Multichannel analytics development
From FarsightWiki
(Difference between revisions)
(Created to document project progress as-it-happens and encourage people to actually touch the wiki) |
|||
| (7 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
| + | =March 2011= | ||
| − | ==August 2011== | + | =April 2011= |
| + | |||
| + | |||
| + | =July 2011= | ||
| + | |||
| + | =August 2011= | ||
| + | |||
| + | The Roysam lab focused on pipeline integration in order to reveal kinks in the planned workflow. | ||
| + | |||
| + | == Image Preprocessing Additions == | ||
| + | |||
| + | *A background subtraction step to remove the low-frequency noise and large imaging artifacts from the raw images. | ||
| + | *A normalization step to equalize foreground contrast and correct for changes in overall brightness as a function of depth. | ||
| + | |||
| + | These additions allowed us to montage a few data sets which had previously defied automated registration. | ||
| + | |||
| + | == Tiling Functions for Microglia Tracing and Nuclei Segmentation: == | ||
| + | |||
| + | *Montage can be split into smaller tiles for processing. | ||
| + | *Workflow of individual tiles can be distributed more easily: | ||
| + | **Tracing microglial tiles | ||
| + | **Segmentation and classification of nuclei tiles | ||
| + | *Resulting files can be stitched back into global coordinates automatically, since the output of each tile only needs a static translation. | ||
| + | |||
| + | == Soma Detection and Segmentation Improvements == | ||
| + | |||
| + | *Bug: previous method of morphological functions on the microglia channel missed some somata. | ||
| + | *Fix: combining the DAPI channel with the microglia channel reduced extrasomatic noise, allowing us to use more relaxed parameters to identify somata. | ||
| + | |||
| + | =September 2011= | ||
| + | |||
| + | == Stress Test - Data 9/12 == | ||
| + | |||
| + | *Largest to-date data set (~48 GB per channel) processed | ||
| + | **Minor bug relating to data registration uncovered, but effort largely successful | ||
| + | |||
| + | =October 2011= | ||
| + | |||
| + | == Pipeline Improvements == | ||
| + | |||
| + | *Added a single-tile analysis step for quick verification of batch processing validity. | ||
Latest revision as of 21:00, 11 October 2011
Contents |
March 2011
April 2011
July 2011
August 2011
The Roysam lab focused on pipeline integration in order to reveal kinks in the planned workflow.
Image Preprocessing Additions
- A background subtraction step to remove the low-frequency noise and large imaging artifacts from the raw images.
- A normalization step to equalize foreground contrast and correct for changes in overall brightness as a function of depth.
These additions allowed us to montage a few data sets which had previously defied automated registration.
Tiling Functions for Microglia Tracing and Nuclei Segmentation:
- Montage can be split into smaller tiles for processing.
- Workflow of individual tiles can be distributed more easily:
- Tracing microglial tiles
- Segmentation and classification of nuclei tiles
- Resulting files can be stitched back into global coordinates automatically, since the output of each tile only needs a static translation.
Soma Detection and Segmentation Improvements
- Bug: previous method of morphological functions on the microglia channel missed some somata.
- Fix: combining the DAPI channel with the microglia channel reduced extrasomatic noise, allowing us to use more relaxed parameters to identify somata.
September 2011
Stress Test - Data 9/12
- Largest to-date data set (~48 GB per channel) processed
- Minor bug relating to data registration uncovered, but effort largely successful
October 2011
Pipeline Improvements
- Added a single-tile analysis step for quick verification of batch processing validity.
