Monoexponential, biexponential, stretched exponential and diffusion kurtosis models of diffusion-weighted imaging: a...

Diffusion-weighted imaging (DWI) can be used for quantitative tumor assessment. DWI with different models may show different aspects of tissue characteristics.

To investigate the diagnostic performance of parameters derived from monoexponential, biexponential, stretched exponential magnetic resonance diffusion weighted imaging (DWI) and diffusion kurtosis imaging (DKI) in differentiating benign from malignant solitary pulmonary lesions (SPLs).

Forty-four SPL subjects were selected according to the inclusion criteria. All patients underwent conventional and multi‑b DWI sequences. Monoexponential DWI and DKI model were fitted using least square method. Levenberg-Marquardt nonlinear fitting biexponential and stretched exponential DWI. Region of interests (ROIs) were described manually. Parameters between benign and malignant SPLs were compared using independent sample t test or the Mann-Whitney U test. Receiver operating characteristic (ROC) curves analysis was used to investigate the diagnostic performance of different DWI parameters. Correlation between all parameters were evaluated by using Spearman correlation.

ADC, ADC_slow, α, DDC and D_app values were significantly lower in malignant SPL than in benign SPL (P < 0.001). K_app was significantly higher in malignant SPL than in benign SPL (P < 0.001). Among all subjects, ADC_slow was significantly lower than ADC (P < 0.05), while DDC and D_app were significantly higher than ADC (P < 0.05). When observing the ROC curves for distinguishing benign and malignant SPL, the AUC values of ADC, ADC_slow, DDC, D_app, and K_app were 0.904, 0.815, 0.942, 0.93, and 0.815, respectively. The DDC value has the highest area under ROC curve value. DeLong analysis showed no statistically significant difference in the area under ADC, DDC, and D_app curves. There were strong correlations among ADC, ADC_slow, ADC_fast, f, α, DDC, D_app, and K_app (P < 0.001).

Multi‑b DWI is a promising method for differentiating benign from malignant SPLs with high diagnostic accuracy. In addition, the DDC derived from stretched‑exponential model is the most promising DWI parameter for the differentiation of benign and malignant SPLs.

This study was a clinical trail study, with study protocol published at ClinicalTrails. Retrospectively registered number ChiCTR2300074258, date of registration 02/08/2023.