Combining Intravoxel Incoherent Motion Diffusion Weighted Imaging and Texture Analysis for a Nomogram to Predict Early Treatment Response to Concurrent Chemoradiotherapy in Cervical Cancer Patients

Table 4

The ROC analysis for IVIM-DWI, TA, and IVIM-DWI with TA.

Features	AUC	95%CI	Accuracy (%)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Overall IVIM-DWI	0.874	0.802–0.947	80.6	91.4	74.1	68.1	93.5
D	0.848	0.768–0.927	78.5	77.1	79.3	69.2	85.2
f	0.825	0.740–0.910	79.6	80.0	79.3	70.0	86.8
Overall TA	0.918	0.857–0.980	88.2	91.4	86.2	80.0	94.4
GLCM-correlation	0.744	0.645–0.843	72.0	60.0	81.0	64.5	75.8
GLRLM-LRE	0.809	0.721–0.897	77.4	74.3	79.3	68.4	83.6
GLSZM-ZE	0.777	0.683–0.870	69.9	85.7	60.3	56.6	87.5
IVIM-DWI with TA	0.975	0.950–1.000	92.5	91.4	93.1	88.9	94.7

IVIM-DWI: intravoxel incoherent motion diffusion weighted imaging; TA: texture analysis; ROC: receiver operating characteristic; AUC: area under curve; CI: confidence interval; D: slow diffusion coefficient; f: perfusion-related diffusion fraction; GLCM: gray level co-occurrence matrix; GLRLM: gray level run length matrix; GLSZM: gray level size zone matrix; LRE: long run emphasis; ZE: zone entro.