Fiber Segmentation

I developed a fiber segmentation script in MATLAB specifically for analyzing stress fibers in airway epithelial cell microscopy images. But the algorithm is likely broadly applicable to any images containing “fiber-like” objects.

I developed the segmentation code in MATLAB by Frankensteining methods from two published stress fiber segmentation methods developed by Rogge+ 2017 and Zhang+ 2017.

Briefly, I implemented a fiber tracing algorithm from by Rogge+ 2017 to segment a pool of stress fiber fragments. Then, I implemented a fragment pairing algorithm based on Zhang+ 2017 to connect longer fibers. The segmentation function segmentfibers.m is available on the repository SegmentFibers.

Creating Synthetic Fiber Images

To test the segmentation algorithm, we generated synthetic stress fiber images with known fiber lengths and widths that are randomly aligned. To mimic stress fiber data, we included an option for generating locally aligned fibers. The code Generate_FiberImage.m in the repository also provides options to adjust the image size and resolution.

Two sample images are shown below:

segmentation example