A new digital analytics tool is helping hospitals of all sizes go digital
Back in 1895, Wilhelm Roentgen accidentally changed the course of medicine. While observing the effects of electricity on vacuum tube equipment, he noticed that a tube made a nearby sheet of barium platinocyanide fluoresce. He had his wife place her hand between the radioactive materials and a fluorescent plate, casting a bony shadow of her hand and wedding ring. He named these rays “X” for their unknown properties and invented a machine that has been scanning patients for more than 100 years.
X-ray is the oldest imaging modality, and it’s the reason GE entered the healthcare industry. But with the emergence of MRI and CT scanners that can show the tiniest crossing fibers of the brain or a single heartbeat in motion, it’s easy to think of X-ray as a dated, less innovative method of seeing inside the body.
However, with age comes experience, and X-ray remains the machine most often on the “front lines” of medicine. A parent rushes their daughter into the ER with a sports-related injury to the wrist, possible fracture – X-ray. An older man sees his doctor, showing symptoms of fever, chills and wheezing, possible pneumonia – X-ray. Just like first impressions with people, that first image taken helps set the care pathway going forward.
Which is why a team of GE Healthcare engineers is using the knowledge gained from more than a century in the industry and access to the world’s largest medical equipment installed base to reinvent X-Ray. They have created the X-ray Quality Application, an analytics tool that in the future, will integrate artificial intelligence (AI) and deep learning to help hospitals set a new standard for intelligent scanners and raise the bar for efficient, high-value patient care – starting with helping to automatically identify and analyze the root causes of rejected X-ray images.
The problem plaguing the X-ray industry
Today’s healthcare environment demands high-quality diagnostic images for accurate diagnosis while keeping patient radiation exposure to a minimum. Rejected images often are a significant barrier to achieving this goal.
A “reject” is an X-ray or other radiology image that is not usable due to poor image quality. X-ray reject rates in the U.S. can be as high as 25 percent, and the U.S. recommended target reject rate is 5-8 percent. A “repeat” means the same image must be retaken because the initial image is not usable, which can contribute to unnecessary patient radiation dose and operational inefficiencies such as decreased throughput, lost revenue potential and wasted staff resources and time, impacting both quality of care and a hospital’s financial outcomes.
Data analysis is an increasingly valuable tool to help reduce the number of rejected and repeated X-ray images. By identifying high repeat/reject rates among technologists, which can occur due to patient positioning errors, image artifacts and improper exposure, radiology departments can actively target training and education efforts to improve technologist skills and therefore patient care, quality and safety. Repeat/reject reporting is recommended by many professional societies and required by several states. According to the American Society of Radiologic Technologists, targeted training with images can reduce reject rates, but the data is often tedious and time consuming to extract. Some hospitals don’t even know their reject rate because reporting is so personnel-driven.
Working with healthcare providers to find a solution
Engineers from GE Healthcare’s X-ray business worked with the University of Washington Medical Center in the U.S. and Humber River Hospital in Canada to develop the X-ray Quality Application, which is part of a suite of GE Healthcare applications, intelligent devices and services that will help hospitals make faster and more informed decisions.
The software helps automatically identify and analyze the root causes of rejected X-ray images for quick and easy review in an “at-glance” dashboard, eliminating hours of manual work, so healthcare providers can take action and implement targeted improvement training.
“The UW QA technologist team was manually collecting this data for years,” said Dr. Kalpana Kanal, PhD, director of the diagnostic physics section at the University of Washington Medical Center Department of Radiology. “This consisted of walking to 13 different X-ray machines around the hospital and retrieving individual reject data files requiring 237 mouse clicks. Someone would then compile the data into a spreadsheet and analyze the data for trends to identify individual technologists for targeted training. The entire retrieval/analysis process took the team up to 7 hours – a whole day of work!”
The automation is more reliable than manual data retrieval and allows for more frequent data collection. Additionally, the software tool displays the rejected image for further analysis to determine why it was faulty.
“The reports help us identify technologists with higher than expected reject rates,” Dr. Kanal said. “We can then offer the technologist opportunities for self-correction and education.”
Humber River Hospital’s Medical Imaging Department is using the X-ray Quality Application as part of their quality assurance and continuous quality improvement strategy. Humber’s clinical reject rate is currently at 4.81 percent. Humber’s clinicians currently take more than 10,000 X-ray images a month.
“Using the X-ray Quality Application, we were able to identify chest and knee X-ray exams as having elevated reject rates, with knee X-ray exams being 13 percent of all the reject images from the department mostly due to patient positioning or a patient moving during the exam,” said Dolores Dimitropoulos, Manager of Medical Imaging at Humber River Hospital.
The team at Humber was able to see which technologists had the highest reject rates and develop a proactive approach to training to help improve performance. One technologist even showed a significant drop in reject rates over just three months.
“This solution has helped us innovate our imaging department while also paving the way for digital transformation throughout the hospital,” Dimitropoulos said. “By delivering actionable insights about rejected X-rays in real-time, we are able to infuse a culture of quality and efficiency that leads to an improved patient experience and empowered technologists who have the information needed to deliver high value care.”
 Little, Kevin J., et al. "Unified database for rejected image analysis across multiple vendors in radiography." Journal of the American College of Radiology 14.2 (2017): 208-216.
 Jones, A. Kyle, et al. "One year’s results from a server-based system for performing reject analysis and exposure analysis in computed radiography." Journal of digital imaging 24.2 (2011): 243-255.
 ASRT, Best Practices in Digital Radiography, ASRT White Paper (2012). Available at: https://www.asrt.org/docs/defaultsource/whitepapers/asrt12_bstpracdigradwhp_final.pdf