Quality Project

Quality in healthcare is more than providing the best medical advice to patients, it’s about ensuring consistent clinical outcomes with predictable business results. Poor quality leaves patients, providers, and administrators frustrated and dissatisfied. Unfortunately, quality is often hard to measure without incurring costs or impediments in the workflow that make it impractical. In this project, we examined a method to automatically extract common medical imaging quality issues so that automated, feedback systems could be created.

Quality is difficult to measure. Everyone can differentiate excellent from poor quality when they see it. However, putting quality metrics in place can be difficult, time consuming, labor intensive, and not always be consistent with the behavior that administration desires. One of the common observations from quality officers is that the right metrics are often too hard to capture because they add additional burdens to front-line healthcare professionals that are not appreciated and therefore not universally adopted.

In this engagement, the client was interested in extracting medical imaging quality in an automated manner. An unobtrusive approach was chosen by examining the self-reported issues in radiology reports. The client recognized that this would be just the “tip of the iceberg” of image quality issues due to self-reporting bias as well as the number of images that were discarded prior to even reaching the radiologist. However, the metric at least provided a minimum assessment of the issue.

The radiology reports were examined using natural language processing techniques to extract a variety of imaging issues from protocol compliance (e.g. contrast injection) to image formation to reconstruction. The metric showed that image quality issues were reported in roughly 2.4% of all captured images ranging by modality from 1 to 6%.

With such a metric in place, the client can now look to correlate image quality issues with other practice management variables such as the technologist, protocol choice, patient handling, and specific imaging equipment.

Snowstone can help you minimize the cost of your quality measurements and maximize the realized benefit for your patients and staff.