Search

Intraoperative diagnosis is essential for providing safe and effective care during cancer surgery1. The existing workflow for intraoperative diagnosis based on hematoxylin and eosin staining of processed tissue is time, resource and labor intensive2,3. Moreover, interpretation of intraoperative histologic images is dependent on a contracting, unevenly distributed, pathology workforce4. In the present study, we report a parallel workflow that combines stimulated Raman histology (SRH)5-7, a label-free optical imaging method and deep convolutional neural networks (CNNs) to predict diagnosis at the bedside in near real-time in an automated fashion. Specifically, our CNNs, trained on over 2.5 million SRH images, predict brain tumor diagnosis in the operating room in under 150 s, an order of magnitude faster than conventional techniques (for example, 20-30 min)2. In a multicenter, prospective clinical trial (n = 278), we demonstrated that CNN-based diagnosis of SRH images was noninferior to pathologist-based interpretation of conventional histologic images (overall accuracy, 94.6% versus 93.9%). Our CNNs learned a hierarchy of recognizable histologic feature representations to classify the major histopathologic classes of brain tumors. In addition, 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 complementary pathway for tissue diagnosis that is independent of a traditional pathology laboratory.

We have implemented an unbiased cell morphology-based screen to identify small-molecule modulators of cellular processes using the Cytometrix (TM) automated imaging and analysis system. This assay format provides unbiased analysis of morphological effects induced by small molecules by capturing phenotypic readouts of most known classes of pharmacological agents and has the potential to read out pathways for which little is known. Four human-cancer cell lines and one noncancerous primary cell type were treated with 107 small molecules comprising four different protein kinase-inhibitor scaffolds. Cellular phenotypes induced by each compound were quantified by multivariate statistical analysis of the morphology, staining intensity, and spatial attributes of the cellular nuclei, microtubules, and Golgi compartments. Principal component analysis was used to identify inhibitors of cellular components not targeted by known protein kinase inhibitors. Here we focus on a hydroxyl-substituted analog (hydroxy-PP) of the known Src-family kinase inhibitor PP2 because it induced cell-specific morphological features distinct from all known kinase inhibitors in the collection. We used affinity purification to identify a target of hydroxy-PP, carbonyl reductase 1 (CBR1), a short-chain dehydrogenase-reductase. We solved the X-ray crystal structure of the CBR1/hydroxy-PP complex to 1.24 A resolution. Structure-based design of more potent and selective CBR1 inhibitors provided probes for analyzing the biological function of CBR1 in A549 cells. These studies revealed a previously unknown function for CBR1 in serum-withdrawal-induced apoptosis. Further studies indicate CBR1 inhibitors may enhance the effectiveness of anticancer anthracyclines. Morphology-based screening of diverse cancer cell types has provided a method for discovering potent new small-molecule probes for cell biological studies and anticancer drug candidates.

New developments in optical studies of single molecules at room temperature have been assembled with concentration on underlying principles and the potential of single-molecule experiments. Single-molecule-studies examples include those that involve photochemistry and photophysics experiments related to spectral fluctuations, single-molecule measurements, Raman spectroscopy, conformational dynamics, diffusional motions, fluorescence resonant energy transfer, enzymatic turnovers and exciton dynamics. These studies are an illustration of the information that can be obtained with the single-molecule approach not easily seen in ensemble-averaged measurements.

Compounds with similar target specificities and modes of inhibition cause similar cellular phenotypes. Based on this observation, we hypothesized that we could quantitatively classify compounds with diverse mechanisms of action using cellular phenotypes and identify compounds with unintended cellular activities within a chemical series. We have developed Cytometrix technologies, a highly automated image-based system capable of quantifying, clustering, and classifying changes in cellular phenotypes for this purpose. Using this system, 45 out of 51 known compounds were accurately classified into 12 distinct mechanisms of action. We also demonstrate microtubule-binding activity in one of seven related cytochalasin actin poisons. This technology can be used for a variety of drug discovery applications, including high-throughput primary screening of chemical and siRNA libraries and as a secondary assay to detect unintended activities and toxicities.