We propose a deep learning approach for user-guided image colorization. The system directly maps a grayscale image, along with sparse, local user "hints" to an output colorization with a Convolutional Neural Network (CNN). Rather than using hand-defined rules, the network propagates user edits by fusing low-level cues along with high-level semantic information, learned from large-scale data. We train on a million images, with simulated user inputs. To guide the user towards efficient input selection, the system recommends likely colors based on the input image and current user inputs. The colorization is performed in a single feed-forward pass, enabling real-time use. Even with randomly simulated user inputs, we show that the proposed system helps novice users quickly create realistic colorizations, and offers large improvements in colorization quality with just a minute of use. In addition, we demonstrate that the framework can incorporate other user "hints" to the desired colorization, showing an application to color histogram transfer. Our code and models are available at https://richzhang.github.io/ideepcolor.

A murine segmental femoral bone graft model was used to show the essential role of donor periosteal progenitor cells in bone graft healing. Transplantation of live bone graft harvested from Rosa 26A mice showed that ∼70% of osteogenesis on the graft was attributed to the expansion and differentiation of donor periosteal progenitor cells. Furthermore, engraftment of BMP-2-producing bone marrow stromal cells on nonvital allografts showed marked increases in cortical graft incorporation and neovascularization, suggesting that gene-enhanced, tissue engineered functional periosteum may improve allograft incorporation and repair.Introduction: The loss of cellular activity in a structural bone allograft markedly reduces its healing potential compared with a live autograft. To further understand the cellular mechanisms for structural bone graft healing and repair and to devise a therapeutic strategy aimed at enhancing the performance of allograft, we established a segmental femoral structural bone graft model in mice that permits qualitative and quantitative analyses of graft healing and neovascularization. Materials and Methods: Using this segmental femoral bone graft model, we transplanted live isografts harvested from Rosa 26A mice that constitutively express ␤-galactosidase into their wildtype control mice. In an attempt to emulate the osteogenic and angiogenic properties of periosteum, we applied a cell-based, adenovirus-mediated gene therapy approach to engraft BMP-2-producing bone marrow stromal cells onto devitalized allografts. Results: X-gal staining for donor cells allowed monitoring the progression of periosteal progenitor cell fate and showed that 70% of osteogenesis was attributed to cellular proliferation and differentiation of donor progenitor cells on the surface of the live bone graft. Quantitative CT analyses showed a 3-fold increase in new bone callus formation and a 6.8-fold increase in neovascularization for BMP-2/stromal cell-treated allograft compared with control acellular allografts. Histologic analyses showed the key features of autograft healing in the BMP-2/stromal cell-treated allografts, including the formation of a mineralized bone callus completely bridging the segmental defects, abundant neovascularization, and extensive resorption of bone graft. Conclusions:The marked improvement of healing in these cellularized allografts suggests a clinical strategy for engineering a functional periosteum to improve the osteogenic and angiogenic properties of processed allografts.

No abstract

Purpose In this work, we tested the hypothesis that microneedles provide a minimally invasive method to inject particles into the suprachoroidal space for drug delivery to the back of the eye. Methods A single, hollow microneedle was inserted into the sclera, and infused nanoparticle and microparticle suspensions into the suprachoroidal space. Experiments were performed on whole rabbit, pig, and human eyes ex vivo. Particle delivery was imaged using brightfield and fluorescence microscopy as well as microcomputed tomography. Results Microneedles were shown to deliver sulforhodamine B as well as nanoparticle and microparticle suspensions into the suprachoroidal space of rabbit, pig, and human eyes. Volumes up to 35 μL were administered consistently. Optimization of the delivery device parameters showed that microneedle length, pressure, and particle size played an important role in determining successful delivery into the suprachoroidal space. Needle lengths of 800–1,000 μm and applied pressures of 250–300 kPa provided most reliable delivery. Conclusions Microneedles were shown for the first time to deliver nanoparticle and microparticle suspensions into the suprachoroidal space of rabbit, pig and human eyes. This shows that microneedles may provide a minimally invasive method for controlled drug delivery to the back of the eye.

Non-healing fractures can result from trauma, disease, or age-related bone loss. While many treatments focus on restoring bone volume, few try to recapitulate bone organization. However, the native architecture of bone is optimized to provide its necessary mechanical properties. Hyaluronic acid (HA) hydrogel scaffold systems with tunable degradation properties were developed for the controlled delivery of osteoinductive and angiogenic growth factors, thus affecting the quantity and quality of regenerated tissue. HA hydrogels were designed to degrade at fast, intermediate, and slow rates due to hydrolysis and further provided controlled release of cationic proteins due to electrostatic interactions. Scaffolds delivering bone morphogenetic protein-2 (BMP-2) were evaluated in a rat calvarial bone critical size defect model. BMP-2 delivery from the HA hydrogels had a clear osteoinductive effect in vivo and, for all hydrogel types, BMP-2 delivery resulted in significant mineralization compared to control hydrogels. The temporal progression of this effect could be modulated by altering the degradation rate of the scaffold. All three degradation rates tested resulted in similar amounts of mineral formation at the latest (six week) time point examined. Interestingly, however, the fastest and slowest degrading scaffolds seemed to result in more organized bone than the intermediate degrading scaffold, which was designed to degrade in 6-8 weeks to match the healing time. Additionally, healing could be enhanced by co-delivery of vascular endothelial growth factor along with BMP-2.

Summary Objective The objective of the present study was to validate the ability of EPIC-μCT to nondestructively assess cartilage morphology in the rat model. Design An appropriate contrast agent (Hexabrix) concentration and incubation time for equilibration were determined for reproducible segmentation of femoral articular cartilage from contrast-enhanced μCT scans. Reproducibility was evaluated by triplicate scans of six femora, and the measured articular cartilage thickness was independently compared to thickness determined from needle probe testing and histology. The validated technique was then applied to quantify age-related differences in articular cartilage morphology between 4, 8, and 16-week old (n=5 each) male Wistar rats. Results A 40% Hexabrix/60% PBS solution with 30 minute incubation was optimal for segmenting cartilage from the underlying bone tissue and other soft tissues in the rat model. High reproducibility was indicated by the low coefficient of variation (1.7-2.5%) in cartilage volume, thickness and surface area. EPIC-μCT evaluation of thickness showed a strong linear relationship and good agreement with both needle probing (r2=0.95, slope=0.81, p<0.01, mean difference 11±22μm, n=43) and histology (r2=0.99, slope=0.97, p<0.01, mean difference 12±10μm, n=30). Cartilage volume and thickness significantly decreased with age while surface area significantly increased. Conclusion EPIC-μCT imaging has the ability to nondestructively evaluate three-dimensional articular cartilage morphology with high precision and accuracy in a small animal model.

Estrogen (E) deficiency leads to an expansion of the pool of tumor necrosis factor (TNF)-producing T cells through an IFN-␥-dependent pathway that results in increased levels of the osteoclastogenic cytokine TNF in the bone marrow. Disregulated IFN-␥ production is instrumental for the bone loss induced by ovariectomy (ovx), but the responsible mechanism is unknown. We now show that mice with T cell-specific blockade of type ␤ transforming growth factor (TGF␤) signaling are completely insensitive to the bone-sparing effect of E. This phenotype results from a failure of E to repress IFN-␥ production, which, in turn, leads to increased T cell activation and T cell TNF production. Furthermore, ovx blunts TGF␤ levels in the bone marrow, and overexpression of TGF␤ in vivo prevents ovx-induced bone loss. These findings demonstrate that E prevents bone loss through a TGF␤-dependent mechanism, and that TGF␤ signaling in T cells preserves bone homeostasis by blunting T cell activation. Thus, stimulation of TGF␤ production in the bone marrow is a critical ''upstream'' mechanism by which E prevents bone loss, and enhancement of TGF␤ levels in vivo may constitute a previously undescribed therapeutic approach for preventing bone loss.A lthough multiple genotropic and nongenotropic effects contribute to explain how estrogen (E) controls bone remodeling (1, 2), most of the bone-sparing activity exerted by E occurs through modulation of bone cell life span and decreased cytokine-driven osteoclastogenesis (3, 4). Among the factors that up-regulate osteoclast (OC) formation and lead to bone loss in estroprevic humans and rodents is tumor necrosis factor (TNF) ␣ (5). This E-regulated cytokine promotes osteoclastogenesis by augmenting the production of receptor activator of nuclear factor-B ligand (RANKL) (1), the nonredundant cytokine responsible for OC development (6) and by increasing the responsiveness of maturing OCs to this factor (7-9). Furthermore, TNF stimulates the production of other cytokines known to be implicated in the pathogenesis of ovariectomy (ovx)-induced bone loss, such as IL-1, IL-6, IL-7, and macrophage colony-stimulated factor (1, 10).ovx increases TNF levels in the bone marrow (BM) via an expansion of the pool of TNF-producing T cells (7, 11) induced by a complex mechanism driven by IFN-␥ (12). This cytokine augments antigen (Ag) presentation by enhancing MHCII expression on BM macrophages (BMMs), through induction of class II transactivator (CIITA) expression (13). Up-regulation of Ag presentation results, in turn, in increased T cell activation. Thus, upregulation of IFN-␥ production induced by ovx leads to increased T cell proliferation and life span, a phenomenon that results in an increase in both the total number of T cells and the pool of TNF-producing T cells (12). T cell-produced TNF plays a pivotal role in the mechanism of ovx-induced bone loss, as demonstrated by the failure of ovx to induce bone loss in T cell-deficient nude mice and by the ability to reconstitute with WT T cells, but not TNFϪ͞Ϫ ...

No abstract