Connie Rees

47 using RevMan 5.3.0. It was not deemed relevant to assess the quality of the other studies investigating adenomyosis on MRI for risk of bias. Data collection and analysis: Study Selection Study selection was conducted in Rayyan (rayyan.qcri.org) in a blinded fashion by two reviewers (CR & IR) based on title and abstract, followed by full-text assessment. If full-text was not available, contact was sought with the corresponding author of the article. The relevant articles were sorted into one or both of two groups upon full-text screening and data extraction done separately (Tables 2.S1 – 2.S4): - Studies investigating diagnostic accuracy of adenomyosis on MRI versus histopathology - Studies evaluating adenomyosis objectively on MRI without assessing diagnostic accuracy Data collection and extraction Data extraction was done independently by two reviewers (CR & IR). Data pertaining to study design, study population, the type of MRI conducted, definition of adenomyosis on MRI, the diagnostic performance (if reported) of MRI parameters, as well as if the MRI diagnosis was confirmed with histopathology was collected. All measured MRI characteristics of adenomyosis on MRI were extracted, with a focus on objective measures. Examples of objective measures include: JZ thickness (in mm), uterine volume or length, adenomyosis lesion size (in mm, or cm3). If clinical or treatment outcomes were mentioned (in relation to MRI characteristics), these were also reported. See supplementary file for the full data extraction table (Tables 2.S1-2.S4). Data synthesis and statistical analysis: Data synthesis was done in two steps, dependent on the study design. For studies investigating diagnostic accuracy, data was synthesized narratively, and the diagnostic performance measures were summarized. We produced forest plots in Graphpad Prism 8.0 showing pairs of sensitivity and specificity together with 95% confidence intervals from each study (based on extracted data in 2 x 2 tables for each study). Summary receiver operating

RkJQdWJsaXNoZXIy MTk4NDMw