Summary Deep learning is a significant step forward for developing autonomous tasks. One of its branches, computer vision, allows image recognition with high accuracy thanks to the use of convolutional neural networks (CNNs). Our goal was to train a CNN with transmitted light microscopy images to distinguish pluripotent stem cells from early differentiating cells. We induced differentiation of mouse embryonic stem cells to epiblast-like cells and took images at several time points from the initial stimulus. We found that the networks can be trained to recognize undifferentiated cells from differentiating cells with an accuracy higher than 99%. Successful prediction started just 20 min after the onset of differentiation. Furthermore, CNNs displayed great performance in several similar pluripotent stem cell (PSC) settings, including mesoderm differentiation in human induced PSCs. Accurate cellular morphology recognition in a simple microscopic set up may have a significant impact on how cell assays are performed in the near future.

Skeletal muscle has the remarkable ability to regenerate. However, with age and disease muscle strength and function decline. Myofiber size, which is affected by injury and disease, is a critical measurement to assess muscle health. Here, we test and apply Cellpose, a recently developed deep learning algorithm, to automatically segment myofibers within murine skeletal muscle. We first show that tissue fixation is necessary to preserve cellular structures such as primary cilia, small cellular antennae, and adipocyte lipid droplets. However, fixation generates heterogeneous myofiber labeling, which impedes intensity-based segmentation. We demonstrate that Cellpose efficiently delineates thousands of individual myofibers outlined by a variety of markers, even within fixed tissue with highly uneven myofiber staining. We created a novel ImageJ plugin (LabelsToRois) that allows processing of multiple Cellpose segmentation images in batch. The plugin also contains a semi-automatic erosion function to correct for the area bias introduced by the different stainings, thereby identifying myofibers as accurately as human experts. We successfully applied our segmentation pipeline to uncover myofiber regeneration differences between two different muscle injury models, cardiotoxin and glycerol. Thus, Cellpose combined with LabelsToRois allows for fast, unbiased, and reproducible myofiber quantification for a variety of staining and fixation conditions.

Real-time reverse transcription PCR (RT-qPCR) is the gold-standard technique for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) detection in nasopharyngeal swabs specimens. The analysis by RT-qPCR usually requires a previous extraction step to obtain the purified viral RNA. Unfortunately, RNA extraction constitutes a bottleneck for early detection in many countries since it is expensive, time-consuming and depends on the availability of commercial kits. Here, we describe an extraction-free protocol for SARS-CoV-2 detection by RT-qPCR from nasopharyngeal swab clinical samples in saline solution. The method includes a treatment with proteinase K followed by heat inactivation (PK+HID method). We demonstrate that PK+HID improves the RT-qPCR performance in comparison to the heat-inactivation procedure. Moreover, we show that this extraction-free protocol can be combined with a variety of multiplexing RT-qPCR kits. The method combined with a multiplexing detection kit targeting N and ORF1ab viral genes showed a sensitivity of 0.99 and a specificity of 0.99 from the analysis of 106 positive and 106 negative clinical samples. In conclusion, PK+HID is a robust, fast and inexpensive procedure for extraction-free RT-qPCR determinations of SARS-CoV-2. The National Administration of Drugs, Foods and Medical Devices of Argentina has recently authorized the use of this method.