Microglia are brain-resident macrophages, which have specialized functions important in brain development and in disease. They colonize the brain in early embryonic stages, but few factors that drive the migration of yolk sac macrophages (YSMs) into the embryonic brain, or regulate their acquisition of specialized properties, are currently known. Here, we present a CRISPR/Cas9-based in vivo reverse genetic screening pipeline to identify new microglia regulators using zebrafish. Zebrafish larvae are particularly suitable due to their external development, transparency and conserved microglia features. We targeted putative microglia regulators, by Cas9/gRNA complex injections, followed by Neutral-Red-based visualization of microglia. Microglia were quantified automatically in 3-day-old larvae using a software tool we called SpotNGlia. We identified that loss of zebrafish colony-stimulating factor 1 receptor (Csf1r) ligand, Il34, caused reduced microglia numbers. Previous studies on the role of IL34 in microglia development in vivo were ambiguous. Our data, and a concurrent paper, show that, in zebrafish, il34 is required during the earliest seeding of the brain by microglia. Our data also indicate that Il34 is required for YSM distribution to other organs. Disruption of the other Csf1r ligand, Csf1, did not reduce microglia numbers in mutants, whereas overexpression increased the number of microglia. This shows that Csf1 can influence microglia numbers, but might not be essential for the early seeding of the brain. In all, we identified il34 as a modifier of microglia colonization, by affecting distribution of YSMs to target organs, validating our reverse genetic screening pipeline in zebrafish.@en

Diagnosis of patients suffering from small-fiber neuropathy is a challenging task and requires accurate measurement of the density of nerve fibers crossing the dermal-epidermal junction in the skin. Currently this is typically done by expert manual counting in microscopy images of sliced and stained skin biopsies. It is a rather subjective and labor-intensive process that would benefit greatly from more automated approaches. Previously we have explored classical image processing methods for this, with very limited success. Here we explore the potential of convolutional neural networks and deep learning for the task. The results of preliminary experiments show the networks perform close to the expert and outperform novices and our previous method.@en

Sparse representations classification (SRC) is a powerful technique for pixelwise classification of images and it is increasingly being used for a wide variety of image analysis tasks. The method uses sparse representation and learned redundant dictionaries to classify image pixels. In this empirical study we propose to further leverage the redundancy of the learned dictionaries to achieve a more accurate classifier. In conventional SRC, each image pixel is associated with a small patch surrounding it. Using these patches, a dictionary is trained for each class in a supervised fashion. Commonly, redundant/overcomplete dictionaries are trained and image patches are sparsely represented by a linear combination of only a few of the dictionary elements. Given a set of trained dictionaries, a new patch is sparse coded using each of them, and subsequently assigned to the class whose dictionary yields the minimum residual energy.We propose a generalization of this scheme. The method, which we call multiple sparse representations classification (mSRC), is based on the observation that an overcomplete, class specific dictionary is capable of generating multiple accurate and independent estimates of a patch belonging to the class. So instead of finding a single sparse representation of a patch for each dictionary, we find multiple, and the corresponding residual energies provides an enhanced statistic which is used to improve classification. We demonstrate the efficacy of mSRC for three example applications: pixelwise classification of texture images, lumen segmentation in carotid artery magnetic resonance imaging (MRI), and bifurcation point detection in carotid artery MRI. We compare our method with conventional SRC, K-nearest neighbor, and support vector machine classifiers. The results show that mSRC outperforms SRC and the other reference methods. In addition, we present an extensive evaluation of the effect of the main mSRC parameters: patch size, dictionary size, and sparsity level.@en