Skip to main content

Module for creating cochleograms from confocal images

Project description

Introduction

This facilitates creating cochleograms from confocal images and will export a frequency map and inner/outer hair cell positions along the tonotopic axis. If position information is stored in the file, it can be used to automatically align multiple confocal z-stacks for a single piece.

Datasets

Right now only Leica LIF files are supported. At one time we had support for the Zeiss CZI format but it needs to be updated to work with the current version of the program.

Naming conventions for LIF files

Files should be named with the identifier (e.g., animal ID and ear) followed byt the label for each channel in the order the channels were imaged. For example:

B009-8L-CtBP2-MyosinVIIa-PMT

Inside each file, the pieces must be numbered sequentially from base (hook) to apex. If more htan one image is required for a piece, use letters for the suffixes (i.e., “piece_2a”, “piece_2b”, etc.). The order in which the images for a single piece are labeled does not matter since the program will automatically align them based on the stage coordinates stored in the file.

Missing pieces

If a piece is missing, you can copy the image stacks for a matching piece from another file (for LIF files, you can use LAS X Office). To indicate that the piece is a copy, it must have the suffix copied_<note>. For example, if you copy piece_4a and piece_4b from B009-8L to the file containing data for B021-3L, the copied images should be named “piece_4a_copy_B009-8L” and “piece_4b_copy_B009-8L”. The note will appear in the composite generated for the frequency map.

Using the program

Mouse interaction

left click

Select tile

left click + drag

Pan image

mouse wheel

Zoom in/out

Keyboard shortcuts

t

Switch to tile mode

i

Switch to IHC mode

1

Switch to OHC1 mode

2

Switch to OHC2 mode

3

Switch to OHC4 mode

4

Switch to extra mode

s

Select spiral tool

e

Select exclude tool

c

Select cell tool

n

Select next tile (tile mode only)

p

Select previous tile (tile mode only)

arrow keys

The behavior of the arrow keys will depend on whether tile mode is selected. If tile mode is selected, then the arrow keys will move the tile. If any other mode is selected, the arrow keys will pan the image (this can be useful when in spiral or cell mode to move through the cochlea when zoomed in). To move the tile (or pan the image) in smaller steps, hold down shift at the same time.

Analysis

Analysis requires the following steps:

  • Aligning the tiles so that they overlap as accurately as possible.

  • Tracing a spiral through each row of hair cells.

  • Marking individual hair cells.

  • Marking regions containing uninterpretable data.

Tools are provided to facilitate each step. Be sure that you are satisfied with the result of the current step before moving to the next step. Although you can go back and edit a previous step, it may affect your analysis (e.g., if you need to move a tile after marking hair cells, you may have to manually edit a large number of hair cells).

Tile mode

Start by selecting “tiles” from the edit buttons, then left-clicking to select the tile that is misaligned. Using the arrow keys, you can move the tile until it is properly aligned with the other tiles. If you need to move the tile in smaller steps, hold down the shift key at the same time as the arrow keys. It may be helpful to toggle “highlight selected” so that you get a transparent overlay. When in “highlight selected” mode, the currently selected tile will be shown with a red border.

left click

Select tile

mouse wheel

Zoom in/out

arrow keys

Move currently selected tile (large steps)

shift + arrow keys

Move currently selected tile (small steps)

n

Select next tile

p

Select previous tile

An “align tiles” tool is provided to facilitate this step. It uses an automated algorithm that attempts to align the tiles based on the correlation between the images (using the MyosinVIIa channel).

Spiral mode

Once you are satisfied with the alignment of the tiles, select “IHC” from the edit buttons and be sure the spiral tool to the right of the edit buttons are selected. The very first point you mark should be on the end of the row of hair cells facing the most basal region of the cochlea. This point will be highlighted with a red circle. If you realize you made a mistake, you can select a different point as the start of the spiral by control + right-clicking that point when in spiral mode.

You must select a minimum of four points to create the spiral. You can add points in between existing points and the spiral will be rerouted through those points. The algorithm assumes that the “next” point in the path is the one closest to it (i.e., the order in which you add the points does not matter).

Repeat the process for OHC1, OHC2, and OHC3. Be sure that the spiral bisects the nuclei (IHCs) or cuticular plate (OHCs) as that will facilitate the semi-automated algorithms implemented by the program to help mark hair cells.

right click

Add point

shift + right click

Remove point

control + right click

Set point as origin for spiral

Cell mode

After marking the spiral, run the algorithm to automatically detect cells. You can play with the settings (each time you run, it will delete the existing cells and create new ones). You will likely have to manually edit the automatically-detected cells. Select the cell tool and then use right click to add cells and shift + left click to delete cells.

right click

Add cell

shift + right click

Remove cell

From time to time there will be a fourth row of OHCs. These should manually be identified by selecting “extra” for the cell you would like to edit and then adding the cells using the cell tool. Since the fourth row tends to be very short in length, you cannot mark a spiral or mark the region as excluded.

Exclude mode

Finally, go back through each row of hair cells. If there was a region you felt you could not intepret properly, select the exclude tool. Right-click the spiral at one end of the region then right-click again at the other end of the region you wish to exclude.

right click

Start region. Click again to end region.

shift + right click

Remove region under mouse cursor.

escape

Cancel current region.

Some additional tools are made available to facilitate this process:

  • You can merge all excluded regions across the OHC spirals into a single set of excluded regions that apply to all OHC spirals (Combine OHC exclusions button).

  • You can simplify a set of excluded regions for a particular spiral if they are overlapping (this will combine overlapping exclusion regions into a single exclusion region) using the Simplify exclusions button.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

cochleogram-0.5.2.tar.gz (61.3 kB view details)

Uploaded Source

Built Distribution

cochleogram-0.5.2-py3-none-any.whl (58.6 kB view details)

Uploaded Python 3

File details

Details for the file cochleogram-0.5.2.tar.gz.

File metadata

  • Download URL: cochleogram-0.5.2.tar.gz
  • Upload date:
  • Size: 61.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.11.5

File hashes

Hashes for cochleogram-0.5.2.tar.gz
Algorithm Hash digest
SHA256 097fd068bc24a087567bd1168f550061a92688aff380d20c014e8e006fb230f8
MD5 8dede825e641215f667bfce915f7d1ea
BLAKE2b-256 8ab6f701eca055d4219c60f0686fd64f7e368767debb83321ec6cfebc3180319

See more details on using hashes here.

Provenance

File details

Details for the file cochleogram-0.5.2-py3-none-any.whl.

File metadata

  • Download URL: cochleogram-0.5.2-py3-none-any.whl
  • Upload date:
  • Size: 58.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.11.5

File hashes

Hashes for cochleogram-0.5.2-py3-none-any.whl
Algorithm Hash digest
SHA256 46b612a6d586bfecd126a4189849188024993825a6c7f63595bdc7f500d0ddb5
MD5 571e99465c783a07e43d628ae33501ae
BLAKE2b-256 dbe8d437fa6147b2d72afc101eb35cbb7235c77191c6940f5d8ba75a3366b1ef

See more details on using hashes here.

Provenance

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page