Summary

• Page properties (2 modified, 0 added, 0 removed)

Details

Page properties

Author

...	...	@@ -1,1 +1,1 @@
1		-XWiki.sharoncy
	1	+XWiki.puchades

Content

...	...	@@ -1,19 +1,19 @@
1	1	== (% style="color:#c0392b" %)Image pre-processing with Nutil Transform(%%) ==
2	2
3	3
4		-**The //Transform// feature in the //Nutil// software **enables **image rotation, renaming, resizing and mirroring** and is used to prepare the images in the series for //QuickNII //alignment and //ilastik //segmentation.
	4	+**The //Transform// feature in the //Nutil// software **enables **image rotation, renaming, resizing and mirroring** and is used to prepare the image series for //QuickNII //alignment and //ilastik //segmentation.
5	5
6		->(% style="color:#27ae60" %)The //QuickNII// and //ilastik// software have different input image size requirements. It is also important that the input images follow the prescribed naming convention.
	6	+>The image input size requirements for the //QuickNII// and //ilastik// software differ. Naming convention of files is important.
7	7
8		-For //QuickNII,// the input requirements described in Puchades et al., 2019 are:
	8	+For //QuickNII,// the input requirements described in (Puchades et al., 2019) are:
9	9	24-bit PNG and JPEG. Images can be loaded up to the resolution of 16 megapixels (e.g.4000x4000 or 5000x3000 pixels), however QuickNII does not benefit from image resolutions exceeding the resolution of the monitor in use. For a standard FullHD or WUXGA display
10	10	(1920x1080 or 1920x1200 pixels) the useful image area is approximately 1500x1000 pixels,using a similar resolution ensures optimal image-loading performance.
11	11
12		-For //ilastik// (Borg et al.2019) the histological images are downscaled in order to enable efficient processing. The pixel classification algorithm relies on input from manual user annotations of the training images, and the features ‒ intensity, edge and/or texture ‒ of the image pixels. The  resizing factor is determined by trial and error, with a test run performed with //ilastik// on images of different sizes to determine the optimal resolution for segmentation. As an example, in Yates et al, 2019, the image were downscaled by a factor of 0.1 and 0.05 for cellular features and Alzheimer's plaques respectively (factor applies to image width).
	12	+For //ilastik// (Borg et al.2019), the resizing was performed in order to enable efficient processing. To clarify, the pixel classification algorithm relies on input from manual user annotations of training images, and the features ‒ intensity, edge and/or texture ‒ of the image pixels.  In practice, a test run was performed with //ilastik// on images of different sizes to find the optimal resolution for segmentation, with a final resize factor of e.g 0.1  for cellular features, and a factor of 0.05 for series with larger features like Alzheimer plaques.
13	13
14	14	=== (% style="color:#c0392b" %)Naming convention(%%) ===
15	15
16		-For QUINT analysis, Nutil Quantifier extracts objects from segmentations and registers them to areas defined in customised atlas maps. To match segmentations and corresponding atlas maps together, Nutil Quantifier relies on a unique ID in the file name of both files.The ID is unique to a particular section and should be in the format: sXXX.., with XXX.. representing the section number. The section number should reflect the serial order and spacing of the sections (e.g. s002, s006, s010 for every 4^^th^^ section starting with section 2).
	16	+For running the QUINT analysis the segmentation files and the customized atlas maps need to match and therefore the segmentation files should follow the same naming format.The file name that correspond to a particular section must contain a unique ID in the format: sXXX.., with XXX.. representing the section number. The section number should reflect the serial order and spacing of the sections (e.g. s002, s006, s010 for every 4^^th^^ section starting with section 2).
17	17
18	18	Example: tg2345_MMSH_s001_segmentation.png
19	19