Comparative performance of multiparametric MRI models for preoperative diagnosis of macrotrabecular-massive hepatocellular carcinoma: A pilot study of DWI, IVIM, and fractional order calculus techniques.
TL;DR
IVIM and FROC models were able to preoperatively diagnose MTM-HCC and were superior to DWI alone and MR imaging features alone.
Abstract
Objectives
To explore the feasibility of diffusion-weighted imaging (DWI), intravoxel incoherent motion (IVIM), and fractional order calculus (FROC) models in preoperatively predicting macrotrabecular-massive subtype of hepatocellular carcinoma (MTM-HCC).
Materials And Methods
101 pathologically confirmed HCC patients underwent multi-b-value DWI. Diffusion parameters including apparent diffusion coefficient (ADC), IVIM metrics (Dt, Dp, and f), and FROC metrics (D, β, and μ) were obtained from the tumor. The differences of diffusion parameters, MRI features-based diagnostic criteria between the MTM and non-MTM HCC groups were compared. Logistic regression was used to identify independent predictors and construct a diagnostic model based on D and μ. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis. Model calibration was assessed using a calibration curve based on 500 bootstrap resamples, and internal validation was performed using bootstrap resampling with 500 repetitions.
Results
30 MTM and 71 non-MTM-HCC patients were enrolled. The D, μ, Dt and ADC values of MTM-HCC were significantly lower than those of non-MTM-HCC (all P < 0.05). MTM-HCC was more frequently associated with arterial phase hypovascular components, satellite nodules and incomplete tumor capsule (P < 0.05). D, μ, Dt with area under the curve (AUC) values of 0.775,0.702,0.744 were higher than ADC, Rhee's criteria (AUC:0.688,0.610). The combination of D and μ demonstrated the highest AUC for identifying MTM-HCC (AUC: 0.885, sensitivity: 73.33%, specificity: 91.55%).The calibration curve showed good agreement between the predicted and observed probabilities. Bootstrap internal validation with 500 resamples yielded a mean AUC of 0.825 (95% CI: 0.741-0.909).
Conclusion
IVIM and FROC models were able to preoperatively diagnose MTM-HCC and were superior to DWI alone and MR imaging features alone.