^<99m>Tc-Sn Colloid肝シンチグラムによるび漫性肝疾患のAutomated Computerized Pattern Characterizationの試み

概要

論文の詳細を見る
The clinical importance of hepatic scintigrams for the differential diagnosis of diffuse liver diseases has been reported in various studies together with the diagnostic criteria. However, most of the criteria proposed are based on manual measurements and / or visual inspection, and, thus, the personal subtle deviation of criteria is inevitable. In this study, the characterization of patterns of right lateral hepatic scintigrams obtained by automated computerized processings is investigated. Several methods are examined to extract the hepatic region from the digitized image, finding that intensity methods with threshold levels based on the variance and gradient criteria can be effectively applied. Two methods, named the rectangular method and the circular method, are also proposed in this study for the characterization of the hepatic patterns. The diagnostic effectiveness of each combination of methods for the region extraction and the pattern characterization was evaluated by applying multivariate discriminant analysis. Seventy six cases, which are already diagnosed pathologically by biopsy, are used for this study, containing 5 of acute hepatitis (abbreviated as AH), 15 of chronic inactive hepatitis (CHI), 20 of chronic active hepatitis (CHA), 7 of liver cirrhosis in early stage (E-LC), 17 of liver cirrhosis (LC) and 12 of control (N). Based on the results of the reclassification of all sample cases by combining the rectangular method and the circular method, 4 of 5 AH (80%), 11 of 15 CHI (73.3%), 17 of 20 CHA (85%), 7 of 7 E-LC (100%) and 16 of 17 LC (94.1%) were accurately diagnosed by the automated computerized differential diagnosis. The over all accuracy rate was 66/76 (86.8%).
公益社団法人日本放射線技術学会の論文
1979-11-01

^<99m>Tc-Sn Colloid肝シンチグラムによるび漫性肝疾患のAutomated Computerized Pattern Characterizationの試み

スポンサーリンク

概要

著者

関連論文

スポンサーリンク