Project

General

Profile

Actions

Feature #1207

closed

Count hair cells in the cochlea

Added by Volker Baecker over 9 years ago. Updated over 8 years ago.

Status:
Closed
Priority:
Normal
Target version:
-
Start date:
10/28/2015
Due date:
% Done:

80%

Estimated time:

Description

The input image is a 3D stack with a fluorescent staining of the cells. The cells build a spiral of three layers of cells. On the inner side are supporting cells which are not hair cells.
They can eventually be eliminated by using a second staining specific to them.
The hair cells must be counted by each cylinder of a given length (200µ).


Files

count-cells-per-200-in-table.ijm (706 Bytes) count-cells-per-200-in-table.ijm Volker Baecker, 03/22/2016 05:57 PM
count_hair_cells_in_cochlea.ijm (1.86 KB) count_hair_cells_in_cochlea.ijm Volker Baecker, 04/07/2016 02:31 PM
Cochlea_Hair_Cell_Counting.ijm (3.95 KB) Cochlea_Hair_Cell_Counting.ijm Volker Baecker, 05/10/2016 03:15 PM
Cochlea_Hair_Cell_Counting.ijm (5.12 KB) Cochlea_Hair_Cell_Counting.ijm Volker Baecker, 05/10/2016 04:35 PM
Actions #1

Updated by Volker Baecker over 9 years ago

Preliminary work:
Done by Julien using Imaris:
J'ai utilisé Imaris Cell pour cela. Ses images n'ont qu'un marquage. On pourrait envisager d'en faire 2 (elles sont en train d'essayer) mais on prend le risque d'augmenter le poids de l'image et que cela ne soit plus gérable avec Imaris.

Donc, je sors un marquage « cellule » avec un isosurface-like. En fonction de son état (écrasé, endommagé), on sort donc la spirale en entier ou en morceaux.
Ensuite je distingue les points (vesicles inside cell) en utilisant le même marquage.
Bon, de là on va s'en sortir, ce n'est pas trop un pb.
J'aurais pu utiliser d'autres outils de Imaris XT pour cela (en gros faire un isosurface, détecter les spots et ensuite restreindre les spots aux objets de l'isosurface).

Ce que je ne sais pas faire c'est découper mon ruban en spirale en tronçons de 200um de long. Et c'est ce qu'elles veulent faire.
Je peux le faire de façon très artisanale et fastidieuse.
On pourrait donc envisager soit de partir des objets 3D que sort Imaris et de découper cela en tronçons 3D de 200um de long, soit de découper l'isosurface en tronçons 3D et de le rerentrer dans Imaris pour restreindre les spots à ces différents tronçons.

Actions #2

Updated by Volker Baecker over 9 years ago

Prelimnary work:
Done by Sylvain and Volker in Imaris and ImageJ:

We do not treat the problem of the supporting cells for the moment, they might be eliminated with the second staining.
  1. Use Imaris to detect spots
  2. Create a synthetic image from the spots (create a channel with constant value 65000 inside and 0 outside of the spots)
  3. Create iso-surface from the green channel - edit - mask all in red channel
  4. Transfer the synthetic image to ImageJ via the provided Imaris mechanism.
  5. Run a min filter with radius 1 to decrease the size of the objects and avoid that objects touch each other in the projection
  6. Calculate the MIP
  7. Threshold the image and create a mask
  8. Draw a line along the spiral in the projection and straighten the image
  9. Re-threshold the image
  10. Use the particle analyzer to add the selections of the hair cells to the roi-manager
  11. Measure the cells from the roi-manager
  12. Use a macro that counts the cells per x-distance of 200µ
Actions #3

Updated by Volker Baecker about 9 years ago

A second channel to mask the unwanted cells is now available.

  • wrote a macro that takes the measurements in the results table and counts by chunks of 200µ.
  • the result table is sorted by the x-coordinate beforehand using the macro from my workshop
  • The pre-processing must still be integrated:
  • z-projection
  • threshold (auto)
  • watershed (optional)
  • wait for user to make a selection
  • straighten
  • threshold (again because of interpolation when straighten)
  • analyze particles (add to manager and measure)
Actions #4

Updated by Volker Baecker about 9 years ago

  • % Done changed from 0 to 70
Actions #5

Updated by Volker Baecker about 9 years ago

Wrote the macro that does the pre-processing and the counting.
The user has to manually select the cochlea.

Actions #8

Updated by Volker Baecker over 8 years ago

  • Status changed from Assigned to Closed
Actions

Also available in: Atom PDF