Skip to content

Latest commit

 

History

History
36 lines (19 loc) · 3.58 KB

File metadata and controls

36 lines (19 loc) · 3.58 KB

Methods Supplement (Draft)

Paste-ready methods text for vesicle/GUV and RBC shape analysis with MorphoStack v0.1.0.
Replace bracketed placeholders with instrument-specific values before submission. Read limitations.md first.

Image preprocessing

Microscopy Z-stacks in TIFF, LSM, or CZI format were loaded in MorphoStack (v0.1.0). Multi-channel or color stacks were converted to grayscale for analysis. Slices were optionally cropped to a rectangular region of interest and trimmed to a selected Z-frame range [zmin, zmax). Voxel spacing was taken from file metadata when available; otherwise manual overrides of [vx, vy, vz] µm were applied after verification against the acquisition log. Analyses that retained the software default spacing of 1 × 1 × 1 µm were treated as uncalibrated and were not used for absolute biological dimensions.

Object selection (crowded fields)

When multiple objects appeared in the field of view, a user-defined seed (circular neighborhood or polygon) on a chosen frame constrained connected-component selection. The selected component was tracked across Z using overlap-first association with centroid-distance fallback. Tracking diagnostics (lost frames, ROI boundary contact, likely neighbor merge) were recorded in the run manifest.

Segmentation and 2D metrics

For vesicle and RBC profiles, each frame was intensity-thresholded and an object contour was extracted. Contour area and perimeter were computed from the polygon representation with anisotropic pixel spacing applied along X and Y. Circularity was evaluated as the isoperimetric quotient (4\pi A / P^2). Additional descriptors included bounding-box aspect ratio, elongation, deformation index, extent, equivalent diameter, and solidity (area relative to convex hull). Frame-level metrics were exported as CSV.

3D surface reconstruction

When enabled, per-frame binary masks derived from contours were stacked and converted to a triangular surface mesh using marching cubes with voxel spacing ((v_z, v_y, v_x)). Reported mesh surface area, volume, equivalent-sphere diameter, and sphericity are computational geometry outputs in micrometers only when voxel calibration was verified. Mesh files were exported as OBJ, STL, PLY, or GLB as needed.

Active surfaces (optional, experimental)

For cases where global thresholding merged neighboring objects or failed on weak membrane signal, an experimental active-surfaces (surfel) profile was optionally applied after seeding. Active-surfaces outputs were compared to threshold-based meshes on the same seed before use in figures. Parameter defaults and validation status are documented in the MorphoStack active-surfaces notes.

Quality control and reproducibility

Frames with lost tracking, ROI boundary contact, or likely neighbor merges were flagged in run warnings. Individual frames could be excluded and the analysis re-run. Each export recorded source path, SHA-256 checksum of the source file, MorphoStack version, profile, threshold, voxel source, ROI, Z range, object seed, excluded frames, and warning codes in a JSON manifest alongside optional Markdown reports.

Limitations statement (short form)

Absolute area, volume, and sphericity require validated microscope calibration. Default voxel spacing is a computational placeholder. The RBC profile currently shares the vesicle threshold-contour engine; RBC-specific morphometrics require separate validation. Active-surfaces segmentation is experimental on crowded or touching objects.

Related

metrics.md · citations.md · validation.md