We address the problem of image upscaling in the form of single image super-resolution based on a dictionary of low-and highresolution exemplars. Two recently proposed methods, Anchored Neighborhood Regression (ANR) and Simple Functions (SF), provide state-ofthe-art quality performance. Moreover, ANR is among the fastest known super-resolution methods. ANR learns sparse dictionaries and regressors anchored to the dictionary atoms. SF relies on clusters and corresponding learned functions. We propose A+, an improved variant of ANR, which combines the best qualities of ANR and SF. A+ builds on the features and anchored regressors from ANR but instead of learning the regressors on the dictionary it uses the full training material, similar to SF. We validate our method on standard images and compare with state-of-the-art methods. We obtain improved quality (i.e. 0.2-0.7dB PSNR better than ANR) and excellent time complexity, rendering A+ the most efficient dictionary-based super-resolution method to date.

No abstract

Activated hepatic stellate cells (aHSC) are the main source of extra cellular matrix in liver fibrosis. Activation is classically divided in two phases: initiation and perpetuation. Currently, HSC-based therapeutic candidates largely focus on targeting the aHSCs in the perpetuation phase. However, the importance of HSC initiation during chronic liver disease (CLD) remains unclear. Here, we identified transcriptional programs of initiating and activated HSCs by RNA sequencing, using in vitro and in vivo mouse models of fibrosis. Importantly, we show that both programs are active in HSCs during murine and human CLD. In human cirrhotic livers, scar associated mesenchymal cells employ both transcriptional programs at the single cell level. Our results indicate that the transcriptional programs that drive the initiation of HSCs are still active in humans suffering from CLD. We conclude that molecules involved in the initiation of HSC activation, or in the maintenance of aHSCs can be considered equally important in the search for druggable targets of chronic liver disease.

Liver sinusoidal endothelial cells have a gatekeeper function in liver homeostasis by permitting substrates from the bloodstream into the space of Disse and regulating hepatic stellate cell activation status. Maintenance of LSEC's highly specialized phenotype is crucial for liver homeostasis. During liver fibrosis and cirrhosis, LSEC phenotype and functions are lost by processes known as capillarization and LSEC dysfunction. LSEC capillarization can be demonstrated by the loss of fenestrae (cytoplasmic pores) and the manifestation of a basement membrane. Currently, no protein or genetic markers can clearly distinguish healthy from damaged LSECs in acute or chronic liver disease. Single cell (sc)RNA sequencing efforts have identified several LSEC populations in mouse models for liver disease and in human cirrhotic livers. Still, there are no clearly defined genesets that can identify LSECs or dysfunctional LSEC populations in transcriptome data. Here, we developed genesets that are enriched in healthy and damaged LSECs which correlated very strongly with healthy and early stage- vs. advanced human liver diseases. A damaged LSEC signature comprised of Fabp4/5 and Vwf/a1 was established which could efficiently identify damaged endothelial cells in single cell RNAseq data sets. In LSECs from an acute CCl4 liver injury mouse model, Fabp4/5 and Vwf/a1 expression is induced within 1–3 days while in cirrhotic human livers these 4 genes are highly enriched in damaged LSECs. In conclusion, our newly developed gene signature of damaged LSECs can be applicable to a wide range of liver disease etiologies, implicating a common transcriptional alteration mechanism in LSEC damage.

Recent algorithms for exemplar-based single image super-resolution have shown impressive results, mainly due to well-chosen priors and recently also due to more accurate blur kernels. Some methods exploit clustering of patches, local gradients or some context information. However, to the best of our knowledge, there is no literature studying the benefits of using semantic information at the image level. By semantic information we mean image segments with corresponding categorical labels. In this paper we investigate the use of semantic information in conjunction with A+, a state-of-the-art super-resolution method. We conduct experiments on large standard datasets of natural images with semantic annotations, and discuss the benefits vs. the drawbacks of using semantic information. Experimental results show that our semantic driven super-resolution can significantly improve over the original settings.