Liver
Conditions
Interventions
Procedure/surgery
E01.370.350.825.500.150
E05.196.867.519.550
The analysis of the magnetic resonance imaging exams in five stages (arms) with a determined time interval between them, by two different radiologists blindly regarding clinical-laboratory data and de
Sponsors
Faculdade de Ciências Médicas da Universidade Estadual de Campinas
Faculdade de Ciências Médicas da Universidade Estadual de Campinas
Eligibility
Age
18 Years to 78 Years
Inclusion criteria
Inclusion criteria: Magnetic resonance imaging of the abdomen using gadoxetic acid as a contrast agent in the assessment of focal liver lesions that had already been identified in previous examinations (ultrasound, computed tomography or magnetic resonance with conventional gadolinium), but which still remained undetermined, requiring complementation diagnostics, performed at the HC of FCM / UNICAMP, from January 2015 to December 2018
Exclusion criteria
Exclusion criteria: Absence of definitive diagnostic criteria for focal liver lesions; radioablation and / or chemoembolization prior to the lesion to be analyzed; examination artifacts preventing adequate characterization of the lesion to be analyzed; absence of detection of the focal lesion in the MR examination; age under 18 and over 78
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| Expected outcome: It is expected to assess interobserver agreement at all stages in the detection and characterization of benign and malignant focal liver lesions. It is expected that the accuracy will be increased by adding the dynamic phases (step 2), hepatobiliary phase (steps 3 and 4) and diffusion sequences (step 5). Only lesions that appeared in the same location at the different stages of MRI and in the criteria for the definitive diagnosis were considered correctly detected and characterized by the observers. The weighted kappa coefficient was used to assess interobserver agreement. To compare the stages and to verify the influence of the type (benign and malignant), and size of the lesion in the detection and characterization of the lesions in the stages, the Generalized Estimating Equations-GEE method was used. The estimates were calculated by maximum likelihood to weight the difference in the number of repetitions for each patient. Multiple comparisons were assessed using the contrast test and odds ratios (OR-odds ratios) and 95% confidence intervals were presented, when necessary. To evaluate the accuracy of each step in relation to the definitive diagnosis, the ROC curve (Receiver Operating Characteristic) was used for repeated measures. The observations within a given patient are not independent and the intra-patient correlation and variation was introduced in the analyzes using a mixed generalized linear model. The comparison of the accuracy between the steps was performed by estimating a logistic regression model for repeated measures by the method of generalized estimation equations (EEG). The level of significance adopted was 5%. The diagnostic performance index was also calculated by adding the lesions correctly classified as definitely benign (score 1) or preferably benign (score 2) among the benign according to the definitive criteria and those categorized as preferably malignant (score 4) or definitely malignant (score 5) among malignant accord | — |
Secondary
| Measure | Time frame |
|---|---|
| Secondary outcomes are not expected | — |
Countries
Brazil
Contacts
Public ContactDaniel Fernandes
Faculdade de Ciências Médicas da Universidade Estadual de Campinas
Outcome results
None listed