DeepHistoClass: A Novel Strategy for Confident Classification of Immunohistochemistry Images Using Deep Learning

Utilization of the QuPath open-source software platform for analysis of mammalian spermatogenesis†

The adult mammalian testis is filled with seminiferous tubules, which contain somatic Sertoli cells along with germ cells undergoing all phases of spermatogenesis. During spermatogenesis in postnatal mice, male germ cells undergo at least 17 different nomenclature changes as they proceed through mitosis as spermatogonia (=8), meiosis as spermatocytes (=6), and spermiogenesis as spermatids (=3). Adding to this complexity, combinations of germ cells at each of these stages of development are clumped together along the length of the seminiferous tubules. Due to this, considerable expertise is required for investigators to accurately analyze changes in spermatogenesis in animals that have spontaneous mutations, have been genetically modified (transgenic or knockout/knockin), or have been treated with pharmacologic agents. Here, we leverage our laboratory's expertise in spermatogenesis to optimize the open-source "Quantitative Pathology & Bioimage Analysis" software platform for automated analyses of germ and somatic cell populations in both the developing and adult mammalian testis.

Rapid detection of mouse spermatogenic defects by testicular cellular composition analysis via enhanced deep learning model

BACKGROUND

Histological analysis of the testicular sections is paramount in infertility research but tedious and often requires months of training and practice.

OBJECTIVES

Establish an expeditious histopathological analysis of mutant mice testicular sections stained with commonly available hematoxylin and eosin (H&E) via enhanced deep learning model MATERIALS AND METHODS: Automated segmentation and cellular composition analysis on the testes of six mouse reproductive mutants of key reproductive gene family, DAZ and PUMILIO gene family via H&E-stained mouse testicular sections.

RESULTS

We improved the deep learning model with human interaction to achieve better pixel accuracy and reduced annotation time for histologists; revealed distinctive cell composition features consistent with previously published phenotypes for four mutants and novel spermatogenic defects in two newly generated mutants; established a fast spermatogenic defect detection protocol for quantitative and qualitative assessment of testicular defects within 2.5-3 h, requiring as few as 8 H&E-stained testis sections; uncovered novel defects in AcDKO and a meiotic arrest defect in HDBKO, supporting the synergistic interaction of Sertoli Pum1 and Pum2 as well as redundant meiotic function of Dazl and Boule.

DISCUSSION

Our testicular compositional analysis not only could reveal spermatogenic defects from staged seminiferous tubules but also from unstaged seminiferous tubule sections.

CONCLUSION

Our SCSD-Net model offers a rapid protocol for detecting reproductive defects from H&E-stained testicular sections in as few as 3 h, providing both quantitative and qualitative assessments of spermatogenic defects. Our analysis uncovered evidence supporting the synergistic interaction of Sertoli PUM1 and PUM2 in maintaining average testis size, and redundant roles of DAZ family proteins DAZL and BOULE in meiosis.

Proteomics mining of cancer hallmarks on a single-cell resolution

Dysregulated proteome is an essential contributor in carcinogenesis. Protein fluctuations fuel the progression of malignant transformation, such as uncontrolled proliferation, metastasis, and chemo/radiotherapy resistance, which severely impair therapeutic effectiveness and cause disease recurrence and eventually mortality among cancer patients. Cellular heterogeneity is widely observed in cancer and numerous cell subtypes have been characterized that greatly influence cancer progression. Population-averaged research may not fully reveal the heterogeneity, leading to inaccurate conclusions. Thus, deep mining of the multiplex proteome at the single-cell resolution will provide new insights into cancer biology, to develop prognostic biomarkers and treatments. Considering the recent advances in single-cell proteomics, herein we review several novel technologies with particular focus on single-cell mass spectrometry analysis, and summarize their advantages and practical applications in the diagnosis and treatment for cancer. Technological development in single-cell proteomics will bring a paradigm shift in cancer detection, intervention, and therapy.

Detection of spermatogonial stem/progenitor cells in prepubertal mouse testis with deep learning

PURPOSE

Rapid and easy detection of spermatogonial stem/progenitor cells (SSPCs) is crucial for clinicians dealing with male infertility caused by prepubertal testicular damage. Deep learning (DL) methods may offer visual tools for tracking SSPCs on testicular strips of prepubertal animal models. The purpose of this study is to detect and count the seminiferous tubules and SSPCs in newborn mouse testis sections using a DL method.

METHODS

Testicular sections of the C57BL/6-type newborn mice were obtained and enumerated. Odd-numbered sections were stained with hematoxylin and eosin (H&E), and even-numbered sections were immune labeled (IL) with SSPC specific marker, SALL4. Seminiferous tubule and SSPC datasets were created using odd-numbered sections. SALL4-labeled sections were used as positive control. The YOLO object detection model based on DL was used to detect seminiferous tubules and stem cells.

RESULTS

Test scores of the DL model in seminiferous tubules were obtained as 0.98 mAP, 0.93 precision, 0.96 recall, and 0.94 f1-score. The SSPC test scores were obtained as 0.88 mAP, 0.80 precision, 0.93 recall, and 0.82 f1-score.

CONCLUSION

Seminiferous tubules and SSPCs on prepubertal testicles were detected with a high sensitivity by preventing human-induced errors. Thus, the first step was taken for a system that automates the detection and counting process of these cells in the infertility clinic.

The human protein atlas-Integrated omics for single cell mapping of the human proteome

Studying the spatial distribution of proteins provides the basis for understanding the biology, molecular repertoire, and architecture of every human cell. The Human Protein Atlas (HPA) has grown into one of the world's largest biological databases, and in the most recent version, a major update of the structure of the database was performed. The data has now been organized into 10 different comprehensive sections, each summarizing different aspects of the human proteome and the protein-coding genes. In particular, large datasets with information on the single cell type level have been integrated, refining the tissue and cell type specificity and detailing the expression in cell states with an increased resolution. The multi-modal data constitute an important resource for both basic and translational science, and hold promise for integration with novel emerging technologies at the protein and RNA level.

Self-supervised learning of cell type specificity from immunohistochemical images

MOTIVATION

Advances in bioimaging now permit in situ proteomic characterization of cell-cell interactions in complex tissues, with important applications across a spectrum of biological problems from development to disease. These methods depend on selection of antibodies targeting proteins that are expressed specifically in particular cell types. Candidate marker proteins are often identified from single-cell transcriptomic data, with variable rates of success, in part due to divergence between expression levels of proteins and the genes that encode them. In principle, marker identification could be improved by using existing databases of immunohistochemistry for thousands of antibodies in human tissue, such as the Human Protein Atlas. However, these data lack detailed annotations of the types of cells in each image.

RESULTS

We develop a method to predict cell type specificity of protein markers from unlabeled images. We train a convolutional neural network with a self-supervised objective to generate embeddings of the images. Using non-linear dimensionality reduction, we observe that the model clusters images according to cell types and anatomical regions for which the stained proteins are specific. We then use estimates of cell type specificity derived from an independent single-cell transcriptomics dataset to train an image classifier, without requiring any human labelling of images. Our scheme demonstrates superior classification of known proteomic markers in kidney compared to selection via single-cell transcriptomics.

AVAILABILITY AND IMPLEMENTATION

Code and trained model are available at www.github.com/murphy17/HPA-SimCLR.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

A Method for Unsupervised Semi-Quantification of Inmunohistochemical Staining with Beta Divergences

In many research laboratories, it is essential to determine the relative expression levels of some proteins of interest in tissue samples. The semi-quantitative scoring of a set of images consists of establishing a scale of scores ranging from zero or one to a maximum number set by the researcher and assigning a score to each image that should represent some predefined characteristic of the IHC staining, such as its intensity. However, manual scoring depends on the judgment of an observer and therefore exposes the assessment to a certain level of bias. In this work, we present a fully automatic and unsupervised method for comparative biomarker quantification in histopathological brightfield images. The method relies on a color separation method that discriminates between two chromogens expressed as brown and blue colors robustly, independent of color variation or biomarker expression level. For this purpose, we have adopted a two-stage stain separation approach in the optical density space. First, a preliminary separation is performed using a deconvolution method in which the color vectors of the stains are determined after an eigendecomposition of the data. Then, we adjust the separation using the non-negative matrix factorization method with beta divergences, initializing the algorithm with the matrices resulting from the previous step. After that, a feature vector of each image based on the intensity of the two chromogens is determined. Finally, the images are annotated using a systematically initialized k-means clustering algorithm with beta divergences. The method clearly defines the initial boundaries of the categories, although some flexibility is added. Experiments for the semi-quantitative scoring of images in five categories have been carried out by comparing the results with the scores of four expert researchers yielding accuracies that range between 76.60% and 94.58%. These results show that the proposed automatic scoring system, which is definable and reproducible, produces consistent results.

Accurate Quantitative Histomorphometric-Mathematical Image Analysis Methodology of Rodent Testicular Tissue and Its Possible Future Research Perspectives in Andrology and Reproductive Medicine

Infertility is increasing worldwide; male factors can be identified in nearly half of all infertile couples. Histopathologic evaluation of testicular tissue can provide valuable information about infertility; however, several different evaluation methods and semi-quantitative score systems exist. Our goal was to describe a new, accurate and easy-to-use quantitative computer-based histomorphometric-mathematical image analysis methodology for the analysis of testicular tissue. On digitized, original hematoxylin-eosin (HE)-stained slides (scanned by slide-scanner), quantitatively describable characteristics such as area, perimeter and diameter of testis cross-sections and of individual tubules were measured with the help of continuous magnification. Immunohistochemically (IHC)-stained slides were digitized with a microscope-coupled camera, and IHC-staining intensity measurements on digitized images were also taken. Suggested methods are presented with mathematical equations, step-by-step detailed characterization and representative images are given. Our novel quantitative histomorphometric-mathematical image analysis method can improve the reproducibility, objectivity, quality and comparability of andrological-reproductive medicine research by recognizing even the mild impairments of the testicular structure expressed numerically, which might not be detected with the present semi-quantitative score systems. The technique is apt to be subjected to further automation with machine learning and artificial intelligence and can be named ‘Computer-Assisted or -Aided Testis Histology’ (CATHI).