
- Current code fits slice to 1 2-D Gaussian, the center of which may be between to bright spots in the image.

First part--Get the locations in the volume:
(done) - threshold: segmentation into foreground and background. Normalize image such that highest pixel value is 1, lowest is 0. Potential threshold is likely 0.3-0.5.
(done) - Reject areas smaller than a prescribed area in square pixels.
(done) - Identify foreground objects (assume circular/round)
(50%) - Compute the inteensity-weighted centroid of each foreground object
- Dialate objects to larger size (likely 2x the object diameter).
- Fit 2D Gaussian to each object to optimize center location. Output here is the position in X and Y.
(deprecated) - Connect centroids/Gaussians between slices by computing the shortest distance between objects in first frame to objects in the second frame.
- look into smoothing to identify and segment volumes better.
- pitch, radius, & handedness
- ultimate output: 

Second part--Visualize, parameterize, quantify results
- Fit resulting positions to extract pitch/diameter
- Visualize, maybe with isosurfaces + isosurface mesh like in the paper Jay sent.

- Notes: It might be a good idea to get a library of the foreground objects just to see the variability in their shapes. If necessary for filtering purposes, computing the circularity here might also be a good metric.
