CVIB software engineers KP Wong and Liza Shrestha achieved the landmark of the first image pushed to PACS by any research or commercial AI software at UCLA. Liza led deep learning efforts to automatically detect endotracheal and nasogastric tubes along with the relevant anatomic landmarks to check their position. KP Wong collaborated with UCLA Health IT, OHIA and ISS, to integrate the software into the clinical workflow. Chest x-rays are automatically routed to an AI server for processing and the resulting image pushed back to PACS within minutes where it is available for ICU physicians to perform an initial check of the tube position. The image is also available for ACI radiologists to support their review and reporting. Significant technical collaborations were required to move this technology from the lab successfully into the clinical workflow at UCLA where it can be most impactful on patient care. A quality improvement project is underway to evaluate the usefulness of this investigational software in clinical practice. Within the UCLA environment the AI is running on Microsoft’s Azure cloud and CVIB computer scientist, Matthew Brown, was interviewed by Microsoft about the collaboration.
