Phasetime: Deep Learning Approach to Detect Nuclei in Time Lapse Phase Images

Decoding the mechanisms of chimeric antigen receptor (CAR) T cell-mediated killing of tumors: insights from granzyme and Fas inhibition

Chimeric antigen receptor (CAR) T cell show promise in cancer treatments, but their mechanism of action is not well understood. Decoding the mechanisms used by individual T cells can help improve the efficacy of T cells while also identifying mechanisms of T cell failure leading to tumor escape. Here, we used a suite of assays including dynamic single-cell imaging of cell-cell interactions, dynamic imaging of fluorescent reporters to directly track cytotoxin activity in tumor cells, and scRNA-seq on patient infusion products to investigate the cytotoxic mechanisms used by individual CAR T cells in killing tumor cells. We show that surprisingly, overexpression of the Granzyme B (GZMB) inhibitor, protease inhibitor-9 (PI9), does not alter the cytotoxicity mediated by CD19-specific CAR T cells against either the leukemic cell line, NALM6; or the ovarian cancer cell line, SkOV3-CD19. We designed and validated reporters to directly assay T cell delivered GZMB activity in tumor cells and confirmed that while PI9 overexpression inhibits GZMB activity at the molecular level, this is not sufficient to impact the kinetics or magnitude of killing mediated by the CAR T cells. Altering cytotoxicity mediated by CAR T cells required combined inhibition of multiple pathways that are tumor cell specific: (a) B-cell lines like NALM6, Raji and Daudi were sensitive to combined GZMB and granzyme A (GZMA) inhibition; whereas (b) solid tumor targets like SkOV3-CD19 and A375-CD19 (melanoma) were sensitive to combined GZMB and Fas ligand inhibition. We realized the translational relevance of these findings by examining the scRNA-seq profiles of Tisa-cel and Axi-cel infusion products and show a significant correlation between GZMB and GZMA expression at the single-cell level in a T cell subset-dependent manner. Our findings highlight the importance of the redundancy in killing mechanisms of CAR T cells and how this redundancy is important for efficacious T cells.

Quantitative Approaches to Study Retinal Neurogenesis

The study of the development of the vertebrate retina can be addressed from several perspectives: from a purely qualitative to a more quantitative approach that takes into account its spatio-temporal features, its three-dimensional structure and also the regulation and properties at the systems level. Here, we review the ongoing transition toward a full four-dimensional characterization of the developing vertebrate retina, focusing on the challenges at the experimental, image acquisition, image processing and quantification. Using the developing zebrafish retina, we illustrate how quantitative data extracted from these type of highly dense, three-dimensional tissues depend strongly on the image quality, image processing and algorithms used to segment and quantify. Therefore, we propose that the scientific community that focuses on developmental systems could strongly benefit from a more detailed disclosure of the tools and pipelines used to process and analyze images from biological samples.

Differentiating Benign from Malignant Renal Tumors Using T2- and Diffusion-Weighted Images: A Comparison of Deep Learning and Radiomics Models Versus Assessment from Radiologists

BACKGROUND

Differentiating benign from malignant renal tumors is important for selection of the most effective treatment.

PURPOSE

To develop magnetic resonance imaging (MRI)-based deep learning (DL) models for differentiation of benign and malignant renal tumors and to compare their discrimination performance with the performance of radiomics models and assessment by radiologists.

STUDY TYPE

Retrospective.

POPULATION

A total of 217 patients were randomly assigned to a training cohort (N = 173) or a testing cohort (N = 44).

FIELD STRENGTH/SEQUENCE

Diffusion-weighted imaging (DWI) and fast spin-echo sequence T2-weighted imaging (T2WI) at 3.0T.

ASSESSMENT

A radiologist manually labeled the region of interest (ROI) on each image. Three DL models using ResNet-18 architecture and three radiomics models using random forest were developed using T2WI alone, DWI alone, and a combination of the two image sets to discriminate between benign and malignant renal tumors. The diagnostic performance of two radiologists was assessed based on professional experience. We also compared the performance of each model and the radiologists.

STATISTICAL TESTS

The area under the receiver operating characteristic (ROC) curve (AUC) was used to assess the performance of each model and the radiologists. P < 0.05 indicated statistical significance.

RESULTS

The AUC of the DL models based on T2WI, DWI, and the combination was 0.906, 0.846, and 0.925 in the testing cohorts, respectively. The AUC of the combination DL model was significantly better than that of the models based on individual sequences (0.925 > 0.906, 0.925 > 0.846). The AUC of the radiomics models based on T2WI, DWI, and the combination was 0.824, 0.742, and 0.826 in the testing cohorts, respectively. The AUC of two radiologists was 0.724 and 0.667 in the testing cohorts.

CONCLUSION

Thus, the MRI-based DL model is useful for differentiating benign from malignant renal tumors in clinic, and the DL model based on T2WI + DWI had the best performance.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY STAGE: 2.

A survey on applications of deep learning in microscopy image analysis

Advanced microscopy enables us to acquire quantities of time-lapse images to visualize the dynamic characteristics of tissues, cells or molecules. Microscopy images typically vary in signal-to-noise ratios and include a wealth of information which require multiple parameters and time-consuming iterative algorithms for processing. Precise analysis and statistical quantification are often needed for the understanding of the biological mechanisms underlying these dynamic image sequences, which has become a big challenge in the field. As deep learning technologies develop quickly, they have been applied in bioimage processing more and more frequently. Novel deep learning models based on convolution neural networks have been developed and illustrated to achieve inspiring outcomes. This review article introduces the applications of deep learning algorithms in microscopy image analysis, which include image classification, region segmentation, object tracking and super-resolution reconstruction. We also discuss the drawbacks of existing deep learning-based methods, especially on the challenges of training datasets acquisition and evaluation, and propose the potential solutions. Furthermore, the latest development of augmented intelligent microscopy that based on deep learning technology may lead to revolution in biomedical research.

Deep Learning for Biospectroscopy and Biospectral Imaging: State-of-the-Art and Perspectives

With the advances in instrumentation and sampling techniques, there is an explosive growth of data from molecular and cellular samples. The call to extract more information from the large data sets has greatly challenged the conventional chemometrics method. Deep learning, which utilizes very large data sets for finding hidden features therein and for making accurate predictions for a wide range of applications, has been applied in an unbelievable pace in biospectroscopy and biospectral imaging in the recent 3 years. In this Feature, we first introduce the background and basic knowledge of deep learning. We then focus on the emerging applications of deep learning in the data preprocessing, feature detection, and modeling of the biological samples for spectral analysis and spectroscopic imaging. Finally, we highlight the challenges and limitations in deep learning and the outlook for future directions.

Pseudo-nuclear staining of cells by deep learning improves the accuracy of automated cell counting in a label-free cellular population

Deep learning has emerged as a breakthrough tool for the segmentation of images without supporting human experts. Here, we propose an automated approach that uses deep learning to generate pseudo-nuclear staining of cells from phase contrast images. Our proposed approach, which has the feature to generate pseudo-nuclear stained images by simple DNN, showed remarkable advantages over existing approaches in the cell-detection and the detection of the relative position of cells for various cell densities, as well as in counting the exact cell numbers. Pseudo-nuclear staining of cells by deep learning will improve the accuracy of automated cell counting in a label-free cellular population on phase contrast images.

A bird's-eye view of deep learning in bioimage analysis

Deep learning of artificial neural networks has become the de facto standard approach to solving data analysis problems in virtually all fields of science and engineering. Also in biology and medicine, deep learning technologies are fundamentally transforming how we acquire, process, analyze, and interpret data, with potentially far-reaching consequences for healthcare. In this mini-review, we take a bird's-eye view at the past, present, and future developments of deep learning, starting from science at large, to biomedical imaging, and bioimage analysis in particular.