Gray level Co‐occurrence Matrices (GLCM) to assess microstructural and textural changes in pre‐implantation embryos

Abstract

The preimplantation embryo is extraordinarily sensitive to environmental signals and events such that perturbations can alter embryo metabolism and program an altered developmental trajectory, ultimately affecting the phenotype of the adult individual; indeed, the physical environment associated with in vitro embryo culture can attenuate development. Defining the underlying metabolic changes and mechanisms, however, has been limited by the imaging technology used to evaluate metabolites and structural features in the embryo. Here, we assessed the impact of in vitro fertilization and culture on mouse embryos using three metabolic markers: peroxyfluor 1 (a reporter of hydrogen peroxide), monochlorobimane (a reporter of glutathione), and Mitotracker Deep Red (a marker of mitochondria). We also evaluated the distribution pattern of histone 2AX gamma (γH2AX) in the nuclei of 2- and 8-cell embryos and blastocysts to investigate the degree of DNA damage caused by in vitro embryo culture. In vitro-fertilized embryos, in vivo-developed embryos, and in vivo-fertilized embryos recovered and cultured in vitro were compared at the 2-, 8-cell, and blastocyst stages. In addition to assessments based on fluorescence intensity, textural analysis using Gray Level Co-occurrence Matrix (GLCM), a statistical approach that assesses texture within an image, was used to evaluate peroxyfluor 1, monochlorobimane, and Mitotracker Deep Red staining in an effort to develop a robust metric of embryo quality. Our data provide strong evidence of modified metabolic parameters identifiable as altered fluorescence texture in embryos developed in vitro. Thus, texture-analysis approach may provide a means of gaining additional insight into embryo programming beyond conventional measurements of staining intensity for metabolic markers. Mol. Reprod. Dev. 83: 701-713, 2016 © 2016 Wiley Periodicals, Inc.

Physiological changes in newborn skin after natural oil massage in rural Nepal

“…The presence of "pebble" texture or miliaria each decreased the skin surface uniformity and increased the randomness ( Figure 5, Figure 5, Panels A and C). [24][25][26] The presence of "pebble" texture increased the randomness (entropy) relative to no "pebble" texture or miliaria ( Figure 5 Figure 5, Panel C). Images with miliaria were more random than images with no visible textural features ( Figure 5 Figure 5, Panel C).…”

“…The GLCM texture outcomes were: (i) angular second moment (ASM, uniformity) that is mathematically the squared sum of GLCM elements, (ii) contrast (local gray level variation), (iii) correlation (gray level linear dependency), (iv) inverse difference moment (IDM, homogeneity) and (v) entropy (randomness or disorder). [24][25][26] Additional details of the GLCM method are provided in the On On-line Supplementary Document line Supplementary Document). 27…”

Gray-Level Co-Occurrence Matrix Analysis of Granule Neurons of the Hippocampal Dentate Gyrus Following Cortical Injury

“…After each resolution unit is given a number based on its grayscale intensity, the software for texture analysis (in our study, we used the CellProfiler program) calculates textural features such as ASM and IDM. Since its establishment, the GLCM protocol has been successfully applied in various fields of medicine, including radiology, histology, pathology, and cell biology (Yang et al, 2012; Chen et al, 2015; Mapayi et al, 2015; Pantic et al, 2016 a ; Tan et al, 2016). Today, textural analysis can be done using various GLCM and non-GLCM algorithms and protocols, and many different programs for quantification of textural parameters have been developed.…”

Cardiac tissue engineering: state-of-the-art methods and outlook

“…In their demonstration, they were able to classify cardiac activity of unknown compounds with an accuracy of roughly 72% and generalize the results to other drug families with an accuracy above 70% [218]. Further, ML and its myriad algorithms can also be used on the protein and gene side of tissue engineering, as it has been demonstrated or proposed for histopathological image analysis [43], ligand affinity [42], folding structure [219], gene expression and biomarker data mining [220, 221], and in evaluation of pre-implantation embryos [222]. Large datasets such as the “Tissue Atlas” [223], a human proteome map categorized by tissue, could easily be used as a training and testing set for ML algorithms targeting identification of impaired tissue or disease onset.…”