Skip to main content
Fig. 3 | BMC Medical Imaging

Fig. 3

From: Prediction of BRCA gene mutation status in epithelial ovarian cancer by radiomics models based on 2D and 3D CT images

Fig. 3

Application of LASSO (Least absolute shrinkage and selection operator)-logistic regression to imaging feature screening in the 2D + 3D model shows that the LASSO-logistic regression model selects tuning parameters (λ) through tenfold cross-validation and obtains the relationship between binomial variance and logarithm (λ) (a). The relationship is retained with the parameters that yield the smallest binomial deviation, and the 12 best features with non-zero coefficients (b) are retained in the final model. The relationships between the features and gene mutation status (correlation coefficient × 100) are shown in the heat map (c)

Back to article page