REAVER Vascular Networks Fluorescent Image Dataset

<strong>Fluorescent Images of Vessel Networks from Various Murine Tissues</strong> <strong>Purpose</strong>: Image dataset of vascular networks with a diverse range of vessel architectures. Dataset is used to evaluate performance of several image processing programs (AngioQuant¹, AngioTool², RAVE³, REAVER). Manual analysis from ImageJ is used as ground truth to compare other programs against. <strong>Labeling</strong>: IB4-Lectin with Alexa Flour 647 <strong>Modality</strong>: Confocal Microscope Nikon 80i CLSM <strong>Objective</strong>: Mixture of 20x and 60x objective images <strong>Image Format</strong>: Images originally acquired in Nikon IDS format, converted to 8-bit greyscale TIFs found in “_Original_Images” folder. <strong>Questions</strong>: Email [e-mail] for inquiries. <strong>External Links</strong> <strong>Manuscript</strong>: <strong>Code repository: </strong>https://github.com/bacorliss/REAVER_public for code to analyze this data (MATLAB 2019a). <strong>Dataset Summary:</strong> Each image folder contains 36 images. For each image: The first channel (red) is the segmented image with values of 0 or 255 (false or true). The second channel (green) is the skeleton image with values of 0 or 255 (false or true). The third channel (blue) is empty except for the Manual images where the third channel contains the original raw image. <strong>Subfolders</strong> <strong>_Original_Images</strong>: contains raw input images. <strong>AngioQuant_Auto</strong>: contains output images from automated analysis in AngioQuant. <strong>AngioTool_Auto</strong>: contains output images from automated analysis in AngioTool. <strong>ImageJ_Auto</strong>: contains output images from automated analysis in ImageJ. <strong>ImageJ_Manual</strong>: contains output images from manual analysis in ImageJ. <strong>RAVE_Auto</strong>: contains output images from automated analysis in RAVE. <strong>REAVER_Auto</strong>: contains output images from automated analysis in REAVER. <strong>Image Metadata and Output data</strong> Each image folder has a .mat file called “Results.mat” containing the results of analysis in the form of the following variables all of which are 1x36 arrays (one entry for each image) unless specified otherwise: <strong>branchpoint_RC</strong>: A 1x36 struct containing the row-column values for each branchpoint in the i^th image (when organized in alphabetic order which is the order given everywhere else); Effectively the same as “BranchpointsByName.mat” <strong>mean_diameter</strong>: The mean diameter of vessels in the image <strong>num_branchpts</strong>: The number of branchpoints in the image <strong>threshold_false_neg</strong>: The number of false negative pixels – a pixel is a false negative if the program has it as “false” and the manual image has the pixel as “true” <strong>threshold_false_pos</strong>: The number of false positive pixels – a pixel is a false positive if the program has it as “true” and the manual image has the pixel as “false” <strong>threshold_true_neg</strong>: The number of true negative pixels – a pixel is a true negative if the program has it as “false” and the manual image has the pixel as “false” <strong>threshold_true_pos</strong>: The number of true positive pixels – a pixel is a false positive if the program has it as “true” and the manual image has the pixel as “true” <strong>umppix</strong>: The length of the edge of one pixel in micrometers <strong>vessel_area</strong>: The number of “true” pixels in the segmented image <strong>vessel_length</strong>: The number of “true” pixels in the skeleton image <strong>Dataset Output Data</strong> The file “image_quantification.csv” in the base folder contains the aggregated results from each image folder. Each row contains the results for a given (Program, Image) pair. The columns are described below: <strong>Program</strong>: Designates the program used to calculate the data for that row <strong>Tissue_Type</strong>: Gives the tissue type for the image <strong>Image_Name</strong>: Gives the specific name of the given image <strong>Vessel_Length</strong>: The number of “true” pixels in the skeleton image <strong>Vessel_Area</strong>: The number of “true” pixels in the segmented image <strong>Mean_Diameter</strong>: The mean diameter of vessels in the image <strong>Num_Branchpoints</strong>: The number of branchpoints in the image <strong>Sensitivity</strong>: (Number of True Positive pixels) / (Number of True Positive pixels + Number of False Negative pixels) <strong>Specificity</strong>: (Number of True Negative pixels) / (Number of True Negative pixels + Number of False Positive pixels) <strong>Accuracy</strong>: (Number of True Positive pixels + Number of True Negative pixels) / (Total number of pixels) <strong>umppix</strong>: The length of the edge of one pixel in micrometers <strong>pix_dim</strong>: The edge length in pixels of the square image <strong>References</strong> 1. Niemisto, A., Dunmire, V., Yli-Harja, O., Wei Zhang &amp; Shmulevich, I. Robust quantification of in vitro angiogenesis through image analysis. <em>IEEE Trans. Med. Imaging</em> <strong>24</strong>, 549–553 (2005). 2. Zudaire, E., Gambardella, L., Kurcz, C. &amp; Vermeren, S. A Computational Tool for Quantitative Analysis of Vascular Networks. <em>PLOS ONE</em> <strong>6</strong>, e27385 (2011). 3. Seaman, M. E., Peirce, S. M. &amp; Kelly, K. Rapid Analysis of Vessel Elements (RAVE): A Tool for Studying Physiologic, Pathologic and Tumor Angiogenesis. <em>PLoS ONE</em> <strong>6</strong>, e20807 (2011).