Clustering knee radiographs from RA patients
In the study, five paremeters were obtained from 831 knee radiographs from RA patients before total knee arthroplasty. Datasets were stored in the folder named 'Data'. (Radiographs were not stored in terms of personal information protection.) Based on five parameters, clustering 831 radiographs into 3 clusters. (The algorism was coded in 'Clusteringkmedoids.m') We found clustering into 3 groups was the best way, caluculating the Calinski-Harabasz Index. (Coded in 'OptimizationClusterNum.m') Comparison of the parameters between 3 clusters were coded in 'CompareParametersBetweenClusters2.m'. Comparison of clinical data between 3 clusters were coded in 'CompareClinicalData_Bonferroni.m'. The comparison of the radiographic parameters between clusters were done in 'ANOVAparametersBonferonni.m'. The functions used in the algorism were stored in the folder named 'function'. The code was described in MATLAB 2022a. For non-MATLAB users, we published the code in PDFs which were stored in the folder named 'ViewAlgorismPDF'.