Moving Toward a Consensus DSC-MRI Protocol: Validation of a Low-Flip Angle Single-Dose Option as a Reference Standard for Brain Tumors.

BACKGROUND AND PURPOSE: DSC-MR imaging using preload, intermediate (60°) flip angle and postprocessing leakage correction has gained traction as a standard methodology. Simulations suggest that DSC-MR imaging with flip angle = 30° and no preload yields relative CBV practically equivalent to the reference standard. This study tested this hypothesis in vivo. MATERIALS AND METHODS: Eighty-four patients with brain lesions were enrolled in this 3-institution study. Forty-three patients satisfied the inclusion criteria. DSC-MR imaging (3T, single-dose gadobutrol, gradient recalled-echo-EPI, TE = 20-35 ms, TR = 1.2-1.63 seconds) was performed twice for each patient, with flip angle = 30°-35° and no preload (P-), which provided preload (P+) for the subsequent intermediate flip angle = 60°. Normalized relative CBV and standardized relative CBV maps were generated, including postprocessing with contrast agent leakage correction (C+) and without (C-) contrast agent leakage correction. Contrast-enhancing lesion volume, mean relative CBV, and contrast-to-noise ratio obtained with 30°/P-/C-, 30°/P-/C+, and 60°/P+/C- were compared with 60°/P+/C+ using the Lin concordance correlation coefficient and Bland-Altman analysis. Equivalence between the 30°/P-/C+ and 60°/P+/C+ protocols and the temporal SNR for the 30°/P- and 60°/P+ DSC-MR imaging data was also determined. RESULTS: Compared with 60°/P+/C+, 30°/P-/C+ had closest mean standardized relative CBV (P = .61), highest Lin concordance correlation coefficient (0.96), and lowest Bland-Altman bias (µ = 1.89), compared with 30°/P-/C- (P = .02, Lin concordance correlation coefficient = 0.59, µ = 14.6) and 60°/P+/C- (P = .03, Lin concordance correlation coefficient = 0.88, µ = -10.1) with no statistical difference in contrast-to-noise ratios across protocols. The normalized relative CBV and standardized relative CBV were statistically equivalent at the 10% level using either the 30°/P-/C+ or 60°/P+/C+ protocols. Temporal SNR was not significantly different for 30°/P- and 60°/P+ (P = .06). CONCLUSIONS: Tumor relative CBV derived from low-flip angle, no-preload DSC-MR imaging with leakage correction is an attractive single-dose alternative to the higher dose reference standard.

Accurate Patient-Specific Machine Learning Models of Glioblastoma Invasion Using Transfer Learning.

BACKGROUND AND PURPOSE: MR imaging-based modeling of tumor cell density can substantially improve targeted treatment of glioblastoma. Unfortunately, interpatient variability limits the predictive ability of many modeling approaches. We present a transfer learning method that generates individualized patient models, grounded in the wealth of population data, while also detecting and adjusting for interpatient variabilities based on each patient's own histologic data. MATERIALS AND METHODS: We recruited patients with primary glioblastoma undergoing image-guided biopsies and preoperative imaging, including contrast-enhanced MR imaging, dynamic susceptibility contrast MR imaging, and diffusion tensor imaging. We calculated relative cerebral blood volume from DSC-MR imaging and mean diffusivity and fractional anisotropy from DTI. Following image coregistration, we assessed tumor cell density for each biopsy and identified corresponding localized MR imaging measurements. We then explored a range of univariate and multivariate predictive models of tumor cell density based on MR imaging measurements in a generalized one-model-fits-all approach. We then implemented both univariate and multivariate individualized transfer learning predictive models, which harness the available population-level data but allow individual variability in their predictions. Finally, we compared Pearson correlation coefficients and mean absolute error between the individualized transfer learning and generalized one-model-fits-all models. RESULTS: Tumor cell density significantly correlated with relative CBV (r = 0.33, P < .001), and T1-weighted postcontrast (r = 0.36, P < .001) on univariate analysis after correcting for multiple comparisons. With single-variable modeling (using relative CBV), transfer learning increased predictive performance (r = 0.53, mean absolute error = 15.19%) compared with one-model-fits-all (r = 0.27, mean absolute error = 17.79%). With multivariate modeling, transfer learning further improved performance (r = 0.88, mean absolute error = 5.66%) compared with one-model-fits-all (r = 0.39, mean absolute error = 16.55%). CONCLUSIONS: Transfer learning significantly improves predictive modeling performance for quantifying tumor cell density in glioblastoma.