Quantifying Fibrillar Collagen Organization with Curvelet Transform-Based Tools

This tool helps to quantify changes in fibrillar collagen organization in the extracellular matrix related to many diseases on a global region of interest and individual fiber basis. This tool uses the curvelet transform, an optimal multi-scale directional image representation method to remove image noises and to enhance fiber edges. Comprehensive and user-friendly modules enable an efficient quantification.

For individual fiber extraction, follow the protocol to open the CT-FIRE and click open files to select one or more images of interest. In the parameters panel, click update and adjust the background threshold and the nucleation searching radius. When all of the parameters have been set, click run.

After the analysis is done, click any element in the output table to visualize the results and check the ctFIREout subfolder for the output files. To analyze a region of interest, click open files to load one or more images and select the image to be annotated. In the run options panel, select region of interest manager in the dropdown menu and click run to launch the region of interest manager module.

Next, click the draw region of interest menu D dropdown menu and draw the regions of interest. If full image CT-FIRE analysis has been conducted and the results are saved in the default directory, click one or more regions of interest in the region of interest list and click CT-FIRE region of interest analyzer to launch the post region of interest analysis module. If full image CT-FIRE analysis has not been conducted, click one or more of the regions of interest in the region of interest list and click apply CT-FIRE on ROI to directly apply the CT-FIRE analysis to the selected regions of interest.

For advanced post-processing, check the out. adv checkbox and click post-processing to launch the advanced post-processing graphic interface. In the analysis module, click select file and visualize fibers to allow the fiber number to be entered based on the labels in the original fibers tab.

Click confirm update to move to the thresholding operation and check the thresholding box to enable the threshold settings. Select one of the four thresholding options from the dropdown menu and enter the desired thresholds in the thresholds panel for one or more fiber properties. Then click threshold now to apply the above thresholding conditions and check the prompt figure with the name ending with metrics visualization to see the selected fibers overlaid on the original images with the customized color maps.

When the threshold has been set, click save fibers to save the selected fiber information and click generate stats and click OK to generate the summary statistics. To combine results from multiple images, check the batch mode and select multiple images to be analyzed before generating the summary statistics as demonstrated. To extract CT-based fiber features, follow the protocol to open the CurveAlign software and click get images in the main graphic user interface and select one or more images or image stacks from the prompt window.

Enter the fraction of coeffs to keep value, the fraction of the largest coefficients of CT that will be used in the fiber analysis, and click run. Repeat to set a desirable threshold. When the process is complete, some summary statistics for each image will be displayed in an output table and all of the output files will be automatically saved in a subfolder named CA_out in the directory of the original images.

To extract CT-FIRE based fiber features, click the fiber analysis method dropdown menu and select CT-FIRE fibers to use the fiber center point and fiber angle to represent the fiber. For relative alignment analysis of the tumor boundary, select the TIFF boundary condition. Open the CurveAlign software and click get images in the main graphic user interface and select one or more images or image stacks from the prompt window.

In the primary parameters panel, enter the distance from the closest boundary to only evaluate the fibers within this distance range. In the output options panel, check the boundary association box to allow visualization of the point on the boundary that is associated with a fiber, fiber segment, or curvelet and click run, then click the output table to visualize the results. If a full image CurveAlign analysis has been conducted and the results are saved in the default directory, click one or more regions of interest in the region of interest list and click CurveAlign region of interest analyzer to run a post region of interest analysis.

If a full image CurveAlign analysis has not been conducted, click one or more regions of interest in the region of interest list and click apply CurveAlign on region of interest to directly apply the CurveAlign analysis to the selected regions of interest. In this image, an overlay image of a tissue microarray core can be observed. In the magnification, the fibers parallel and perpendicular to the tumor boundary within the region of interest can be observed.

The CurveAlign output overlay image shows the tumor boundaries and the representative fiber locations and orientations. This method can be used to quantify not only collagen fibers, but also other line-like structures such as elastin and can potentially facilitate early diagnosis and prognostic analysis of cancer.