System Review Of Mammography, Magnetic Resonance Image, And Radionuclide Imaging On Breast Cancer With Exploring Involved Statistic Technique.

Part 1 System review of mammography, magnetic resonance image, and radionuclide imaging on breast cancer: a Meta analysisObjective: To compare the performance of several imaging modalities (mammography, magnetic resonance image, radionuclide imaging) in the detection of breast cancer by Meta analysis.Materials and Methods: PUBMED, OVID, database (1986-2006) were searched for the related articles. By the evaluation criterion of evidence based medicine(EBM), articles about the diagnosis of breast cancer using MG, MRI, PET and/or SPECT were find out. Related diagnostic index, such as TPR, FPR, FNR, and TNP, was acquired. Using software Excel 2003, RevMan 4.2, the pooled sensitivity, specificity, odds ratio (OR) and likelihood (LR) were worked out. And Summary receiver operating characteristic curve (SROC) was drawn to compare the accuracy of the several imaging modality vividly.Results: Thirty articles including 14MG (2941 lesions), 10 MRI (1428 lesions), 17 PET/SPECT (2247 lesions) met the inclusion criteria. The pooled sensitivity (mean( 95%CI)) of US,MG,MRI and PET/SPECT were 82%(76-86%), 86%(83-88%), and 87%(83-90%) respectively, pooled specificity, 69%(62-75%), 65%(62-69%)and 82%(76-86%). Area under SROC ( AUC-SROC) was 0.84, 0.89and 0.90 respectively, Q* value, 0.77, 0.81 and 0.88. Conclusion: The pooled sensitivity of the three modalities is similar. The pooled specificity of radionuclide imaging is higher than other two modalities. The pooled diagnostic efficacy of radionuclide imaging and MRI is higher than MG. Therefore, MG is still suitable for preliminary screening of breast cancer, MRI is recommended for young women, and Radionuclide imaging can be used to further confirm a breast cancer suspected by other two modalities. Part 2 Summary ROC plotting and its parameters estimation using Excel2003Objective: To set up a VBA based on Excel2003 to estimate and draw the SROC. Methods: Based on Excel2003, first, the forms and buttons of the VBA were produced by VB toolbox. Second, related Excel built-in functions were called to perform the logit transformation, plot the SROC, and work out the area under the SROC as well as the value of the Q*. At the same time initial data storage and management capabilities was established. Principles of mathematical statistics of SROC were based on the article of Guanjian LIU and Taixian WU. After the building of the VBA the data from the reported sample was processed several times to validate the in-house software by comparing its results with the results from Guanjian LIU and Taixian WU.results:The established VBA has user friendly interface and is very convenient for data entry and revision. Several times processing produced one result. The produced SROC plot was satisfactory and the area under the curve and the value of the Q* were consistent with those reported by Guanjian LIU, Taixian WU. Conclusion: the VBA based on Excel2003 can be used conveniently and reliably for Meta-analysis of the diagnostic tests. It's a useful tool for Meta-analysis of Evidence-based imaging medicine.Part 3 Meta analysis of diagnostic test with Review Manager 4.2The objective of this article was to introduce a new method to perform Meta analysis of diagnostic test with Reviewmanager 4.2. With an instance we demonstrate how to collect, range the raw data from a diagnostic test, and then input it into the Reviewmanager 4.2 for pooled positive likelihood ratio, pooled negative likelihood ratio, and pooled diagnostic odds ratio (DOR). According to these data the pooled sensitivity and specificity can be worked out.