Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Quality Control

This chapter presents the visual quality control (QC) review of all axon segmentations across the 8 corpus callosum regions. Every axon detected by AxonDeepSeg was visually inspected and either retained or excluded.

Methodology

For each image, the following QC pipeline was applied:

  1. Morphometrics computation with instance segmentation using AxonDeepSeg v5.4.0

  2. Automated flagging based on: border-touching, invalid g-ratio, no myelin, sub-pixel artifacts (d < 0.05 um), myelin-dominant (g < 0.3), low solidity (< 0.75), myelin >> axon (mt/ad > 1.0)

  3. Visual review of every axon closeup, confirming auto-flags and checking for missed problems

  4. Exclusion decision with per-axon justification

Overall summary

Exclusions by category

Per-image summary

SampleSliceTotalExcludedRetained% Excluded
Region 114966043612.1%
Region 134457237316.2%
Region 174969440219.0%
Region 213084626214.9%
Region 233695231714.1%
Region 253645530915.1%
Region 312765721920.7%
Region 331666610039.8%
Region 352765921721.4%
Region 41144836157.6%
Region 451807210840.0%
Region 472007812239.0%
Region 512147014432.7%
Region 532166714931.0%
Region 552817021124.9%
Region 612716820325.1%
Region 632798619330.8%
Region 653318125024.5%
Region 713297225721.9%
Region 733238024324.8%
Region 753226226019.3%
Region 811956812734.9%
Region 832557518029.4%
Region 8589256428.1%

Based on the visual review, the following automated filters capture all exclusions:

  1. image_border_touching == True

  2. gratio >= 1.0 or gratio <= 0

  3. myelin_area <= 0 or myelin_thickness <= 0

  4. axon_diam < 0.05 um

  5. gratio < 0.3

  6. solidity < 0.75

  7. myelin_thickness / axon_diam > 1.0

The full exclusion database is available as exclusion_database.csv.

Per-sample reports