Journal:Analyzing huge pathology images with open source software - Revision history

Shawndouglas: /* Notes */ Added cat.

2016-01-12T16:38:25Z

Notes: Added cat.

← Older revision		Revision as of 16:38, 12 January 2016
Line 322:		Line 322:
	[[Category:LIMSwiki journal articles (all)‎]]		[[Category:LIMSwiki journal articles (all)‎]]
	[[Category:LIMSwiki journal articles on health informatics‎‎]]		[[Category:LIMSwiki journal articles on health informatics‎‎]]
			[[Category:LIMSwiki journal articles on imaging informatics‎‎]]

Shawndouglas: Updated caption formats.

2015-12-28T22:40:40Z

Updated caption formats.

← Older revision		Revision as of 22:40, 28 December 2015
Line 41:		Line 41:
	[[File:Fig1 Deroulers DiagnosticPath2013 8.jpg\|700px]]		[[File:Fig1 Deroulers DiagnosticPath2013 8.jpg\|700px]]
	{{clear}}		{{clear}}
	<blockquote>'''Figure 1. A sample slide. (a)''': macroscopic view of the whole slide (the black rectangle on the left is 1x2 cm).~~<br />~~'''(b,c)''': Influence of the magnification on the quality of results. '''(b)''': a portion of the slide scanned at magnification~~<br />~~level 10x. The white contours show the result of an automatic detection of the dark cell nuclei with the ImageJ~~<br />~~software. A significant fraction of the cell nuclei is missed and the contours are rather pixelated. '''(c)''': the same~~<br />~~portion of the slide scanned at magnification 40x. The white contours show the result of the same automatic~~<br />~~detection. Almost all cell nuclei are detected and the shapes of the contours are much more precise.~~<br />~~Scale bar: 4 ''μ''m.</blockquote>		{\|
			\| STYLE="vertical-align:top;"\|
			{\| border="0" cellpadding="5" cellspacing="0" width="700px"
			\|-
			\| style="background-color:white; padding-left:10px; padding-right:10px;"\| <blockquote>'''Figure 1. A sample slide. (a)''': macroscopic view of the whole slide (the black rectangle on the left is 1x2 cm). '''(b,c)''': Influence of the magnification on the quality of results. '''(b)''': a portion of the slide scanned at magnification level 10x. The white contours show the result of an automatic detection of the dark cell nuclei with the ImageJ software. A significant fraction of the cell nuclei is missed and the contours are rather pixelated. '''(c)''': the same portion of the slide scanned at magnification 40x. The white contours show the result of the same automatic detection. Almost all cell nuclei are detected and the shapes of the contours are much more precise. Scale bar: 4 ''μ''m.</blockquote>
			\|-
			\|}
			\|}

	However, this technique involves the manipulation of huge images (of the order of 10 billions of pixels for a full-size slide at magnification 40x with a single focus level) for which the approach taken by most standard software, loading and decompressing the full image into RAM, is impossible (a single slice of a full-size slide needs of the order of 30 GiB of RAM). As a result, standard open-source software such as ImageJ<ref name="RasbandImageJ12">{{cite web \|url=http://imagej.nih.gov/ij/ \|title=ImageJ \|author=Rasband, W.S. \|date=2012}}</ref>, ImageMagick<ref name="IMSImage13">{{cite web \|url=http://www.imagemagick.org/ \|title=ImageMagick \|author=ImageMagick Studio, LLC \|date=2013}}</ref> or GraphicsMagick<ref name="GMGGraphics13">{{cite web \|url=http://www.graphicsmagick.org/ \|title=GraphicsMagick \|author=GraphicsMagick Group \|date=2013}}</ref> completely fails or is prohibitively slow when used on these images. Of course, commercially available software exists<ref name="GarciaRev09" />, but it is usually quite expensive, and very often restricted to a single operating system. It uses proprietary source code, which is a problem if one wants to control or check the algorithms and their parameters when doing image analysis for research.		However, this technique involves the manipulation of huge images (of the order of 10 billions of pixels for a full-size slide at magnification 40x with a single focus level) for which the approach taken by most standard software, loading and decompressing the full image into RAM, is impossible (a single slice of a full-size slide needs of the order of 30 GiB of RAM). As a result, standard open-source software such as ImageJ<ref name="RasbandImageJ12">{{cite web \|url=http://imagej.nih.gov/ij/ \|title=ImageJ \|author=Rasband, W.S. \|date=2012}}</ref>, ImageMagick<ref name="IMSImage13">{{cite web \|url=http://www.imagemagick.org/ \|title=ImageMagick \|author=ImageMagick Studio, LLC \|date=2013}}</ref> or GraphicsMagick<ref name="GMGGraphics13">{{cite web \|url=http://www.graphicsmagick.org/ \|title=GraphicsMagick \|author=GraphicsMagick Group \|date=2013}}</ref> completely fails or is prohibitively slow when used on these images. Of course, commercially available software exists<ref name="GarciaRev09" />, but it is usually quite expensive, and very often restricted to a single operating system. It uses proprietary source code, which is a problem if one wants to control or check the algorithms and their parameters when doing image analysis for research.
Line 86:		Line 93:
	[[File:Fig2 Deroulers DiagnosticPath2013 8.jpg\|700px]]		[[File:Fig2 Deroulers DiagnosticPath2013 8.jpg\|700px]]
	{{clear}}		{{clear}}
	<blockquote>'''Figure 2. A typical session using ImageJ and the NDPITools plugins'''. A NDPI file has been opened with the NDPITools plugins~~<br />~~and it is displayed as a preview image (image at largest resolution which still fits into the computer’s screen) — top window. A~~<br />~~rectangular region has been selected and extracted as a TIFF image, then opened — bottom window.</blockquote>		{\|
			\| STYLE="vertical-align:top;"\|
			{\| border="0" cellpadding="5" cellspacing="0" width="700px"
			\|-
			\| style="background-color:white; padding-left:10px; padding-right:10px;"\| <blockquote>'''Figure 2. A typical session using ImageJ and the NDPITools plugins'''. A NDPI file has been opened with the NDPITools plugins and it is displayed as a preview image (image at largest resolution which still fits into the computer’s screen) — top window. A rectangular region has been selected and extracted as a TIFF image, then opened — bottom window.</blockquote>
			\|-
			\|}
			\|}

	[[File:Fig3 Deroulers DiagnosticPath2013 8.jpg\|600px]]		[[File:Fig3 Deroulers DiagnosticPath2013 8.jpg\|600px]]
	{{clear}}		{{clear}}
	<blockquote>'''Figure 3. Preview image of a NDPI file with several focalization levels in ImageJ'''. The NDPI file <tt>08.ndpi</tt~~><br /~~>contains images at 5 different focalization levels. Therefore, its preview image is displayed as a stack of 5 images.</blockquote>		{\|
			\| STYLE="vertical-align:top;"\|
			{\| border="0" cellpadding="5" cellspacing="0" width="600px"
			\|-
			\| style="background-color:white; padding-left:10px; padding-right:10px;"\| <blockquote>'''Figure 3. Preview image of a NDPI file with several focalization levels in ImageJ'''. The NDPI file <tt>08.ndpi</tt> contains images at 5 different focalization levels. Therefore, its preview image is displayed as a stack of 5 images.</blockquote>
			\|-
			\|}
			\|}

	When producing a mosaic, the user can request that pieces be JPEG files. Since the <tt>File > Open</tt> command of versions 1.x of ImageJ is unable to open TIFF files with JPEG compression (one has to use plugins), this is way to produce mosaics which can be opened by click-and-drag onto the window or icon of ImageJ while still saving disk space thanks to efficient compression. Figure 4 shows how the mosaic production options can be set inside ImageJ through the NDPITools plugins.		When producing a mosaic, the user can request that pieces be JPEG files. Since the <tt>File > Open</tt> command of versions 1.x of ImageJ is unable to open TIFF files with JPEG compression (one has to use plugins), this is way to produce mosaics which can be opened by click-and-drag onto the window or icon of ImageJ while still saving disk space thanks to efficient compression. Figure 4 shows how the mosaic production options can be set inside ImageJ through the NDPITools plugins.
Line 96:		Line 117:
	[[File:Fig4 Deroulers DiagnosticPath2013 8.jpg\|600px]]		[[File:Fig4 Deroulers DiagnosticPath2013 8.jpg\|600px]]
	{{clear}}		{{clear}}
	<blockquote>'''Figure 4. Dialog box for customized extraction in ImageJ from an NDPI file with production of a mosaic'''.~~<br />~~The dialog box shows some options which can be customized while producing a mosaic from a rectangular~~<br />~~selection of a NDPI file preview image (here, using the file previewed in Figure 3).</blockquote>		{\|
			\| STYLE="vertical-align:top;"\|
			{\| border="0" cellpadding="5" cellspacing="0" width="600px"
			\|-
			\| style="background-color:white; padding-left:10px; padding-right:10px;"\| <blockquote>'''Figure 4. Dialog box for customized extraction in ImageJ from an NDPI file with production of a mosaic'''. The dialog box shows some options which can be customized while producing a mosaic from a rectangular selection of a NDPI file preview image (here, using the file previewed in Figure 3).</blockquote>
			\|-
			\|}
			\|}

	==Results and discussion==		==Results and discussion==
Line 189:		Line 217:
	[[File:Fig5 Deroulers DiagnosticPath2013 8.jpg\|600px]]		[[File:Fig5 Deroulers DiagnosticPath2013 8.jpg\|600px]]
	{{clear}}		{{clear}}
	<blockquote>'''Figure 5.''' The positions of the 154240 identified nuclei were obtained from the analysis with ImageJ of the digital slide~~<br />~~on a laptop computer. Since the slide was too large to fit into the computer’s memory, it was turned into a mosaic of 16 pieces~~<br />~~with overlap of 60 pixels, and each piece underwent automated analysis independently. Then the results were aggregated. The graph~~<br />~~shows the probability density function of the distance of a cell nucleus to its nearest neighbor in the whole sample.</blockquote>		{\|
			\| STYLE="vertical-align:top;"\|
			{\| border="0" cellpadding="5" cellspacing="0" width="600px"
			\|-
			\| style="background-color:white; padding-left:10px; padding-right:10px;"\| <blockquote>'''Figure 5.''' The positions of the 154240 identified nuclei were obtained from the analysis with ImageJ of the digital slide on a laptop computer. Since the slide was too large to fit into the computer’s memory, it was turned into a mosaic of 16 pieces with overlap of 60 pixels, and each piece underwent automated analysis independently. Then the results were aggregated. The graph shows the probability density function of the distance of a cell nucleus to its nearest neighbor in the whole sample.</blockquote>
			\|-
			\|}
			\|}

	==Conclusions==		==Conclusions==

Shawndouglas: /* Notes */ Added cat.

2015-12-28T20:51:13Z

Notes: Added cat.

← Older revision		Revision as of 20:51, 28 December 2015
Line 284:		Line 284:

	<!--Place all category tags here-->		<!--Place all category tags here-->
			[[Category:LIMSwiki journal articles (added in 2015)‎]]
	[[Category:LIMSwiki journal articles (all)‎]]		[[Category:LIMSwiki journal articles (all)‎]]
	[[Category:LIMSwiki journal articles on health informatics‎‎]]		[[Category:LIMSwiki journal articles on health informatics‎‎]]

Shawndouglas: Italicizing journal title.

2015-10-05T17:33:41Z

Italicizing journal title.

← Older revision		Revision as of 17:33, 5 October 2015
Line 5:		Line 5:
	\|caption =		\|caption =
	\|title_full = Analyzing huge pathology images with open source software		\|title_full = Analyzing huge pathology images with open source software
	\|journal = Diagnostic Pathology		\|journal = ''Diagnostic Pathology''
	\|authors = Deroulers, Christophe; Ameisen, David; Badoual, Mathilde; Gerin, Chloé; Granier, Alexandre; Lartaud, Marc		\|authors = Deroulers, Christophe; Ameisen, David; Badoual, Mathilde; Gerin, Chloé; Granier, Alexandre; Lartaud, Marc
	\|affiliations = Université Paris Diderot, Université Paris-Sud, Montpellier RIO Imaging, CIRAD		\|affiliations = Université Paris Diderot, Université Paris-Sud, Montpellier RIO Imaging, CIRAD

Shawndouglas: Added rest of content.

2015-09-29T17:38:01Z

Added rest of content.

Show changes

Shawndouglas: Added content. Saving and adding more.

2015-09-26T23:09:43Z