uw-loci / curvelets

Curvelet-Transform based fibrillar collagen quantification (CurveAlign and CT-FIRE)

Curvelet-Transform based fibrillar collagen quantification (CurveAlign and CT-FIRE)

This software package includes two tools, i.e. CurveAlign[1-4] and CT-FIRE[5-6] for collagen fiber quantification. CurveAlign is a curvelet transform (CT)[7]-based quantitative tool for interpreting the regional interaction between collagen and tumors by assessment of up to ~thirty fiber features, including angle, alignment, and density. CT-FIRE analyzes individual fiber metrics such as length, width, angle, and curvature. The individual fibers extracted by CT-FIRE can be imported into the CurveAlign as one of the two fiber analyse modes. The approach of CT-FIRE is described in [5], which combines the advantages of the fast discrete curvelet transform for denoising images, enhancement of the fiber edge features, and the fiber extraction (FIRE) algorithm [6] for extracting individual fibers. For now, CurveAlign should be used for bulk assessment of collagen features including angles/density and CT-FIRE for individual fiber quantification.

Since the release in 2020, we have been adding new features, mainly including 1) Bio-Formats importer/exporter to enhance image format support; 2) auto-threshold for CT-FIRE to optionally apply different thresholds for image stack or images from different acquisition settings; 3) fiber intersection points calculation based on CT-FIRE fiber extraction; 4) ROI based fiber density analysis; 5) deep learning based cell-fiber analysis. We have started running python-based deep learning modules for individual cell analysis which is integrated to the MATLAB-based fiber analysis pipeline to quantify the fiber-cell/tumor interactions.

CurveAlign Example	CT-FIRE Example

A quick manual to a cell-based fiber and tumor analysis module for the newest version CurveAlign 6.0 can be found here (testing only for now). An illustration of three relative measurements between fibers and associated annotation(can be tumor, cell, and any ROI) is shown below.

CurveAlign and CT-FIRE are licensed under the 2-Clause BSD license as described LICENSE.txt, except for some third-party code whose licenses are described in LICENSE-third-party.txt. One third-party code, CurveLab 2.1.2 MATLAB package for curvelet transform, can only be downloaded from http://www.curvelet.org/software.html. To run CurveAlign or CT-FIRE, the CurveLab needs to be downloaded and added to the MATLAB searching path.

Installation and Usage:

1. General instructions for downloads and installation of CurveAlign or CT-FIRE

2. Quick Manual and testimages

3. CurveAlign FAQs and CT-FIRE FAQs

4. Initial protocol for alignment quantification

References:

CurveAlign

1. Schneider, C.A., Pehlke, C.A., Tilbury, K., Sullivan, R., Eliceiri, K.W., and Keely, P.J. (2013). Quantitative Approaches for Studying the Role of Collagen in Breast Cancer Invasion and Progression. In Second Harmonic Generation Imaging, F.S. Pavone, and P.J. Campagnola, eds. (New York: CRC Press), p. 373.

2. Bredfeldt, J.S., Liu, Y., Conklin, M.W., Keely, P.J., Mackie, T.R., and Eliceiri, K.W. (2014). Automated quantification of aligned collagen for human breast carcinoma prognosis. J Pathol Inform 5.

3. Liu, Y., Keikhosravi, A., Mehta, G.S., Drifka, C.R., and Eliceiri, K.W. (2017). Methods for quantifying fibrillar collagen alignment. In Fibrosis: Methods and Protocols, L. Rittié, ed. (New York: Springer) p.429-451

4. Liu Y, Keikhosravi A, Pehlke CA, Bredfeldt JS, Dutson M, Liu H, Mehta GS, Claus R, Patel AJ, Conklin MW, Inman DR, Provenzano PP, Sifakis E, Patel JM, Eliceiri KW. (2020) Fibrillar collagen quantification with curvelet transform based computational methods. Front Bioeng Biotechnol 8:198.

CT-FIRE

5. Bredfeldt, J.S., Liu, Y., Pehlke, C.A., Conklin, M.W., Szulczewski, J.M., Inman, D.R., Keely, P.J., Nowak, R.D., Mackie, T.R., and Eliceiri, K.W. (2014). Computational segmentation of collagen fibers from second-harmonic generation images of breast cancer. Journal of Biomedical Optics 19, 016007–016007.

6. Stein, A. M., Vader, D. A., Jawerth, L. M., Weitz, D. A. & Sander, L. M. An algorithm for extracting the network geometry of three-dimensional collagen gels. J. Microsc. 232, 463–475 (2008).

Fast Discrete Curvelet Transform

7. Candes, E., Demanet, L., Donoho, D. & Ying, L. Fast discrete curvelet transforms. Multiscale Model. Simul. 5, 861–899 (2006).

Lab page links

CurveAlign at the website of Eliceiri lab

CT-FIRE at the website of Eliceiri lab

curvelets's Issues

CT-fire V2.0 installation issue in macOS Catalina and I am using MATLAB 2020a

I am using macOS Catalina trying to install CT-fire using it with MATLAB 2020a, but I am facing an installation error, is it because I am using the 2020a version, or is it from macOS Catalina, if there is a solution for my issue please contact me.

Use the uifigure UserData to share data among Callbacks instead of global variable

More details can be seen https://www.mathworks.com/help/matlab/creating_guis/share-data-among-callbacks.html

how to make layers in an image using cuvrelets transformation

please if any one know how to group elements after applying curvelets transformation to create layer by layer image.. n again in need to repack the image by adding them, i need to delete layer for finding snr,, please help

CT-FIRE on macOS Catalina

Hello everyone,
How can I download CT-Fire Standalone application on macOS Catalina, without facing any problem.

Bio-Formats GUI

Incorporate the BFinMatlabFiugreSlider.m (visualization function) into the bioFormatsMatlabGUI function after loading an image. Use structure "BFcontrol" to contain all the control parameters read or set by the BF GUI. More details can refer to the currently hardwired parameters in BFinMatlabFiugreSlider.m function. The colormap option, uieidtfields on the BF GUI can be kept and the corresponding values can be passed to the visualization function.

bfsave in Bio-Formats matlab library saves an incorrect BitDepth information

When using "bfsave" to split channels or timepoints, the resulted images of separate channels or timepoints have a BitDepth that is different from the original image. This error can be reproduced in Bio-Formats MATLAB importer and exporter GUI by following steps:
(1) Import a multi-channel image of 8-bit or 16-bit image.

(2) Choose "Each Channel" in the "Export" panel and then click "Save"

(2) Click the "Individual Image" to import the saved image in step (2)

The OME-XML metadata shows that SignificantBits="64"

Compile and release the visualization module developed in Matlab App Designer

Create a standalone release of this module in the visualization branch and release it.

Compile and test the latest version of CurveAlign and CT-FIRE using the latest Mac OS system

To check the potential Mac OS compatibility issue.

Version control of code in Matlab App Designer

App designer code is binary file and is not straight forward to version control it. Need to find a practical way for its version control.

Mask Functionality?

How do I get the CT partial reconstruction image without the border, axis and subtitles when running CT-FIRE? I am looking to create a mask of the collagen-rich regions of my image.

Additionally, is there any sort of functionality that outputs the location/ coordinates of specific fiber regions? I want to know where certain fibers are.

Re-organize/design the layout of the GUI of the intersection point(IP) detection module

1. Put all the buttons to the left column.
2. Make the coordinates table occupy the full column from top to bottom of the figure.
3. Refer to the ROI manager of Fiji/CurveAlign to design the button function.
4. Add two check boxes to show/not show "IP Index" and "all IP points".
5. Add a button to load an image with the CT-FIRE output .mat file. This can make this module do a post-processing of CT-FIRE results.
6. Add a dropdown menu to list all IP detection methods.

Restructure source directory

what we want for curvelets repository:

• src directory
  • ctfire directory
  • curvealign directory

These source directories should contain the minimum amount of stuff so that all we need to do is:

1. copy CurveLab to one of these source directories on local machine
2. run mcc -m -a ./ CurveAlign.m ormcc -m -a ./ ctFIRE.m

or for headless executables we use goCTFK.m or goCAK.m

So we can document this in a README.md

Table refresh in IP GUI is slow.

Need to find a way to make the visualization of the X-Y list faster.

Test the functionality of all the current Bio-Formats GUI features

Use the images of different formats/dimensions.

Add the data related to IP detection to a new field of the CT-FIRE output .mat file

If IP detection is conducted on an image processed by CT-FIRE, add a new field in the CT-FIRE .mat file for all the data related to IP points including methods, operations, and parameters.

Evaluate UDCT vs. FDCT and other possible options

The Uniform Discrete Curvelet Transform (UDCT) is a possible alternative to the Fast Discrete Curvelet Transform (FDCT). There is also a review paper from 2010 on curvelets.

The UDCT is appealing because there is a partial open-source implementation written in Julia on GitHub. We could easily port this to Java, MATLAB, and/or other useful platforms. Whereas the flagship FDCT implementation, CurveLab, is unfortunately not open source.

Major goals of this evaluation are:

• Possible improvements in both performance and results/correctness.
• Deploy curvelet-based analysis routines in ImageJ and other OSS tools which give scientists improved control over their in toto workflow.

Set proper figure properties for the IP detection GUI.

1. The figure should be plotted in a fixed given window/area.
2. Should enable the operations of the figure such as zoom, move, etc.

Warning message starting CurvAlign in Matlab R2014b on Linux (Ubuntu 16.04)

When you type 'CurveAlign' into command window to start the GUI MatLab warns the following.

Warning: Invalid file or directory '<HOME PATH/CURVEALIGN PATH>/CurveAlign
V4.0 Beta sourcecode for August 31 2017
release/20130227_xlwrite/poi_library/dom4j-1.6.1.jar'. 
> In javaclasspath>local_validate_dynamic_path at 274
  In javaclasspath>local_javapath at 182
  In javaclasspath at 119
  In javaaddpath at 71
  In CurveAlign at 57 

1. Can the warning be fixed or suppressed if not critical?
2. Is this only related to Linux/Mac builds or is it an issue for Windows?

Redesign the functionality of some buttons/operations in IP detection module.

1. "Reset" should make the module start over .
2. Any operation (deletion, addition, combination) of the X-Y table should update the corresponding data file.
3. Currently selected one or more IPs in the X-Y table should be visualized in the figure.
4. Add a button to set the properties of the marker/index size and color.

MATLAB save a wrong physical size of the pixel

When using "imwrite" to save an 8-bit image, the physical size of X and Y can not be set by the user. It was to set to a fixed value by the system (in Windows 10, the value was 352. 778) . This error can be reproduced in Bio-Formats MATLAB importer and exporter GUI by following steps:
(1) Choose "Each Plane MATLAB 8-bit" in the "Export" panel and then click "Save"

(2) Click the "Individual Image" to load the saved image in (1)

The Pixel size will be shown to be 352.778.