Wiki source code of 4. Image segmentation with ilastik
Last modified by annedevismes on 2021/06/08 11:56
Hide last authors
| author | version | line-number | content |
|---|---|---|---|
 |
7.1 | 1 | == [[image:ilastik_logo.PNG||style="float:right"]] == |
| |
1.1 | 2 | |
| |
20.2 | 3 | == [[image:Pixel_classification workflow.png||style="float:left"]](% style="color:#c0392b" %)Analysis approach for series of rodent-brain section image(%%) == |
| |
7.1 | 4 | |
| |
20.2 | 5 | //Ilastik //is a versatile image analysis tool specifically designed for the classification, segmentation, and analysis of biological images based on supervised machine-learning algorithms. |
| |
8.1 | 6 | |
| |
20.2 | 7 | **There are two main approaches for the analysis of rodent-brain section image~:** |
| |
15.1 | 8 | |
| |
20.2 | 9 | 1. pixel classification only (with two or more classes) and |
| 10 | 1. pixel classification with two classes (//immunoreactivity// and //background//), followed by Object classification with two classes (//objects of interest// and //artefact//). | ||
| |
1.1 | 11 | |
| |
1.5 | 12 | **Which approach is best for my dataset?** |
| 13 | |||
| |
20.2 | 14 | As a general rule, pixel classification is suitable for images in which there are clear differences in the colour, intensity, and / or texture of the feature of interest (labelling) versus the background and other structures. If there is non-specific labelling in the image that is very similar in appearance to the labelling of interest, object classification may allow the non-specific labelling to be filtered out on the basis of the object-level features such as shape and size. The best approach is determined by trial and error. |
| |
1.5 | 15 | |
| |
5.1 | 16 | === (% style="color:#c0392b" %)Pixel classification workflow(%%) === |
| |
1.5 | 17 | |
| |
11.1 | 18 | For a quick introduction, [[watch this video>>https://www.youtube.com/watch?v=5N0XYW9gRZY&feature=youtu.be]]. |
| |
1.1 | 19 | |
| |
20.2 | 20 | **Basic steps** |
| |
2.4 | 21 | |
| |
20.2 | 22 | * Train the classifier with two classes (labelling and background). |
| 23 | * Apply the classifier to the rest of the images (batch processing). | ||
| 24 | * Export the probability maps in HDF5 format and simple_segmentation images in PNG format with the default settings. | ||
| |
11.1 | 25 | * Review the results. |
| |
2.4 | 26 | |
| |
20.2 | 27 | === (% style="color:#c0392b" %)Object-classification workflow(%%) === |
| |
1.1 | 28 | |
| |
20.3 | 29 | There are three options on the //ilastik //start-up page for running "Object Classification." Choose the //Object Classification with Raw Data and Pixel Prediction Maps //as input. |
| |
1.1 | 30 | |
| |
20.2 | 31 | * Save the object-classification file in the same folder as the raw images for analysis. If the images are moved after the //ilastik// file is created, the link between the //ilastik //file and the images may be lost, resulting in a corrupted file. |
| 32 | * In the "Input Data" applet, upload the original images and their respective probability maps in HDF5 format (output from the Pixel Classification). | ||
| 33 | * Train the classifier with two classes (labelling and artefacts). | ||
| 34 | * In the "Object Information Export" applet, export “Object Predictions” in PNG format. Do not change the default export location. | ||
| |
11.1 | 35 | * Review the results. |