Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – New research on Central Nervous System Diseases and Conditions - Meningeal Neoplasms is the subject of a report. Accor ding to news originating from Qingdao, People’s Republic of China, by NewsRx cor respondents, research stated, “This study evaluates the efficacy of radiomics-ba sed machine learning methodologies in differentiating solitary fibrous tumor (SF T) from angiomatous meningioma (AM). A retrospective analysis was conducted on 1 71 pathologically confirmed cases (94 SFT and 77 AM) spanning from January 2009 to September 2020 across four institutions.” Our news journalists obtained a quote from the research from the Affiliated Hosp ital of Qingdao University, “The study comprised a training set (n=137) and a va lidation set (n=34). All patients underwent contrast-enhanced T1-weighted (CE-T1 WI) and T2-weighted(T2WI) MRI scans, from which 1166 radiomics features were ext racted. Subsequently, seventeen features were selected through minimum redundanc y maximum relevance (mRMR) and the least absolute shrinkage and selection operat or (LASSO). Multivariate logistic regression analysis was employed to assess the independence of these features as predictors. A clinical model, established via both univariate and multivariate logistic regression based on MRI morphological features, was integrated with the optimal radiomics model to formulate a radiom ics nomogram. The performance of the models was assessed utilizing the area unde r the receiver operating characteristic curve (AUC), accuracy (ACC), sensitivity (SEN), specificity (SPE), positive predictive value (PPV), and negative predict ive value (NPV). The radiomics nomogram demonstrated exceptional discriminative performance in the validation set, achieving an AUC of 0.989. This outperformanc e was evident when compared to both the radiomics algorithm (AUC= 0.968) and the clinical model (AUC = 0.911) in the same validation sets. Notably, the radiomic s nomogram exhibited impressive values for ACC, SEN, and SPE at 97.1% , 93.3%, and 100%, respectively, in the validation set .”
NETL
NSTL