论文部分内容阅读
作者从国内52份资料(包括作者自己的资料)统计了经纤维胃镜检查的35,529病例,根据检查结果分为四组(胃良性疾病、其它、正常、胃癌),分别计算其条件概率与验前概率,与选择的十八个参数作为信息输入计算机,编制程序。将待诊新病人信息送入计算机,即可输出此患者四种疾病的概率,其中最大的概率值作为电子计算机助诊(CAD)的结果。经病理证实的58例胃癌病人 CAD 的准确率为87.93%,可靠性为89.4%,四组总计132例,总符合率为81.82%。CAD 对胃癌的鉴别诊断可能有价值。统计学显示 CAD 的准确性优于钡餐透视。
The author collected 35,529 cases of endoscopic gastroscopy from 52 domestic data (including the author’s own data) and divided them into four groups (benign gastric diseases, normal gastric cancer, and gastric cancer) according to the examination results, and separately calculated the conditional probabilities and pretest The probability, with the selected eighteen parameters, is entered into the computer as information, programmed. The new patient information to be diagnosed into the computer, you can output the probability of the four diseases in this patient, the largest of which probability value as a result of computerized aided diagnosis (CAD). The accuracy of CAD was 87.93% in 58 gastric cancer patients confirmed by pathology, with a reliability of 89.4% and a total of 132 cases in four groups with a total coincidence rate of 81.82%. The differential diagnosis of gastric cancer by CAD may be of value. Statistics show that the accuracy of CAD better than barium meal perspective.