97 information. Access to this software was restricted to only those researchers who were directly involved in the study. In order to maintain patient privacy and confidentiality, each patient was assigned an anonymised study ID, which ensured that their personal information could not be linked back to them. MRI protocol In the participating centre, all MRIs were carried out with a Philips 3T MRI system (Philips, Ingenia, the Netherlands). An MRI protocol other than ‘cervix’, ‘endometriosis’, ‘abdomen’ or ‘pelvis’ was excluded from this study. In almost all cases, the protocol involved acquiring T2-weighted turbo spin echo (T2-TSE) images in the sagittal, coronal and transverse planes. Also, the protocol involves acquiring a T1-weighted sequence, often using the Dixon method. In general, a slice thickness of 3-5 millimetres was used. At the MST hospital, patients did not receive any antispasmodic medication prior to MRI. When patients received multiple MRI scans prior to hysterectomy, the most recent MRI to hysterectomy was assessed. Predictive model A multivariate logistic regression model was developed to predict the probability of adenomyosis diagnosis (204). The final model included age and BMI at MRI, symptoms such as dysmenorrhea and abnormal uterine bleeding and several MRI features. The developed and internally validated formula for the likelihood of the histopathological diagnosis of adenomyosis was as follows: != ! "( A$; 7% B& (%&@(9 .:*:+5,9$% (--(.;%%∗7. 0?("*@) $9 ?( 203) ∗4. ∗0. C0 (* ),$) $( D( E5 627 8(9;:%;7 ?9 <" =@>9:?% 808 )-∗.. 0% *( F; )%$7 ?( B" %@-9@? G0H)∗∗. .",((F())$$ (GH I6<< ³J BB (;%7?" @9?0)∗.(*0)&K# ! $ " %L.0+ (;%7?" @9?0)∗".**,M$ (NO4 P9=6 (;%7?" @9?0)∗"."+F) Statistical analysis The baseline characteristics of the patients with and without a histopathological adenomyosis diagnosis after hysterectomy were compared. These characteristics of the external dataset were compared with the characteristics of the original dataset. Categorical variables were reported as numbers and frequencies, and continuous variables as means with standard deviations. For group differences continuous variables were assessed using an independent ttest if they were normally distributed, and the Mann-Whitney U test if not. Categorical variables were analysed using the Chi-square test. Missing values
RkJQdWJsaXNoZXIy MTk4NDMw