Background: Computer-aided methods have been widely applied to diagnose lesions detected on breast MRI, but fullyautomatic diagnosis using deep learning is rarely reported. Purpose: To evaluate the diagnostic accuracy of mass lesions using region of interest (ROI)-based, radiomics and deeplearning methods, by taking peritumor tissues into consideration. Study Type: Retrospective. Population: In all, 133 patients with histologically confirmed 91 malignant and 62 benign mass lesions for training (74 patients with 48 malignant and 26 benign lesions for testing). Field Strength/Sequence: 3T, using the volume imaging for breast assessment (VIBRANT) dynamic contrast-enhanced (DCE) sequence. Assessment: 3D tumor segmentation was done automatically by using fuzzy-C-means algorithm with connectedcomponent labeling. A total of 99 texture and histogram parameters were calculated for each case, and 15 were selected using random forest to build a radiomics model. Deep learning was implemented using ResNet50, evaluated with 10-fold crossvalidation. The tumor alone, smallest bounding box, and 1.2, 1.5, 2.0 times enlarged boxes were used as inputs. Statistical Tests: The malignancy probability was calculated using each model, and the threshold of 0.5 was used to make a diagnosis. Results: In the training dataset, the diagnostic accuracy was 76% using three ROI-based parameters, 84% using the radiomics model, and 86% using ROI + radiomics model. In deep learning using the per-slice basis, the area under the receiver operating characteristic (ROC) was comparable for tumor alone, smallest and 1.2 times box (AUC = 0.97-0.99), which were significantly higher than 1.5 and 2.0 times box (AUC = 0.86 and 0.71, respectively). For per-lesion diagnosis, the highest accuracy of 91% was achieved when using the smallest bounding box, and that decreased to 84% for tumor alone and 1.2 times box, and further to 73% for 1.5 times box and 69% for 2.0 times box. In the independent testing dataset, the perlesion diagnostic accuracy was also the highest when using the smallest bounding box, 89%. Data Conclusion: Deep learning using ResNet50 achieved a high diagnostic accuracy. Using the smallest bounding box containing proximal peritumor tissue as input had higher accuracy compared to using tumor alone or larger boxes. Level of Evidence: 3 Technical Efficacy: Stage 2

No abstract

The Chinese Imaging Genetics (CHIMGEN) study establishes the largest Chinese neuroimaging genetics cohort and aims to identify genetic and environmental factors and their interactions that are associated with neuroimaging and behavioral phenotypes. This study prospectively collected genomic, neuroimaging, environmental, and behavioral data from more than 7000 healthy Chinese Han participants aged 18-30 years. As a pioneer of large-sample neuroimaging genetics cohorts of non-Caucasian populations, this cohort can provide new insights into ethnic differences in genetic-neuroimaging associations by being compared with Caucasian cohorts. In addition to micro-environmental measurements, this study also collects hundreds of quantitative macro-environmental measurements from remote sensing and national survey databases based on the locations of each participant from birth to present, which will facilitate discoveries of new environmental factors associated with neuroimaging phenotypes. With lifespan environmental measurements, this study can also provide insights on the macro-environmental exposures that affect the human brain as well as their timing and mechanisms of action.

Attention deficit hyperactivity disorder (ADHD) is a common childhood neuropsychiatric disorder that has been linked to the dopaminergic system. This study aimed to investigate the effects of regulation of the dopamine D4 receptor (DRD4) on functional brain activity during the resting state in ADHD children using the methods of regional homogeneity (ReHo) and functional connectivity (FC). Resting-state functional magnetic resonance imaging data were analyzed in 49 children with ADHD. All participants were classified as either carriers of the DRD4 4-repeat/4-repeat (4R/4R) allele (n = 30) or the DRD4 2-repeat (2R) allele (n = 19). The results showed that participants with the DRD4 2R allele had decreased ReHo bilaterally in the posterior lobes of the cerebellum, while ReHo was increased in the left angular gyrus. Compared with participants carrying the DRD4 4R/4R allele, those with the DRD4 2R allele showed decreased FC to the left angular gyrus in the left striatum, right inferior frontal gyrus, and bilateral lobes of the cerebellum. The increased FC regions included the left superior frontal gyrus, medial frontal gyrus, and rectus gyrus. These data suggest that the DRD4 polymorphisms are associated with localized brain activity and specific functional connections, including abnormality in the frontal-striatal-cerebellar loop. Our study not only enhances the understanding of the correlation between the cerebellar lobes and ADHD, but also provides an imaging basis for explaining the neural mechanisms underlying ADHD in children.

Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disease featuring executive control deficits as a prominent neuropsychological trait. Executive functions are implicated in multiple sub-networks of the brain; however, few studies examine these sub-networks as a whole in ADHD. By combining resting-state functional MRI and graph-based approaches, we systematically investigated functional connectivity patterns among four control-related networks, including the frontoparietal network (FPN), cingulo-opercular network, cerebellar network, and default mode network (DMN), in 46 drug-naive children with ADHD and 31 age-, gender-, and intelligence quotient-matched healthy controls (HCs). Compared to the HCs, the ADHD children showed significantly decreased functional connectivity that primarily involved the DMN and FPN regions and cross-network long-range connections. Further graph-based network analysis revealed that the ADHD children had fewer connections, lower network efficiency, and more functional modules compared with the HCs. The ADHD-related alterations in functional connectivity but not topological organization were correlated with clinical symptoms of the ADHD children and differentiated the patients from the HCs with a good performance. Taken together, our findings suggest a less-integrated functional brain network in children with ADHD due to selective disruption of key long-range connections, with important implications for understanding the neural substrates of ADHD, particularly executive dysfunction.