Validation of the novel International Association for the Study of Lung Cancer grading system and prognostic value of filigree micropapillary and discohesive growth pattern in invasive pulmonary adenocarcinoma

Diagnostic Performance of Radiomics in Prediction of Ki-67 Index Status in Non-small Cell Lung Cancer: A Systematic Review and Meta-Analysis

“…We observed a spectrum of cut-off values for Ki-67 across studies, encompassing values of 5% (12,13), 10% (14,15), 14% (16), 30% (17), 40% (18)(19)(20), and 50% (21). Most studies utilized immunohistochemical analysis to assess the Ki-67 index in samples obtained from surgeries or biopsies (12-15, 18, 19, 21).…”

“…Image segmentation was mainly carried out manually by one or two skilled radiologists (12)(13)(14)(15)(16)(17)(18)21). However, in a particular study, two blinded radiologists were involved in performing semiautomatic segmentation (19).…”

“…The predominant segmentation tool employed across the majority of studies was ITK-SNAP (60%) (12)(13)(14)(16)(17)(18). A single study utilized MaZda software (21), another study leveraged 3D Slicer (15).…”

“…Different software tools were used in extracting radiomic features across studies. GE Healthcare's Analysis Kit (AK) software emerged as a common choice, employed in multiple studies for this purpose (12,13,17,18). Additionally, Feature Explorer (FAE) open-source software (14), the 3D Slicer PyRadiomics plugin (15), the python-based PyRadiomics implementation (16,20), the Radcloud platform (19), and MaZda software (21) were among the tools utilized.…”

“…Various methods for selecting radiomics features were utilized in the included papers, with a majority incorporating Least Absolute Shrinkage and Selection Operator (LASSO) in 80% (12)(13)(14)(15)(17)(18)(19)(20) of the studies and minimum redundancy maximum relevant (mRMR) in 30% (12,14,18) of them. Additionally, other methods were employed, such as Recursive Feature Elimination (RFE) (13), variance thresholding (19), SelectKBest (19), Random Forest Feature Selection (RFFS) (21), Pearson correlation coefficient (20), and a combination of Mann-Whitney U test, Spearman test, RF-based Boruta algorithm (16).…”