Kimme-Smith C, Haus A G, DeBruhl N, Bassett L W
Iris Cantor Center for Breast Imaging, UCLA School of Medicine 90095, USA.
AJR Am J Roentgenol. 1997 Mar;168(3):775-8. doi: 10.2214/ajr.168.3.9057533.
Viewing conditions can affect diagnostic performance differently depending on background optical densities. We quantified detection accuracy when viewing calcifications in glandular tissue under recommended viewing conditions versus accuracy with lower view box luminance and higher ambient lighting.
A phantom with adipose, 50% adipose and 50% glandular, and glandular-simulating material was imaged, and images were interpreted by five medical imaging physicists using two lighting conditions: the recommended one, high view box luminance (4365 nits) with low ambient light (25 lx), and a suboptimal one, low view box luminance (1763 nits) with moderate ambient lighting (290 lx). Then, a dense (Breast Imaging Reporting and Data System breast composition pattern type 4) unfixed cadaveric breast with numerous native calcifications was imaged 28 times. Nineteen of the films had added clusters of simulated calcifications. Three radiology fellows, each with 11 months of training in mammography, identified the added calcification clusters in the images under the two lighting conditions. Changes in phantom analysis and accuracy of the clinical diagnosis were compared for each lighting condition.
On mammograms of the phantom, both speck and fibril identification were degraded by an average of 1.4 objects for the adipose-simulating section (with its darker optical density). For medium optical densities, found in the section with the simulation of 50% glandular and 50% adipose tissue, suboptimal lighting conditions had little or no effect on speck and fibril identification. For sections of the phantom that simulated glandular tissue, an average of 0.6 specks or fibers were not seen when lighting was suboptimal. With the dense cadaveric breast, the fraction of added calcification clusters detected by the three observers improved by an average of 17% when low luminance viewers and high ambient light were replaced with recommended viewing conditions; individual scores of the observers improved significantly: p values ranged from .02 to .05.
Luminance of the view box and ambient lighting significantly affect detection of calcifications in dense breasts when images are interpreted by radiologists with about 1 year of training in mammography. Detection of calcifications in phantoms is primarily degraded for adipose tissue with its darker optical density. However, when lighting conditions are suboptimal, some observers also have trouble detecting calcifications in glandular tissue with its low optical density.
视场条件会根据背景光学密度不同程度地影响诊断性能。我们对比了在推荐视场条件下观察腺体组织中的钙化灶时的检测准确性，以及视箱亮度较低且环境光照较高时的准确性。
对含有脂肪、50%脂肪和50%腺体以及模拟腺体材料的体模进行成像，由五名医学影像物理学家在两种光照条件下解读图像：推荐条件，视箱亮度高（4365尼特）且环境光低（25勒克斯）；次优条件，视箱亮度低（1763尼特）且环境光照适中（290勒克斯）。然后，对一个致密的（乳腺影像报告和数据系统乳腺组成模式类型4）未固定的尸体乳腺进行成像，该乳腺有大量天然钙化灶，共成像28次。其中19张胶片添加了模拟钙化灶簇。三名各有11个月乳腺摄影培训经验的放射科住院医师在两种光照条件下识别图像中添加的钙化灶簇。比较每种光照条件下体模分析的变化和临床诊断的准确性。
在体模的乳腺造影片上，对于模拟脂肪的部分（其光学密度较暗），斑点和纤维的识别平均减少了1.4个。对于中等光学密度，即在模拟50%腺体和50%脂肪组织的部分，次优光照条件对斑点和纤维的识别几乎没有影响。对于模拟腺体组织的体模部分，光照次优时平均有0.6个斑点或纤维未被看到。对于致密的尸体乳腺，当将低亮度视箱和高环境光替换为推荐视场条件时，三名观察者检测到的添加钙化灶簇的比例平均提高了17%；观察者的个人得分有显著提高：p值范围为0.02至0.05。
当由有大约1年乳腺摄影培训经验的放射科医生解读图像时，视箱亮度和环境光照会显著影响致密乳腺中钙化灶的检测。对于光学密度较暗的脂肪组织，体模中钙化灶的检测主要受到影响。然而，当光照条件次优时，一些观察者在检测光学密度较低的腺体组织中的钙化灶时也会遇到困难。
