On radiographic image analysis

On radiographic image analysis
复制标题

关于射线照相图像分析

DOI:
10.1007/3540075798_21
复制
发表时间:
1976
期刊:
J. Electronic Imaging
影响因子:
--
通讯作者:
G. Lodwick
G. Lodwick
中科院分区:
--
文献类型:
--
作者:
C. Harlow;S. Dwyer;G. Lodwick

文献摘要

被引文献

相似文献

当放射线图像分析的初步工作开始时,人们对放射线照片计算机分析中涉及的几个因素感到担忧。这种担忧主要集中在三点。首先,分析简单性质的图像(例如字符和块)的已知困难似乎表明,进行放射线图像分析这样困难的项目是不明智的。第二个问题是射线照片中存在大量信息,可能需要 1024 x 1024 或超过一百万个数据点的数字化 [3.1]。这一事实似乎表明有必要使用大型计算机,而这在临床环境中是不切实际的。第三点是放射科医生对使用增强或修改的图像进行诊断缺乏热情。这意味着一些最先进的图像分析方法将具有不确定的价值。拥有用于分析射线照相图像的系统可能带来的优势在一定程度上缓解了这些问题。也就是说,用于分析射线照片的计算机系统提供了潜在的优势,允许放射科医生处理更多的患者,提供改进的诊断,并使他的诊断更加可重复或标准化。理想情况下,计算机系统应该是放射科医生使用的诊断工具,为他提供有关射线照片中感兴趣的项目的定量和诊断信息。目前分析射线照相图像的一大困难是需要分析的胶片量很大。例如,1964 年,人类拍摄了 5.06 亿张 X 光照片,1970 年每年增加到 6.6 亿张 [3.2]。仅 1970 年就有 7100 万次胸部放射检查(一张或多张胶片)I-3.3]。此外,在 15 年期间,放射服务的需求以 7.1% 的速度增长,而放射科医生的数量仅增加了 5.9%。据估计,对放射科医生的需求是现有数量的两倍[3.4]。当前分析放射线图像的方法的另一个困难是不同放射科医生在不同时间观看同一胶片的差异。例如,一些
When the initial work on radiographic image analysis was begun, there was concern about several factors involved in the computer analysis of radiographs. This concern focused on three points. First, the known difficulty of analyzing images of a simple nature such as characters and blocks seemed to indicate that it would be unwise to undertake such a difficult project as the analysis of radiographic images. The second point of concern was the tremendous amount of information present in a radiograph, possibly requiring a digitization of 1024 x 1024 or in excess of one million data points [3.1]. This fact would seem to indicate the necessity of utilizing a large computer which would be impractical in a clinical environment. The third point was the lack of enthusiasm of radiologists for using enhanced or modified images for diagnoses. This means that some of the best developed methodology for image analysis would be of uncertain value. These problems are somewhat mitigated by the advantages that might accrue from having a system for the analysis of radiographic images. That is, a computer system for analyzing radiographs offers the potential advantages of allowing a radiologist to handle more patients, to give improved diagnosis, and to make his diagnosis more repeatable or standardized. Ideally, the computer system should be a diagnostic tool used by the radiologist to give him quantitative and diagnostic information on items of interest in the radiograph. One difficulty at present in the analysis of radiographic images is the large volume of film to be analyzed. For example, in 1964 there were 506 million radiographs taken of humans, and thus increased to 660 million per year in 1970 [3.2]. In 1970 alone there were 71 million radiographic chest examinations (one or more films) I-3.3]. Also, over a fifteen-year period, it was determined that the demand for radiological services increased at a rate of 7.1% while the number of radiologists increased by only 5.9%. It has been estimated that the need for radiologists is twice the number available [3.4]. Another difficulty in current methods of analyzing radiographic images is the variation among different radiologists seeing the same film at different times. For example, some of