REAVER Vascular Networks Fluorescent Image Dataset

Fluorescent Images of Vessel Networks from Various Murine Tissues

Purpose: 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.

• Labeling: IB4-Lectin with Alexa Flour 647
• Modality: Confocal Microscope Nikon 80i CLSM
• Objective: Mixture of 20x and 60x objective images
• Image Format: Images originally acquired in Nikon IDS format, converted to 8-bit greyscale TIFs found in “_Original_Images” folder.
• Questions: Email bac7wj@virginia.edu for inquiries.

External Links

1. Manuscript:
2. Code repository: https://github.com/bacorliss/REAVER_public for code to analyze this data (MATLAB 2019a).

Dataset Summary:

Each image folder contains 36 images. For each image:

1. The first channel (red) is the segmented image with values of 0 or 255 (false or true).
2. The second channel (green) is the skeleton image with values of 0 or 255 (false or true).
3. The third channel (blue) is empty except for the Manual images where the third channel contains the original raw image.

Subfolders

1. _Original_Images: contains raw input images.
2. AngioQuant_Auto: contains output images from automated analysis in AngioQuant.
3. AngioTool_Auto: contains output images from automated analysis in AngioTool.
4. ImageJ_Auto: contains output images from automated analysis in ImageJ.
5. ImageJ_Manual: contains output images from manual analysis in ImageJ.
6. RAVE_Auto: contains output images from automated analysis in RAVE.
7. REAVER_Auto: contains output images from automated analysis in REAVER.

Image Metadata and Output data

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:

1. branchpoint_RC: 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”
2. mean_diameter: The mean diameter of vessels in the image
3. num_branchpts: The number of branchpoints in the image
4. threshold_false_neg: 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”
5. threshold_false_pos: 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”
6. threshold_true_neg: 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”
7. threshold_true_pos: 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”
8. umppix: The length of the edge of one pixel in micrometers
9. vessel_area: The number of “true” pixels in the segmented image
10. vessel_length: The number of “true” pixels in the skeleton image

Dataset Output Data

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:

1. Program: Designates the program used to calculate the data for that row
2. Tissue_Type: Gives the tissue type for the image
3. Image_Name: Gives the specific name of the given image
4. Vessel_Length: The number of “true” pixels in the skeleton image
5. Vessel_Area: The number of “true” pixels in the segmented image
6. Mean_Diameter: The mean diameter of vessels in the image
7. Num_Branchpoints: The number of branchpoints in the image
8. Sensitivity: (Number of True Positive pixels) / (Number of True Positive pixels + Number of False Negative pixels)
9. Specificity: (Number of True Negative pixels) / (Number of True Negative pixels + Number of False Positive pixels)
10. Accuracy: (Number of True Positive pixels + Number of True Negative pixels) / (Total number of pixels)
11. umppix: The length of the edge of one pixel in micrometers
12. pix_dim: The edge length in pixels of the square image

References

1.         Niemisto, A., Dunmire, V., Yli-Harja, O., Wei Zhang & Shmulevich, I. Robust quantification of in vitro angiogenesis through image analysis. IEEE Trans. Med. Imaging 24, 549–553 (2005).

2.         Zudaire, E., Gambardella, L., Kurcz, C. & Vermeren, S. A Computational Tool for Quantitative Analysis of Vascular Networks. PLOS ONE 6, e27385 (2011).

3.         Seaman, M. E., Peirce, S. M. & Kelly, K. Rapid Analysis of Vessel Elements (RAVE): A Tool for Studying Physiologic, Pathologic and Tumor Angiogenesis. PLoS ONE 6, e20807 (2011).