| Object:In this paper, the method of meta-analy sis, systematic review ofdomesti c and foreign related clinical data, diagnosis value to evaluate ultrasound, CT, MR I and other imaging methods in pancreatic cancer.Methods:"Pancreatic cancer""malignant tumor of the pancreas""contrast""CT""MRI" as the key words, the VIP, CNKI, CECDB, PUBMED, SD, OVID. Retriev al time for the years2000-2013at the same time, MESH combined with free word retrieval of. The experimental data in the literature into the extraction, rando m effects model meta-analy sis, using Meta-disc1.4software for data processing, u sing quantitative chi square test the research measure of heterogeneity index I2.Qu ality evaluation of the literature QUAD AS (quality assessment of diagnostic ac curacy studies) literature quality assessment, each item according to the "yes","no ","not clear"3criteria evaluation by2reviewers independently through literatur e, to resolve the disagreement. If meet all14criteria of quality evaluation, is "is ", to study the possibility of bias is extremely low, rated A-class; if any one or m ore of the quality evaluation standard can only partially satisfied, is "not clear", t he possibility of bias for the middle, rated B; any or a plurality of quality evalu ation standards do not meet is "no", the possibility of high bias, rated C. Inclusion criteria: literature language Chinese or English. The experimental sample is greater than20. The standard diagnosis of pancreatic cancer: confirmed by operation o r fine-needle aspiration in the diagnosis of pancreatic carcinoma, clear, or biopsy pathology, Chest. It can directly or indirectly to obtain the original data of diag nostic tests, such as the number of true positive false, positive, true negative rat e and false negative rate. Exclusion criteria: the information is not comprehensiv e, such as abstract, review articles and conference proceedings; the selected sampl e does not meet the diagnostic criteria; insufficient number of samples in20cas es; the data, data is not complete or repeat data published literature.Results:1. Weighted sensitivity, specificity, positive likelihood ratio of1contra st-enhanced ultrasound, negative likelihood ratio, DOR=0.913(95%CI:0.870-0.945),0.901(95%CI:0.852-0.939),6.556(95%CI:3.812-11.277),0.124(95%CI:0. 081-0.189),71.541(95%CI:36.254-141.18), area under the SROC curve of0.9544, Q*=0.8965.2. Weighted sensitivity, specificity, positive likelihood ratio of2CT examination, negative likelihood ratio, DOR=0.916(95%CI:0.871-0.950),0.879(95%CI:0.814-0.928),6.309(95%CI:4.146-9.506),0.134(95%CI:0.086-0.208),66.517(95%CI:30.824-143.54), area under the SROC curve of0.9444, Q*=0.8330.3. Weighted sensitivity, specificity, positive likelihood ratio of3MR examination, negative likelihood ratio, DOR=0.945(95%CI:0.914-0.967),0.837(95%CI:0.767-0.893),4.868(95%CI:3.443-6.883),0.086(95%CI:0.0052-0.141),72.160(95%CI:35.029-148.65), area under the SROC curve of0.9443, Q*=0.8830.4. In this study, the statistical software output of ultrasound contr ast, CT, MRI ROC scatter distribution are not typical "shoulder shape", no het erogeneity threshold effect. Inspection, the sources of heterogeneity of other res ults, have varying degrees of heterogeneity in the study of three kinds of exa mination room. Contrast enhanced ultrasound: Heterogeneity chi-squared=12.98(d.f.=7) P=0.073, I2=46.1%, has the medium heterogeneity; CT:Heterog eneity chi-squared=14.41(d.f.=8),P=0.072, I2=44.5%, with moderate het erogeneity; MRI:Heterogeneity chi-squared=8.70(d.f.=8),P=0.369, I2=8%, with mild heterogeneity.Conclusion: The accurate rate of ultrasound, CT, MRI in pancreatic cancer detect ion is of high clinical value in diagnosis, evaluation of resection of pancreatic ca ncer staging and operation. Three methods of checking the diagnostic value of no statistical difference. |
PDF Full Text Request