BACKGROUND: Evaluating longitudinal changes in gliomas is a time-intensive process with significant interrater variability. Automated segmentation could reduce interrater variability and increase workflow efficiency for assessment of treatment response. We sought to evaluate whether neural networks would be comparable to expert assessment of pre- and posttreatment diffuse gliomas tissue subregions including resection cavities. METHODS: A retrospective cohort of 647 MRIs of patients with diffuse gliomas (average 55.1 years; 29%/36%/34% female/male/unknown; 396 pretreatment and 251 posttreatment, median 237 days post-surgery) from 7 publicly available repositories in The Cancer Imaging Archive were split into training (536) and test/generalization (111) samples. T1, T1-post-contrast, T2, and FLAIR images were used as inputs into a 3D nnU-Net to predict 3 tumor subregions and resection cavities. We evaluated the performance of networks trained on pretreatment training cases (Pre-Rx network), posttreatment training cases (Post-Rx network), and both pre- and posttreatment cases (Combined networks). RESULTS: Segmentation performance was as good as or better than interrater reliability with median dice scores for main tumor subregions ranging from 0.82 to 0.94 and strong correlations between manually segmented and predicted total lesion volumes (0.94 < R 2 values < 0.98). The Combined network performed similarly to the Pre-Rx network on pretreatment cases and the Post-Rx network on posttreatment cases with fewer false positive resection cavities (7% vs 59%). CONCLUSIONS: Neural networks that accurately segment pre- and posttreatment diffuse gliomas have the potential to improve response assessment in clinical trials and reduce provider burden and errors in measurement.

Purpose

Restriction spectrum imaging (RSI) decomposes the diffusion-weighted MRI signal into separate components of known apparent diffusion coefficients (ADCs). The number of diffusion components and optimal ADCs for RSI are organ-specific and determined empirically. The purpose of this work was to determine the RSI model for breast tissues.

Methods

The diffusion-weighted MRI signal was described using a linear combination of multiple exponential components. A set of ADC values was estimated to fit voxels in cancer and control ROIs. Later, the signal contributions of each diffusion component were estimated using these fixed ADC values. Relative-fitting residuals and Bayesian information criterion were assessed. Contrast-to-noise ratio between cancer and fibroglandular tissue in RSI-derived signal contribution maps was compared to DCE imaging.

Results

A total of 74 women with breast cancer were scanned at 3.0 Tesla MRI. The fitting residuals of conventional ADC and Bayesian information criterion suggest that a 3-component model improves the characterization of the diffusion signal over a biexponential model. Estimated ADCs of triexponential model were D_1,3 = 0, D_2,3 = 1.5 × 10^-3 , and D_3,3 = 10.8 × 10^-3 mm² /s. The RSI-derived signal contributions of the slower diffusion components were larger in tumors than in fibroglandular tissues. Further, the contrast-to-noise and specificity at 80% sensitivity of DCE and a subset of RSI-derived maps were equivalent.

Conclusion

Breast diffusion-weighted MRI signal was best described using a triexponential model. Tumor conspicuity in breast RSI model is comparable to that of DCE without the use of exogenous contrast. These data may be used as differential features between healthy and malignant breast tissues.

Purpose

Diffusion-weighted MRI (DW-MRI) is a contrast-free modality that has demonstrated ability to discriminate between predefined benign and malignant breast lesions. However, how well DW-MRI discriminates cancer from all other breast tissue voxels in a clinical setting is unknown. Here we explore the voxelwise ability to distinguish cancer from healthy breast tissue using signal contributions from the newly developed three-component multi-b-value DW-MRI model.

Experimental design

Patients with pathology-proven breast cancer from two datasets (n = 81 and n = 25) underwent multi-b-value DW-MRI. The three-component signal contributions C ₁ and C ₂ and their product, C ₁ C ₂, and signal fractions F ₁, F ₂, and F ₁ F ₂ were compared with the image defined on maximum b-value (DWI _max), conventional apparent diffusion coefficient (ADC), and apparent diffusion kurtosis (K _app). The ability to discriminate between cancer and healthy breast tissue was assessed by the false-positive rate given a sensitivity of 80% (FPR₈₀) and ROC AUC.

Results

Mean FPR₈₀ for both datasets was 0.016 [95% confidence interval (CI), 0.008-0.024] for C ₁ C ₂, 0.136 (95% CI, 0.092-0.180) for C ₁, 0.068 (95% CI, 0.049-0.087) for C ₂, 0.462 (95% CI, 0.425-0.499) for F ₁ F ₂, 0.832 (95% CI, 0.797-0.868) for F ₁, 0.176 (95% CI, 0.150-0.203) for F ₂, 0.159 (95% CI, 0.114-0.204) for DWI _max, 0.731 (95% CI, 0.692-0.770) for ADC, and 0.684 (95% CI, 0.660-0.709) for K _app. Mean ROC AUC for C ₁ C ₂ was 0.984 (95% CI, 0.977-0.991).

Conclusions

The C ₁ C ₂ parameter of the three-component model yields a clinically useful discrimination between cancer and healthy breast tissue, superior to other DW-MRI methods and obliviating predefining lesions. This novel DW-MRI method may serve as noncontrast alternative to standard-of-care dynamic contrast-enhanced MRI.