Study: Pediatric Fracture Detection
Reading Time: 3 minutes read
As a research scientist at Carestream Health, my recent work has focused on pediatric X-ray imaging – specifically, on the goal of achieving high-quality diagnostic images while minimizing dose. To explore potential solutions to this challenge, I initiated a study in collaboration with Samuel Richard, PhD, a fellow researcher at Carestream, and Sosamma T. Methratta, MD, of the Milton S. Hershey Medical Center in Hershey, PA. The study had two specific goals: 1) Determine the impact of a simulated reduced dose rendering on the detection of skeletal fractures in children, and 2) Evaluate the effect of enhanced skeletal processing on the same detection task. The methodology and results of this study were on display at RSNA 2015.
Background: Imaging of pediatric patients is notoriously difficult. First, the xray exam process itself is unfamiliar to young children and can frighten them. Moreover, children who have sustained an injury are understandably distressed and agitated. Consequently, it’s often difficult to keep them still during the exam. Their near-continuous movement can make obtaining a clear, high-quality image problematic – and often necessitates multiple retakes, leading to increased X-ray dose to the child.
This issue of compromised pediatric image quality is of particular concern for radiologists who are called to testify on their findings where child abuse has been alleged. Clearly, these doctors would like to feel a high level of confidence in their conclusions – but that confidence can be eroded in the face of sub-par image quality.
The study: 150 DR pediatric exams (50 with fractures, 100 without) were presented to five pediatric radiologists for reading and evaluation. Each exam was viewed in the following ways: (1) original exposure with standard processing; (2) simulated reduced exposure (720 equivalent film speed) with standard processing; (3) the view in (1) paired with an enhanced processing view; (4) the view in (2) paired with an enhanced processing view. Original exposure renderings in (1) were acquired with site-specific default techniques, and then processed with default parameters of a pediatric-tuned multi-frequency rendering algorithm. Enhanced processing-view renderings added a companion image to either the “original exposure/standard processing view” or the “reduced exposure/standard processing view” to accentuate skeletal interruptions. Base images used for reduced exposure images were simulated from original exposure images with a validated noise-add model.
The participating radiologists read the images, marking them where they judged a fracture to be present. Their findings were then evaluated for accuracy. Fractures that were correctly identified were categorized as “true positives.” Non-fractures mistakenly marked as fractures were labeled as “false positives.” “True negatives” were identified when a radiologist correctly perceived no fractures where none existed. Missed fractures were labeled as “false negatives.”
Results: The accuracy with which the radiologists read exams in the four categories demonstrated that:
- Diagnostic-quality images of pediatric patients may be captured with DR using exposures at least as low as 720 speed.
- Viewing the “simulated reduction/standard processing image” paired with the “simulated reduction/enhanced processing image” is comparable to viewing only the “original exposure/standard processed image” for detecting fractures using the “original exposure/enhanced image.”
- DR coupled with the Enhanced Skeletal Processing can improve the visibility of fine detail in pediatric images – for an enhanced level of diagnostic confidence.
- Enhanced processing offers the opportunity to reduce cumulative dose exposure to pediatric patients – reducing long-term radiation risk.
Read the related blogs on:
Shriners Hospital for Children share their insights for imaging pediatric patients
An experienced rad tech shares his tips for imaging young patients
#RSNA #pediatricimaging #radiology
Lynn La Pietra, PhD, is a Senior Research Scientist at Carestream Health