Journal: Communications Biology

Article Title: A framework for multiplex imaging optimization and reproducible analysis

doi: 10.1038/s42003-022-03368-y

Figure Lengend Snippet: Reproducible image analytics.

Article Snippet: Additionally, we demonstrate processing Zeiss Axioscan, Akoya CODEX, and Miltenyi MACSima prototype images in our pipeline example Jupyter notebooks, here: https://gitlab.com/engje/mplexable/-/tree/master/jupyter .

Techniques:

Journal: Frontiers in Immunology

Article Title: A real-time GPU-accelerated parallelized image processor for large-scale multiplexed fluorescence microscopy data

doi: 10.3389/fimmu.2022.981825

Figure Lengend Snippet: Overview of image processing methods for CODEX multiplexed imaging data. (A) CODEX workflow and the structure of a raw imaging dataset. (B) Overview of the CODEX Uploader pipeline. (C) Overview of the RAPID pipeline. (D) Plot of processing time versus number of regions analyzed with the two pipelines on a local system and for the RAPID run on a cloud-based system.

Article Snippet: We demonstrated that the open-source algorithm reduced blur and improved resolution of CODEX images as effectively as the commercial software utilized by the CODEX Uploader ( ).

Techniques: Imaging

Journal: Frontiers in Immunology

Article Title: A real-time GPU-accelerated parallelized image processor for large-scale multiplexed fluorescence microscopy data

doi: 10.3389/fimmu.2022.981825

Figure Lengend Snippet: Comparison between the old and new CODEX image processing pipelines.

Article Snippet: We demonstrated that the open-source algorithm reduced blur and improved resolution of CODEX images as effectively as the commercial software utilized by the CODEX Uploader ( ).

Techniques:

Journal: Frontiers in Immunology

Article Title: A real-time GPU-accelerated parallelized image processor for large-scale multiplexed fluorescence microscopy data

doi: 10.3389/fimmu.2022.981825

Figure Lengend Snippet: RAPID corrects axial and lateral drifts in CODEX multiplexed data. (A) Representative CODEX nuclear images from cycles 1, 4, and 5 of a PDAC tissue section processed by the CODEX Uploader and by RAPID. (B) Percentage of in-focus tiles across nine tiles and nine imaging cycles for one example tissue region. (C) Representative CODEX images of PDAC tissue section stained for CD45 and CD3 processed by the two pipelines, and the corresponding cell gating using CD3 and CD45 protein expression. (D) Representative images of fibroblasts and epithelium processed by the two pipelines. (E) Percentage of the non-distorted cells across the twelve tile-tile interfaces of an example tissue region. Wilcoxon matched-pairs signed rank test was used in (B, E) . *P < 0.05, ***P < 0.001.

Article Snippet: We demonstrated that the open-source algorithm reduced blur and improved resolution of CODEX images as effectively as the commercial software utilized by the CODEX Uploader ( ).

Techniques: Imaging, Staining, Expressing

Journal: Frontiers in Immunology

Article Title: A real-time GPU-accelerated parallelized image processor for large-scale multiplexed fluorescence microscopy data

doi: 10.3389/fimmu.2022.981825

Figure Lengend Snippet: RAPID improves down-stream cell-type identification. (A) Unsupervised clustering and cell-type annotation of CODEX data processed using the CODEX Uploader (total number of cells = 8967) and RAPID (total number of cells = 9597). (B) Bar graphs of the numbers of annotated cells identified by the two pipelines. (C) Overlays of a blank image (Cy3 channel) with cell centroids (green crosses) assigned to the RBC-contaminated cell clusters after processing with the two pipelines. (D) Overlays of Hoechst nuclear stain with CD3, CD20, CD31, αSMA, and Granzyme B stainings of images processed with CODEX Uploader and with RAPID. Noise from RBC autofluorescence is indicated by white (due to overlay of all colors). (E) Scatter plots of x/y coordinates of cell centroids showing the spatial distribution of the annotated clusters in the tissue. Centroids of cells from the RBC contaminated cluster (orange) co-localize with white spots in images in panel (D) .

Article Snippet: We demonstrated that the open-source algorithm reduced blur and improved resolution of CODEX images as effectively as the commercial software utilized by the CODEX Uploader ( ).

Techniques: Staining