Intraoperative diagnosis is essential for providing safe and effective care during cancer surgery 1. The existing workflow for intraoperative diagnosis based on hematoxylin and eosin-staining of processed tissue is time-, resource-, and labor-intensive 2,3. Moreover, interpretation of intraoperative histologic images is dependent on a contracting, unevenly distributed pathology workforce 4. Here, we report a parallel workflow that combines stimulated Raman histology (SRH) 5-7 , a label-free optical imaging method, and deep convolutional neural networks (CNN) to predict diagnosis at the bedside in near real-time in an automated fashion. Specifically, our CNN, trained on over 2.5 million SRH images, predicts brain tumor diagnosis in the operating room in under 150 seconds, an order of magnitude faster than conventional techniques (e.g., 20-30 minutes) 2. In a multicenter, prospective clinical trial (n = 278) we demonstrated that CNN-based diagnosis of SRH images was non-inferior to pathologist-based interpretation of conventional histologic images (overall accuracy, 94.6% vs. 93.9%). Our CNN learned a hierarchy of recognizable histologic feature representations to classify the major histopathologic classes of brain tumors. Additionally, we implemented a semantic segmentation method to identify tumor infiltrated, diagnostic regions within SRH images. These results demonstrate how intraoperative cancer diagnosis can be streamlined, creating a complimentary pathway for tissue diagnosis that is independent of a traditional pathology laboratory.

In the developing pancreas, the basic helix-loop-helix (bHLH) protein Neurogenin3 (Ngn3) specifies which precursor cells ultimately will become endocrine cells and initiates the islet differentiation program. NeuroD1, a closely related bHLH protein and a downstream target of Ngn3, maintains the differentiation program initiated by Ngn3. We have developed an in vitro model of Ngn3-dependent differentiation by infecting pancreatic duct cell lines with an Ngn3-expressing adenovirus. We found that both Ngn3 and its downstream target NeuroD1 activated the islet differentiation program in these cells by inducing the expression of genes with early roles in the differentiation cascade, as well as genes characteristic of fully differentiated islet cells. Induction of these genes, as exemplified by the insulin1 gene, involved alteration of the local chromatin structure. Interestingly, the subsets of genes activated by Ngn3 and NeuroD1 were not completely overlapping, indicating that these two bHLH proteins serve specific functions in the development of the endocrine pancreas. In addition, microarray gene expression analysis identified a previously uncharacterized group of Ngn3-induced genes with potentially important roles in islet development and function. These studies demonstrate how Ngn3 initiates islet differentiation and provide us with a model for testing methods for producing islet cells for people with diabetes. During pancreatic development, differentiation of endocrine and exocrine cells from a common endodermal progenitor cell requires the coordinated regulation of specific sets of genes. This process can be envisioned as a hierarchy or cascade of transcription factors that initiate and maintain the distinct gene expression programs that define the various pancreatic cell types (1). Among these factors, the basic helix-loop-helix (bHLH) protein Neurogenin3 (Ngn3) plays a dominant role in the specification of the endocrine͞islet cell lineage.During embryonic development, Ngn3 appears transiently in scattered pancreatic epithelial cells (2, 3). Several lines of evidence indicate that the expression of Ngn3 in these undifferentiated cells directs them to an endocrine cell fate and initiates the program of islet differentiation. First, lineage tracing shows that these transient Ngn3-expressing cells differentiate exclusively into islet cells (4). Second, mice homozygous for a targeted deletion of the ngn3 gene fail to generate any islet cells (5). Third, ectopic expression of Ngn3 drives embryonic endoderm to an endocrine fate (2, 3, 6).Ngn3 may play a similar role in the generation of new islet cells postnatally. It has been suggested that cells along the pancreatic ducts may act as progenitors for new islet cells in the postnatal period, although recent lineage tracing experiments suggest that the bulk of newly generated beta cells in adult mice result from the replication of preexisting beta cells (7).Ngn3 initiates islet cell differentiation, but other factors downstream of ngn3 must complete the task. Genetic...

The development of controllable and reproducible animal models of intracerebral hemorrhage (ICH) is essential for the systematic study of the pathophysiology and treatment of hemorrhagic stroke. In recent years, we have used a modified version of a murine ICH model to inject blood into mouse basal ganglia. According to our protocol, autologous blood is stereotactically infused in two stages into the right striatum to mimic the natural events of hemorrhagic stroke. Following ICH induction, animals demonstrate reproducible hematomas, brain edema formation and marked neurological deficits. Our technique has proven to be a reliable and reproducible means of creating ICH in mice in a number of acute and chronic studies. We believe that our model will serve as an ideal paradigm for investigating the complex pathophysiology of hemorrhagic stroke. The protocol for establishing this model takes about 2 h.