Detection of diabetes from whole-body MRI using deep learning.

Obesity is one of the main drivers of type 2 diabetes, but it is not uniformly associated with the disease. The location of fat accumulation is critical for metabolic health. Specific patterns of body fat distribution, such as visceral fat, are closely related to insulin resistance. There might be further, hitherto unknown, features of body fat distribution that could additionally contribute to the disease. We used machine learning with dense convolutional neural networks to detect diabetes-related variables from 2371 T1-weighted whole-body MRI data sets. MRI was performed in participants undergoing metabolic screening with oral glucose tolerance tests. Models were trained for sex, age, BMI, insulin sensitivity, HbA1c, and prediabetes or incident diabetes. The results were compared with those of conventional models. The area under the receiver operating characteristic curve was 87% for the type 2 diabetes discrimination and 68% for prediabetes, both superior to conventional models. Mean absolute regression errors were comparable to those of conventional models. Heatmaps showed that lower visceral abdominal regions were critical in diabetes classification. Subphenotyping revealed a group with high future diabetes and microalbuminuria risk.Our results show that diabetes is detectable from whole-body MRI without additional data. Our technique of heatmap visualization identifies plausible anatomical regions and highlights the leading role of fat accumulation in the lower abdomen in diabetes pathogenesis.

Quantitative microstructural studies of the armor of the marine threespine stickleback (Gasterosteus aculeatus).

In this study, a quantitative investigation of the microstructure and composition of field-caught marine Gasterosteus aculeatus (threespine stickleback) armor is presented, which provides useful phylogenetic information and insights into biomechanical function. Micro-computed tomography (microCT) was employed to create full three-dimensional images of the dorsal spines and basal plate, lateral plates, pelvic girdle and spines and to assess structural and compositional properties such as the spatial distribution of thickness (approximately 100-300 microm), the heterogeneous cross-sectional geometry (centrally thickened), plate-to-plate juncture and overlap (approximately 50% of the plate width), and bone mineral density (634-748 HA/cm(3)). The convolution of plate geometry in conjunction with plate-to-plate overlap allows a relatively constant armor thickness to be maintained throughout the assembly, promoting spatially homogeneous protection and thereby avoiding weakness at the armor unit interconnections. Plate-to-plate junctures act to register and join the plates while permitting compliance in sliding and rotation in selected directions. Mercury porosimetry was used to determine the pore size distribution and volume percent porosity of the lateral plates (20-35 vol.%) and spines (10-15 vol.%). SEM and microCT revealed a porous, sandwich-like cross-section beneficial for bending stiffness and strength at minimum weight. Back-scattered electron microscopy and energy dispersive X-ray analysis were utilized to quantify the weight percent mineral content (58-68%). Scanning electron microscopy and surface profilometry were used to characterize the interior and exterior surface topography (tubercles) of the lateral plates. The results obtained in this study are discussed in the context of mechanical function, performance, fitness, and survivability.