A customized deep learning tool is accurate in the detection and quantification of hemorrhage on NCCT. Demonstrated high performance on prospective NCCTs ordered from the emergency department suggests the clinical viability of the proposed deep learning tool.

Purpose:To investigate the association between parameters obtained from dynamic contrast enhanced MRI (DCE-MRI) of breast cancer using different analysis approaches, as well as their correlation with angiogenesis biomarkers (vascular endothelial growth factor and vessel density).Materials and Methods: DCE-MRI results were obtained from 105 patients with breast cancer (108 lesions). Three analysis methods were applied: 1) whole tumor analysis, 2) regional hot-spot analysis, and 3) intratumor pixel-by-pixel analysis. Early enhancement intensities and fitted pharmacokinetic parameters were studied. Paraffin blocks of 71 surgically resected specimens were analyzed by immunohistochemical staining to measure microvessel counts (with CD31) and vascular endothelial growth factor (VEGF) expression levels.Results: MRI parameters obtained from the three analysis methods showed significant correlations (P Ͻ 0.0001), but a substantial dispersion from the linear regression line was noted (r ϭ 0.72-0.97). The entire region of interest (ROI) vs. pixel population analyses had a significantly higher association compared to the entire ROI vs. hot-spot analyses. Cancer specimens with high VEGF expression had significantly higher CD31 microvessel densities than did specimens with low VEGF levels (P Ͻ 0.005). No significant association was found between MRI parameters obtained from the three analysis strategies and IHC based measurements of angiogenesis. Conclusion:A consistent analysis strategy was important in the DCE-MRI study. In this series, none of these strategies yielded results for MRI based quantitation of tumor vascularity that were associated with IHC based measurements. Therefore, different analyses could not account for the lack of association.

No abstract