ray images as PACS. The majority of observed x-ray images passed the quality criteria on ordinary PC (97.5 %). Conclusion: Use of Ordinary core i3 PC based systems is possible and can provide accurate image evaluation and diagnosis, and significantly reduce waiting time in a resource-limited setting