Fine-tuning the reading room

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon

A series of recent studies presented at SCAR on Friday shone new light on reading room efficiency. The three presentations explored on and off axis viewing, optimization of reading conditions for flat panels, consumer displays for primary interpretation, performance differences among 1 to 5 megapixel monitors and voice recognition macro use.

Elizabeth Krupinski, PhD, of University of Arizona, presented a study of on and off axis viewing of images with CRT and LCD monitors.
   
With LCDs quickly replacing CRTs, facilities need to be aware of one disadvantage of LCDs: off-axis perception distortions. Krupinski aimed to determine whether or not the distortions affect performance in a study with radiologists reading 400 images derived from 80 mammograms.
   
The study found similar performance among monitors when images were viewed straight on. However, at a 45 degree angle, contrast and performance on LCD monitors dropped. JNDMetrix and Channelized model results confirmed the findings, leading Krupinski to conclude that LCDs are better than CRTs for on-axis review, but not off-axis review. Radiologists need to be aware of this and avoid off-axis viewing, Krupinski said.

Jerry A. Thomas, MS, of Uniformed Services University of the Health Sciences hypothesized that there may be a need to limit minimum illumination in the reading room.
   
Thomas completed the study with a 5 megapixel CRT monitor and two 9.2 megapixel LCDs and found non-linear range of luminance vs. illuminance in the flat panels. "The introduction of a requirement of minimum room illuminance could eliminate possible contrast non-linearities," concluded Thomas. He cited a need for further studies to determine the best approach.
   
David Hirschorn, MD, of Massachusetts General Hospital, questioned, "Is the monochrome LCD necessary and optimal for modalities other than radiography?"
   
Hirschorn compared industry-standard 3 megapixel grayscale monitors and consumer, color flat panel displays for reading of chest CT studies. He found no significant differences in negative studies, lesion detection or conspicuity for lesions on both systems. Moreover, reader concordance was near perfect. The conclusion? There is no difference in performance between consumer color and 3 megapixel grayscale monitors for detection of lung lesions on CT studies.
   
Nabile Sadfar, MD, of the University of Maryland School of Medicine compared one, 2, 3 and 5 megapixel LCD monitors for detection of cervical spine fractures.
   
"There is a trend toward decreasing resolution," Sadfar said. "What is the correct matrix size?" The study entailed 200 lateral cervical x-rays with radiologists reading the studies on the four monitors. Sadfar found similar sensitivity, specificity and accuracy on all systems, demonstrating 1, 2, 3 and 5 megapixel monitors are adequate for the detection of cervical fractures on cervical x-rays.
   
Mark Zucherman of StructuRad presented research conducted at Geisinger Medical Center and shared insights into the use of macros in voice recognition.
   
Geisinger radiologists are stratified in their use of voice recognition, with the bulk of radiologists either embracing or avoiding the technology. Moreover, successful voice recognition users employ macros, showing a correlation between use of macros and voice recognition, but not a cause and effect.
   
Zucherman provided an overview of types of macros. Complete or near complete macros produce a report, such as a normal chest x-ray. A report skeleton outline offers a fill-in-the-blank approach, and report-build macros allow the radiologist to build a report with short, specific statements.
   
Issues affecting macro use include name dependency (the number of macro names to remember), report formatting issues, ease of editing and the ability to share macros across an organization. Zucherman suggested that users understand and build naming conventions, relying on an organized structure to enhance their ability to recall names. Also, users can build libraries of personal and organizational macros to drive use across the facility.