Radiomics-based machine learning analysis and characterization of breast lesions with multiparametric diffusion-weighted MR

Table 1 Comparisons between radiomics and mean diffusion metrics

Maps	AUC of mean metrics	AUC of Radiomics	P value
ME
mADC_0–1000	0.79 (0.76–0.83)	0.83 (0.80–0.87)	0.002
mADC_all-b	0.77 (0.73–0.81)	0.84 (0.80–0.88)	0.001
BE_IVIM
mD	0.75 (0.72–0.79)	0.85 (0.81–0.89)	< 0.001
mD*	0.67 (0.63–0.71)	0.80 (0.74–0.83)	< 0.001
mf	0.54 (0.50–0.58)	0.82 (0.77–0.86)	< 0.001
SE
mDDC	0.77 (0.74–0.81)	0.81 (0.77–0.85)	0.030
mα	0.62 (0.58–0.66)	0.80 (0.77–0.84)	< 0.001
DKI
mD	0.78 (0.73–0.80)	0.83 (0.80–0.88)	0.005
mK	0.75 (0.71–0.80)	0.84 (0.80–0.88)	0.001

ME: mono-exponential model; mADC_0–1000: mean value of ADC_0–1000; mADC_{all b}: mean value of ADC_{all b}; BE-IVIM: biexponential intravoxel incoherent motion model; mD: mean value of true diffusion coefficient; mD*: mean value of pseudo-diffusion coefficient; mf: mean value of fractional perfusion; SE: stretched exponential model; mDDC: mean value of distributed diffusion coefficient; mα: mean value of low intravoxel diffusion heterogeneity; DKI: diffusion kurtosis imaging; mD: mean value of diffusivity coefficient; mK: mean value of kurtosis coefficient

ISSN: 1479-5876