Premium
Impact of calibration methods for color medical displays on interpreting brain SPECT images
Author(s) -
Shibutani Takayuki,
Onoguchi Masahisa,
Yoneyama Hiroto,
Nakajima Kenichi,
Matsuo Shinro,
Bamba Yusuke,
Saito Shinji
Publication year - 2019
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1002/mp.13532
Subject(s) - imaging phantom , grayscale , nuclear medicine , artificial intelligence , computer vision , computer science , medicine , image (mathematics)
Purpose Color images are visualized on medical monitors that are adjusted by a grayscale standard display function ( GSDF ) or γ 2.2. Although the GSDF is visually displayed as a linear graded grayscale, it does not specify how color medical images should be presented. On the other hand, the usual gamma setting for color images is γ 2.2, but it has not been standardized. The color standard display function ( CSDF ) has recently been proposed as a standardized gamma setting for color medical monitors. However, the influence of various gamma settings on image characteristics should be determined. The present study aimed to identify differences in color‐scale characteristics on nuclear medicine images displayed on medical monitors adjusted by CSDF , GSDF , and γ 2.2. Methods Transverse normal (n = 1) and abnormal (n = 5) brain perfusion images were generated using a mathematical digital phantom. Transverse phantom and clinical brain images are shown using the clinically applied eight‐color scale. Five nuclear medicine experts visually assessed phantom and clinical images using a defect severity scale that ranged from zero (no defect) to four (defect). Receiver operating characteristic curves were created and areas under the curves ( AUC s) were analyzed. Defect scores for the clinical study were evaluated in the nine segments on basal ganglia slice, and defect scores were summed for each patient. Results The average defect score for color A significantly differed in multiple comparison tests, but not in post hoc tests. The ranges of AUC for CSDF , GSDF , and γ 2.2 were 0.86–0.94, 0.82–0.94, and 0.88–0.97, respectively. The AUC s of CSDF in all color scales did not significantly differ from other gamma settings. The summed defect scores of CSDF were similar to those of other gamma settings. Conclusion Nuclear medicine images were equally valid when adjusted by CSDF even at various gamma settings. Nuclear medicine images can be evaluated equally using any gamma setting. Nonetheless, the color gamma setting for medical monitors should be standardized.