TH‐C‐330A‐03: Does Display Bit‐Depth Influence Observer Performance?

Elizabeth A Krupinski, H. Roehrig, J. Fan

Research output: Contribution to journalArticle

Abstract

Purpose: The display a radiologist reads images from represents a crucial link in the information chain. There has been much debate recently on what bit‐depth is necessary for optimal reading. Greater bit depth permits more gray levels to be displayed initially, possibly obviating the need for the observer to window/level the image during viewing. This project compared an 8‐bit with 11‐bit LCD display in terms of observer performance. Method and Materials: Two Totoku LCD displays (3 Mpixel, grayscale) were used that were identical in every way except for the bit‐depth. One was 8‐bit and the other 11‐bit. Max luminance was set to 500 cd/m2 and min luminance was set to 0.6 cd/m2. The displays were DICOM calibrated and room lights were set to the average display luminance with an image displayed. A set of 100 DR chest images, half with nodules and half without, were shown to 6 radiologists in a counter‐balanced Receiver Operating Characteristic study. They reported whether a nodule was absent or present and rated heir confidence. Results: Performance (ROC Az values) was essentially equivalent for the two bit‐depth displays. There were no significant differences in viewing time or how often the readers used the window/level functions. Conclusion: For chest images with nodules it appears that having greater than 8‐bit depth does not significantly enhance observer performance. It is not clear whether the results would be the same with different types of images or lesions. Conflict of Interest: Totoku provided the displays, software and funds to reimburse the radiologist observers.

Original languageEnglish (US)
Pages (from-to)2266
Number of pages1
JournalMedical Physics
Volume33
Issue number6
DOIs
StatePublished - 2006

Fingerprint

Thorax
Conflict of Interest
Financial Management
ROC Curve
Reading
Software
Light
Radiologists

ASJC Scopus subject areas

  • Biophysics
  • Radiology Nuclear Medicine and imaging

Cite this

TH‐C‐330A‐03 : Does Display Bit‐Depth Influence Observer Performance? / Krupinski, Elizabeth A; Roehrig, H.; Fan, J.

In: Medical Physics, Vol. 33, No. 6, 2006, p. 2266.

Research output: Contribution to journalArticle

@article{3372cc7e7e434839b87455071a13f0f4,
title = "TH‐C‐330A‐03: Does Display Bit‐Depth Influence Observer Performance?",
abstract = "Purpose: The display a radiologist reads images from represents a crucial link in the information chain. There has been much debate recently on what bit‐depth is necessary for optimal reading. Greater bit depth permits more gray levels to be displayed initially, possibly obviating the need for the observer to window/level the image during viewing. This project compared an 8‐bit with 11‐bit LCD display in terms of observer performance. Method and Materials: Two Totoku LCD displays (3 Mpixel, grayscale) were used that were identical in every way except for the bit‐depth. One was 8‐bit and the other 11‐bit. Max luminance was set to 500 cd/m2 and min luminance was set to 0.6 cd/m2. The displays were DICOM calibrated and room lights were set to the average display luminance with an image displayed. A set of 100 DR chest images, half with nodules and half without, were shown to 6 radiologists in a counter‐balanced Receiver Operating Characteristic study. They reported whether a nodule was absent or present and rated heir confidence. Results: Performance (ROC Az values) was essentially equivalent for the two bit‐depth displays. There were no significant differences in viewing time or how often the readers used the window/level functions. Conclusion: For chest images with nodules it appears that having greater than 8‐bit depth does not significantly enhance observer performance. It is not clear whether the results would be the same with different types of images or lesions. Conflict of Interest: Totoku provided the displays, software and funds to reimburse the radiologist observers.",
author = "Krupinski, {Elizabeth A} and H. Roehrig and J. Fan",
year = "2006",
doi = "10.1118/1.2241845",
language = "English (US)",
volume = "33",
pages = "2266",
journal = "Medical Physics",
issn = "0094-2405",
publisher = "AAPM - American Association of Physicists in Medicine",
number = "6",

}

TY - JOUR

T1 - TH‐C‐330A‐03

T2 - Does Display Bit‐Depth Influence Observer Performance?

AU - Krupinski, Elizabeth A

AU - Roehrig, H.

AU - Fan, J.

PY - 2006

Y1 - 2006

N2 - Purpose: The display a radiologist reads images from represents a crucial link in the information chain. There has been much debate recently on what bit‐depth is necessary for optimal reading. Greater bit depth permits more gray levels to be displayed initially, possibly obviating the need for the observer to window/level the image during viewing. This project compared an 8‐bit with 11‐bit LCD display in terms of observer performance. Method and Materials: Two Totoku LCD displays (3 Mpixel, grayscale) were used that were identical in every way except for the bit‐depth. One was 8‐bit and the other 11‐bit. Max luminance was set to 500 cd/m2 and min luminance was set to 0.6 cd/m2. The displays were DICOM calibrated and room lights were set to the average display luminance with an image displayed. A set of 100 DR chest images, half with nodules and half without, were shown to 6 radiologists in a counter‐balanced Receiver Operating Characteristic study. They reported whether a nodule was absent or present and rated heir confidence. Results: Performance (ROC Az values) was essentially equivalent for the two bit‐depth displays. There were no significant differences in viewing time or how often the readers used the window/level functions. Conclusion: For chest images with nodules it appears that having greater than 8‐bit depth does not significantly enhance observer performance. It is not clear whether the results would be the same with different types of images or lesions. Conflict of Interest: Totoku provided the displays, software and funds to reimburse the radiologist observers.

AB - Purpose: The display a radiologist reads images from represents a crucial link in the information chain. There has been much debate recently on what bit‐depth is necessary for optimal reading. Greater bit depth permits more gray levels to be displayed initially, possibly obviating the need for the observer to window/level the image during viewing. This project compared an 8‐bit with 11‐bit LCD display in terms of observer performance. Method and Materials: Two Totoku LCD displays (3 Mpixel, grayscale) were used that were identical in every way except for the bit‐depth. One was 8‐bit and the other 11‐bit. Max luminance was set to 500 cd/m2 and min luminance was set to 0.6 cd/m2. The displays were DICOM calibrated and room lights were set to the average display luminance with an image displayed. A set of 100 DR chest images, half with nodules and half without, were shown to 6 radiologists in a counter‐balanced Receiver Operating Characteristic study. They reported whether a nodule was absent or present and rated heir confidence. Results: Performance (ROC Az values) was essentially equivalent for the two bit‐depth displays. There were no significant differences in viewing time or how often the readers used the window/level functions. Conclusion: For chest images with nodules it appears that having greater than 8‐bit depth does not significantly enhance observer performance. It is not clear whether the results would be the same with different types of images or lesions. Conflict of Interest: Totoku provided the displays, software and funds to reimburse the radiologist observers.

UR - http://www.scopus.com/inward/record.url?scp=85024785319&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85024785319&partnerID=8YFLogxK

U2 - 10.1118/1.2241845

DO - 10.1118/1.2241845

M3 - Article

AN - SCOPUS:85024785319

VL - 33

SP - 2266

JO - Medical Physics

JF - Medical Physics

SN - 0094-2405

IS - 6

ER -