Search

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is widely used to evaluate tumor permeability, yet measurements have not been directly validated in brain tumors. Our purpose was to compare estimates of forward leakage Ktrans derived from DCE-MRI to the estimates K obtained using [14C]aminoisobutyric acid quantitative autoradiography ([14C]AIB QAR), an established method of evaluating blood--tumor barrier permeability. Both DCE-MRI and [14C]AIB QAR were performed in five rats 9 to 11 days following tumor implantation. Ktrans in the tumor was estimated from DCE-MRI using the threeparameter general kinetic model and a measured vascular input function. Ki was estimated from QAR data using regions of interest (ROI) closely corresponding to those used to estimate Ktrans. Ktrans and Ki correlated with each other for two independent sets of central tumor ROI (R = 0.905, P = .035; R = 0.933, P = .021). In an additional six rats, Ktrans was estimated on two occasions to show reproducibility (intraclass coefficient = 0.9993; coefficient of variance = 6.07%). In vivo blood-tumor permeability parameters derived from DCE-MRI are reproducible and correlate with the gold standard for quantifying blood tumor barrier permeability, [14C]AIB QAR. [ABSTRACT FROM AUTHOR]

• We propose a novel cross-modality medical image segmentation method. • Our model can perform segmentation for a target domain without labeled training data. • A diverse data augmentation approach is used to augment the training data for segmentation. • Experiments in two different tasks demonstrate the effectiveness of proposed method. [Display omitted] The dearth of annotated data is a major hurdle in building reliable image segmentation models. Manual annotation of medical images is tedious, time-consuming, and significantly variable across imaging modalities. The need for annotation can be ameliorated by leveraging an annotation-rich source modality in learning a segmentation model for an annotation-poor target modality. In this paper, we introduce a diverse data augmentation generative adversarial network (DDA-GAN) to train a segmentation model for an unannotated target image domain by borrowing information from an annotated source image domain. This is achieved by generating diverse augmented data for the target domain by one-to-many source-to-target translation. The DDA-GAN uses unpaired images from the source and target domains and is an end-to-end convolutional neural network that (i) explicitly disentangles domain-invariant structural features related to segmentation from domain-specific appearance features, (ii) combines structural features from the source domain with appearance features randomly sampled from the target domain for data augmentation, and (iii) train the segmentation model with the augmented data in the target domain and the annotations from the source domain. The effectiveness of our method is demonstrated both qualitatively and quantitatively in comparison with the state of the art for segmentation of craniomaxillofacial bony structures via MRI and cardiac substructures via CT. [ABSTRACT FROM AUTHOR]