Vous êtes sur la page 1sur 13

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/258262937

A framework for image processing analysis


and visualization of materials microstructures
using ImageJ package
ARTICLE JANUARY 2012

CITATION

READS

453

5 AUTHORS, INCLUDING:
Asad Ullah

Guoquan Liu

Karakoram International University

University of Science and Technology Beijing

29 PUBLICATIONS 92 CITATIONS

111 PUBLICATIONS 603 CITATIONS

SEE PROFILE

SEE PROFILE

Dil Faraz Khan

Matiullah Khan

University of Science & Technology Bannu

Kohat University of Science and Technology

31 PUBLICATIONS 87 CITATIONS

44 PUBLICATIONS 173 CITATIONS

SEE PROFILE

SEE PROFILE

Available from: Asad Ullah


Retrieved on: 23 March 2016

2012 17 4
CHINESE JOURNAL OF STEREOLOGY AND IMAGE ANALYSIS Vol 17 No 4 Dec 2012

1007 1482 2012 04 0301 0312

301

A framework for image processing


analysis and visualization of materials
microstructures using ImageJ package
Asad Ullah12 Guoquan Liu13 Hao Wang1 Dil Faraz Khan1 Matiullah Khan1
1. School of Materials Science and EngineeringUniversity of Science and Technology Beijing
Beijing 100083China

2. Department of MathematicsKarakoram International University Gilgit-Baltistan

Gilgit 15100Pakistan

3. State Key Laboratory for Advanced Metals and MaterialsUniversity of

Science and Technology BeijingBeijing 100083China


Abstract Digital image processingsegmentation and analysis of microstructural images are crucial to obtain three dimensional 3D information about the features present in microstructure such as particles or
grains There are several commercial as well as public domain packages availible for processing and analysis of images ImageJis one of them whose wide adoptionlong existence and extensible plugin style
has made it a tool of choice for scientists from a broad range of disciplines It contains almost all of the
basic and latest functionalities required to processsegmentreconstruct and visualize materials microstructural images along with analysis tools for instance Particle Analyzer3D object counter3D
Roi Managerand so on for sophisticated statistical processing of groups of particles Although it is very
popular and is considered to be an useful and efficient open source image processing and analysis software
in biomedical research fieldsa little is known about it in materials science sector In this article we introduce ImageJbriefly to materials science community particularly those professionals who have no experience of image processing and analysis and propose a general frame work to processsegmentanalyze
and visualize microstructural serial section images in 3D by using it
Key words material stereology microstructure serial section images imageJ image segmentation image analysis
CLC number TG113. 1 TP391

Document code A

2012-11-20
50901008
50871017
51071019
Asad Ullah 1983 Gilgit-BaltistanPakistan
E-mail g liu@ustb edu cn

302

2012

17

ImageJ
Asad Ullah12 13

1
1
1
Dil Faraz Khan Matiullah Khan

1. 100083 2. Department of Mathematics


Karakoram International University Gilgit-BaltistanGilgit 15100Pakistan
3. 100083

ImageJ

Particle Analyzer3D object counter3D Roi Manager


ImageJ

ImageJ
transparent and everyone can know the full details of al-

Introduction

gorithmswhich is very useful to decide which algorithm is better suited according to the needs of user

The contribution of technological advances in the

ImageJan open-source Java-based image processing

field of materials science and engineering is well recog-

and analysis software created by Wayne Rasband at the

nized Three dimensional microstructural image pro-

US National Institutes of Health NIH 8 9 is a cost-

cessing and analysis is a key step towards materials

effective alternative for processing and analysis of mi-

characterization in which many researchers and materi-

crostructural images ImageJ is unique among all the

als scientists have used technological advances latest

open source softwares because it is not only in the pub-

computer vision and other image processing approa-

lic domainbut also runs on any operating system It

ches to good advantage There are several open source

has a particularly outstanding application programming

for exampleVaa3D
3

er

BioImageXD

CellProfiler
5

Reconstruct

ImageSurf-

interface API The reason for its attractiveness is be-

cause of its ease of useand it can perform a full set

IMOD

and

others and commercial for exampleAmiraVolocity

of imaging manipulations
10

According to a current

the ImageJ website11 has 7000 visitors a

and Imaris software packages for the processing and a-

study

nalysis of imagesa complete review of imaging soft-

day and a huge and knowledgeable community compri-

ware tools can be found in 7 Most of the commer-

sing of 1900 members are subscribed to the ImageJ

cial image processing and analysis softwares are often

mailing list as of May 2012 According to a PubMed

expensive and one of their important drawbacks is hid-

search conducted in May 2012there have been found

ing the details of the implemented algorithms Con-

over 20 000 papers which used imageJ10 A screen

versely in open-source softwares these details are

shot of the main window of imageJ is shown in fig 1

2012

17

Asad Ullah et al A framework for image processinganalysis and


visualization of materials microstructures using ImageJ package

Fig 1

303

Screen shot of imageJ's main window

ImageJ can read most of the common and impor-

of our knowledgethere are few studies in materials

tant image formats including BMPTIFFJPEGGIF

science which used imageJ for processing and analysis

DICOMFITSand PNGRAW and many other for-

of microstructural imagesfor instance 12 16but

mats It can displaysaveeditprintprocess and

most of them only used a specific plugin for a specific

analyze 8-bit grayscale and color16-bit integer and

task and used other commercial softwares for the rest of

32-bit floating point images It supports all types of

processing This seems to be the result of lack of

common image manipulations including whole images

awareness about the functionalities and plugins of im-

volumesspecific image regions and individual pixels

ageJ and also lack of a strong collaboration between bi-

The concept of volume in ImageJ is actually the stack of

ologists and materials science researchers ultimately

a sequence of images that share a single window It can

there exists a communication gap Thereforewe felt it

perform many image processingsegmentation and a-

is necessary to put an effort towards reducing this com-

nalysis operationsfor instancelogical and arithmeti-

munication gap to some extent by proposing a general

cal operations between imagescontrast manipulation

framework to processsegmentanalyzereconstruct

convolution Fourier analysis distance transforma-

and visualize materials microstructural images in three

tionssharpeningsmoothingedge detectionmorpho-

dimensions by using imageJ As a case studythis arti-

logical operationsmedian filteringwatershed segmen-

cle used optical microscopy serial section images of pure

tationparticle analysis and many other advanced oper-

iron

ations It performs image processing and analysis tasks


using filtersthese filters are written as plugins Due to

1. The framework

its open source natureit is extensible via its Java plugins and recordable macros user community can devel-

In order to convert the experimental serial section

op the plugins according to their needs and add them to

images into three dimensional volume datasets for fur-

ImageJ to extend its functionality According to a re-

ther characterization and visualization purposesseveral

cent study10ImageJ has more than 500 plugins to

processing steps are required These processing steps

perform different operations on images A full review of

include importing images into imageJ setting the

the evolution of ImageJ and its role in image processing

scaleregistering aligning selecting the region of

and analysis is well beyond the scope of this article

interest ROI and croppingsegmenting particles or

but one can refer to 8 10for an excellent overview

grains in imagesmaking the image stack isotropican-

about its all aspects

alyzing grains or particles and finally visualizing the

Because of its popularity in biomedical and biological research fieldsImageJ has emerged as increasingly used image processing and analysis software in diverse applications ranging from astronomysoil sciences

particles to study the morphology and other shape factors The details of each of these steps are given below
1 1

Importing images into imageJ and setting the


measurement scale

and climate sciencesto crystallography and materials


science As for as materials science research sector is

ImageJ can read many image formats as described

concernedstill a little is known about it To the best

in section 1 To import the images into imageJthe im-

304

2012

17

ages have to be placed into a folder in a proper

line the length of the scale bar This line length is giv-

sequence according to the acquisition order These im-

en in pixels Then Analyze set scale command is used

ages can be accessed by browsing through File import

to fill the information about the distance the value of

command from imageJ menu Thusthe sequence of

scale bar in Known distancefield and desired unit

images is imported into imageJ in form of a scrollable

in Unit of length fieldthe Distance in pixels

stack displayed in a single window In this studywe

field is filled automatically due to the drawing of line on

imported 100 image sections which formed our stack

scale bar In our casewe have the known distance

as shown in fig 2

value of scale bar as 100 m as shown in fig 2

To set the measurement scalethe scale bar on the

and Distance in pixelsas 126this sets the in-section

original microscope images can be usedby selecting

resolution of our images which is 0 974 m ImageJ

the Straight linetool from menu bar and drawing a

uses this scale for all the measurements on images

a Showing the position of scroll bar at 25 th image section b A different orientation view depicting the stack hierarchy of imageJ

Fig. 2

Stack of imported images

translating the second image with respect to the first


1 2

Image registration

which reduced some major rotational and translational

The rotational and translational shifts produced in

shifts Howeverthis process cannot completely align

images during experiment due to the manual placing of

the stack because of the lack of sub-pixel accuracy

sample in microscope stage or some other reasons can

The sub-pixel accuracy can be achieved by using

be corrected in imageJ by manual and automatic

Stackregplugin The algorithm used in this plugin

means There are many imageJ plugins available for

provides the recursive alignment of stacks where each

image registrationwe used two of them in this study

image is used as the template with respect to which the

and Stackreg The

next image is alignedbased on a coarse-to-fine itera-

Align3_TPplugin can be located from imageJ menu

tive strategy known aspyramid approach For exam-

bar by selecting Plugins Align Stacks Align3 _ TP

plefirst image is used as the template for second im-

and Stackreg can be located by selecting Plugins

age and after alignment with first one this resulting im-

Stacks Stackreg Align3_TPplugin allows the user

age will be used as the template for third image and so

to register two images or two image stacks by suc-

on Thusthe alignment process is finished only when

cessively aligning two dimensional 2D sections We

the images are perfectly aligned on each other Com-

used it for an initial alignment of images by overlaying

plete details of the algorithm can be found in 18

two consecutive images and manually rotating and

When the plugin is launchedthe current slice acts as

17

namely Align3 _ TP

18

2012

17

Asad Ullah et al A framework for image processinganalysis and


visualization of materials microstructures using ImageJ package

305

a global anchor In this studywe used the first image

such as dilationerosion and Euclidean distance map

of our stack as global anchor and thus got the entire im-

which may improve the thresholded imagesbut a man-

age stack completely registered with sub-pixel accuracy

ual segmentation is necessary to completely clean the

The stack of registered images is shown in fig 3 d

remaining artifacts
Herewe have cropped an area of 300 270 m

1 3

Image cleanup and segmentation


Image segmentation is an important step towards

analysis of grains Its main objective is to separate the


objects of interest from background and from each other
as well Several automatic and manual image segmentation methods are available in the literaturesuch as
thresholdingregion growingwatershed segmentation
edge detection and so on The selection of a specific
method depends on the imaging condition under which
the image is acquired Howevera suitable segmentation method must be applied for each type of image acquired An overview of image segmentation methods
can be found in 19 20
Before starting the segmentation processthe region of interest ROI has to be selected in the image
stack and croppedthis can be done by selecting the
rectangular toolbutton from below the menu bar to

note that imageJ can also handle massive datasets but


here we selected a small ROI from our dataset for demonstration purposes from our image stack and applied
a brightness threshold to convert the stack into a binary
volumetric data set in a way that each pixel belonging to
the interior of grain is set to whitewhile each belonging to the boundaries is set to black Howeverafter
thresholdingthe images were not cleaned properly and
there remained some spots and other artifacts To remove the spotswe have applied some morphological
operations e g morphological openingin which an erosion operation is followed by a dilatation that removed
some small spots in the background These commands
can be reached via Process Binary Dilate and
Process Binary Erode from the menu Other remaining artifacts are corrected manually It is also
usual that in case of space-filling grainssome grain
boundaries left unfilled within the image sections after

outline the desired area and cropped by using Image

thresholding fig 3 a and in case of particles dis-

crop command from menu bar After cropping the ROI

persed in matrix phasesome particles may overlap

the boundaries of objects grains / particles have to be

This is actually the result of low contrast between parti-

identified in each 2D image This is accomplished by

cle-particle contacts There are several options in im-

applying a threshold using Image Adjust Thresh-

ageJ to address these issuesbut for such type of prob-

old commandprovided that the objects have homoge-

lems a 3D watershed segmentation algorithm is always

neous pixel values and there is a significant difference

considered to be a best choice It does not only correct

between the pixel values of objects and background

these boundary problems but also labels each 3D parti-

Otherwisethe image background should be subtracted

cle with a unique identifierwhich is very useful for

by using Process Subtract Background command be-

analysis of particles In some recent studies it was ap-

fore thresholding There are different options available

plied to overcome similar problems21 22

in imageJ for thresholding 2D images or 3D stacks vol-

ImageJ have some plugins to implement 3D water-

umetric datasets including automatic by selecting

shed segmentation on image stacksbut unfortunately

different algorithmse g HuangIsodataLiOtsu

its results are not satisfactory Insteadsome alternative

MaxEntroppyYen and so on and manual by selec-

methods can be applied to fill these gaps in grain

ting an interval of values manually thresholding It is

boundary interfaces One way is to apply a binary wa-

often not the case that the objects are thresholded in a

tershed algorithm note that this algorithm works only

satisfactory way Mathematical morphology provides

on 2D binary images on the stack of images by selec-

some sophisticated tools for morphological operations

ting Process Binary Watershed command from

306

2012

17

menuit automatically fills the boundary gaps or sepa-

ming Fig 3 d shows the final stack of thresholded

rates the overlapping particles in each 2D imageand

and cleaned imagesin which the images are perfectly

this method works best for smooth convex objects that

aligned on each other with sub-pixel accuracy

do not overlap too much The second method is to cre-

The next step in the segmentation process is labe-

ate an Euclidean Distance Map EDM of each 2D im-

ling of objects for quantitative analysis There exist some

age in the stack and then find the local maxima by se-

commands and plugins in imageJ for labeling of parti-

lecting Process Binary Distance Map and then

cles The Analyze Analyze Particles command can be

Process Find Maxima Herewe used the second

used to label particles in 2Dit can be applied to 3D

method to fill the grain boundary gaps exist in each 2D

stacks but it labels particles independently in each 2D

image by selecting a suitable noise tolerance Actually

image section so that the same particle gets different

theFind Maximacommand finds the local maxima in

identifiers in different sections This problem can be

an image and creates one segmented particle per maxi-

avoided by applying a 3D labeling method In this in-

mum upon selecting Segmented Particlesfrom drop

stancethe 3D object counterplugin23 is the best

down menu of Output typeas shown in fig 3 b

choice for 3D labeling of particles It first applies a

c To apply it on whole image stack theFindStack-

threshold to image stack and labels each particle in 3D

Maximamacro was used Thusit automatically filled

and also computes geometrical measures Once installed

the missing gaps in each image of our stack Although

in plugins folderthis plugin can be reached via Plugins

these plugins operate in only 2D on a section by section

3D Object counter The output of3D object counter

basisbut is far better than manually tracing the grain

applied to our images stack is shown in fig 4

boundaries on each 2D imagewhich is time consu-

a Showing the gaps exist in grain boundaries after thresholding b Showing the euclidean distance map of image
followed byfind maximaoperation The + marks inside the image depict the local maxima of each grain c
Shows the result offind maximaoperationwhich filled the gaps in grain boundaries d Shows segmented
and perfectly aligned with sub-pixel accuracy image stack
Fig 3 A step by step process of image segmentation and cleanup

2012

17

Fig 4

Asad Ullah et al A framework for image processinganalysis and


visualization of materials microstructures using ImageJ package

307

The output of3D Object Counterplugin showing 3D segmented and labeled grains Labels are
shown as unique colors as well as integers see the integers inside small rounded rectangles on the
upper left portion of the screen shot The Log window shows the count of grains and the results
window shows geometrical measurements

mand twice The other way is using the plugins Make

Image calibration and scaling

Isotropicdeveloped by Julian Cooper or TransformJ


developed by Erik Meijering24 In this studyTrans-

Before starting measurements the image must be

formJplugin is used to scale the stack anisotropically

calibrated in 3Dwhich can be done by entering the

and selected the cubic convolution interpolation method

pixel widthheight and depth using image Properties

to get the isotropic voxel dimensions Once installed

command It is often the case that the Z spacing is gen-

this plugin can be reached via Plugins TransformJ

erally much larger than the X Y spacing and hence

TransformJ Scale from the menu bar

the voxels are taller along Z-direction as compared to X


and Ythis is due to the acquisition systems These

Quantitative analysis of 3D grains

anisotropic voxel dimensions ultimately form an anisotropic image stack Although these anisotropic voxel di-

The overarching aim of image analysis is the quan-

mensions do not affect the measurement processhow-

titative measurement of 3D grains or particles in image

everisotropic voxel dimensions are usually preferred

stackpreferably in an automatic way Once the grains

for 3D analysis and especially for effective high-resolu-

have been segmented and labeledand images have

tion reconstruction and visualization

been calibrated in 3Dthe quantitative analysis can be

ImageJ have several options to rescale the image

performed on each individual grain ImageJ have sever-

stacks and make them isotropic One way is to reslice

al plugins to perform geometrical and intensity based

the data by using Image Stacks Reslice/ com-

measurements on 3D grains for example the basic 3D

308

23
analysis plugins include 3D object counter
and
25

2012

17

It can also compute the 3D distance measurements from

3D Roi Manager In this studywe firstly used

one particle to others for examplethe distance from

3D object counter plugin to label each 3D grain

centre to centreborder to borderborder to centre

this plugin also computes geometrical measures as

eccentricityradial distance 3D angles between three

shown in fig 4 and then used 3D Roi Managerto

particles and 3D colocalization and contact surface A

get the geometrical and intensity measurements

screen shot of these measurements on our image stack is

As the particle size and surface area are key char-

shown in fig 5 According to the measurementsthere

acteristics which control many properties of materials

are a total of 273 grains in our image stack However

for example flowabilityagglomerationdusting tenden-

for an unbiased quantitative analysis it is better to re-

cy and other mechanical properties Hall-Petch effect

move the grains which touching the border of the sam-

and so on The 3D Roi Managercan compute these

plethis can be done in 3D object counter Thus

characteristics and many other geometrical measures in-

after removing the outside grainsa total of 91 unbiased

cluding ferret diametersphericity or compacity ratio

grains left in our sample with the largest grain having

between volume and surface and ellipsoid fitting the

volume 229 201 41 m3 ferret diameter 120 875 m

minimummaximum and mean distance from the center

and the smallest grain having volume 3404 36 m3 and

of the particle to its boundaryetc The intensity meas-

ferret diameter 25 445 m Both of these largest and

ures include minmaxmeanintegrated densityetc

smallest grains are shown in fig 7

Fig 5

The upper right window is the3D Roi Managershowing a list which contains ID's of grains which are added
via segmented image of3D Object Counterby usingAdd Imagebutton can be seen clearly The upper
left window is the segmented image sack which is added for measurements The windows3D Measure3D
Quantifyand3D Distanceshow the geometricalintensity and distance measurement results respectively

2012

17

Asad Ullah et al A framework for image processinganalysis and


visualization of materials microstructures using ImageJ package

309

Besides these basic pluginsthere available many

umesurface areaEuler characteristiclocal thick-

other plugins which can be used for quantitative analy-

nessmoments of inertiaFerret diameternumber of

sis of grains or particles For instancethe Particle

topological holes handles in the particlenumber of

26

Analyserin BoneJ

plugin can be used to meas-

enclosed cavities in the particle and coordinates of cen-

ure particles in a 3D binary image stack This plugin

troid of particles It calculates the surface area by first

was particularly developed for bone image analysises-

creating a surface mesh of each particle by using the

pecially to study osteocyte star shaped cells lacu-

well known marching cubes algorithm27 and then

naein bonesbut it can be used to study any porous

adding the areas of triangles making up the mesh The

media Like3D object counterand3D Roi Manag-

volume enclosed by surface mesh and the best fit ellip-

erit also labels unique particles and analyze each

soids to the mesh of each particle can also be computed

particle separately It can compute many important

by using it Fig 6 shows the ellipsoids fitted to each

quantities for each individual particle including vol-

grain's surface mesh

Fig 6

Results ofParticle Analyser The upper left window shows the surface renderings of each grain during
quantification processwhile the ellipsoids fitted to each grain are also visible clearly The upper right
window shows the segmented 3D grains in the image stack The Results window shows the measurement
results Log window shows warnings about grains which could not be rendered due to small size

28
Another useful plugin is3D Convex Hull
it
calculates the 3D shape descriptors for example 3D so-

propriate analysis methods or parameters ImageJ is

lidity and 3D Convexity based on a convex hull con-

considered to be a popular tool to analyze and visualize

structed from the image stack Other numerous plugins

massive datasets of microscopic images It provides

are freely available online which can cater the needs of

practical solutions for memory management and displa-

materials science researchers

ying large datasets without unrealistic hardware require-

ualize it in 3Dwhich can be helpful to determine ap-

ments29

3D visualization of microstructures
To better understand the data it is necessary to vis-

Volumes can be visualized by either volume rendering or surface rendering Volume rendering can be

310

2012

17

defined as the rendering of all the volume data by spec-

rithm32 which is a useful algorithm to preserve the to-

ifying color and opacity of each voxel and surface ren-

pology of particles or grains during rendering

dering is the rendering of the iso-surface of the region

In this articlewe used ImageJ 3D Viewerplu-

of interest ROI from volume data There available

gin for both volume and surface renderings As before

two important and powerful imageJ plugins Volume

visualizationthe brightness and contrast of the image

Viewerand ImageJ 3D Viewerfor 3D visualization

stack must be adjusted for an improved visualization

and manipulation of massive datasets The Volume

We adjusted the brightness and contrast by using the

Viewerplugin can be used for more complex volume

command Process Enhance Contrastin which there

visualizations in which one can slice in any direction

are several options but the Normalizeoption is best

inside the volume The hardware accelerated ImageJ

among allwhich allows all the image sections in the

30

3D Viewerplugin

can be used to visualize the data

stack to have almost the same brightness and contrast

as volumeorthogonal slices and surfaces at the same

The 3D volume and surface renderings of our image

time There exist several ways to implement volume

stack are shown in fig 7 By using ImageJ 3D Vie-

renderingssuch as ray-casting31 and texture map-

werone can also render each individual grain from

ping ImageJ 3D Vieweruses 2D texture mapping

image stack for visualization and the rendered dataset

method for volume renderingthat s why it is user

can be animated and saved as movies We rendered

friendly and does not need dedicated workstations

here the largest and the smallest grains in our dataset

rather it can be run on common laptops For surface

as shown in fig 7

renderingit uses a variant of marching cubes algo-

a Surface rendered microstructure formed from 100 sectionscomposing a volume of 300 270 170 m3 containing
273 grains b Surface rendered microstructure containing 91 grainsborder touching grains are removed for unbiased
quantitative analysis c The largest and smallest grains in the system d Volume rendered microstructure

Fig 7

Different views of surface rendered and volume rendered microstructures

2012

17

Asad Ullah et al A framework for image processinganalysis and


visualization of materials microstructures using ImageJ package

311

modern software engineering practices It is easy to use

Discussion and conclusions

due to its organization of plugins One can use any of


imageJ or Fiji for full image manipulation and visualiza-

Although there are available hundreds of imageJ

tion

plugins for a full set of image manipulation and visual-

To sum upwe introduced imageJ briefly to mate-

izationwe tried here to report the best from each cate-

rials science community and outlined a general frame-

gory that can do the best job according to the situation

work to processsegmentanalyzereconstruct and

and image conditionsand materials microstructural im-

visualize materials microstructural images in three di-

ages remained our main foci in this study Specially

mensions by using it We hope that this framework will

our main target in this study was thosematerials science

be helpful to materials science and engineering re-

and engineering professionals who have no experience

searchersespecially those professionals who do not

in image processing and analysis As image processing

have any experience in image processing and analysis

and analysis tasks can be performed by using a variety


of computer languages for example lower level compiled

Acknowledgements

languages such as FORTRAN and C + + and scripting languages such as MATLAB TM and IDL TM Al-

The authors gratefully acknowledge the financial

though the scripting languages are tend to be easier

support from National Natural Science Foundation of

than compiled languages but for a novice person it is a

China 5107101950901008 and 50871017

very difficult and time consuming task to first learn programming and then apply it to solve problems in the

References

relevant field Insteadmost of the researchers prefer


to look for cost-effective and easy to use softwares

1 Peng HRuan ZLong Fet al V3D enables real-time

which can cater their needs in an optimal way In this

3D visualization and quantitative analysis of large-scale

instanceimageJdue to its ease of use and open

biological image data sets J Nature Biotechnology


2010
28 348 353

source natureis a best choice for materials science


and engineering professionals It allows to do a full set

2 Carpenter A EJones T RLamprecht M Ret al CellProfiler image analysis software for identifying and quan-

of imaging manipulations which can hardly be found in

tifying cell phenotypes J Genome Biology2006 7

an open-source software As it is discussed above that


imageJ has hundreds of plugins to perform image pro-

10 R100
3 Feng DMarshburn DJen Det al Stepping into the

cessinganalysis and visualization tasks so to choose

third dimension J Journal of Neuroscience200727

best plugin for the right job is very important and at the

47 12757 12760

same time very difficult also In this article we mainly

Kankaanpaa PPaavolainen LTiitta Set al BioIm-

focused on this issue and tried to resolve it by selecting

ageXD an opengeneral-purpose and high-throughput

best plugins from each category to do the right job and

image-processing platform J Nature Methods20129

practically demonstrated the importance of each plugin

7 683 689

by applying it to the optical microscopic serial sections


of real microstructures
Finallyit is worth introducing here a variant of
imageJ namely FijiFiji33 is an open-source distribution of imageJ in which the plugins are organized category-wise in a well mannered way It enhances imageJ
by updating the infrastructure of the programusing

5 Fiala J C Reconstruct a free editor for serial section microscopy J Journal of Microscopy2005218 pt 1
52 61
6 Kremer J RMastronarde D NMcIntosh J R Computer
visualization of three-dimensional image data using IMOD
J Journal of Structural Biology1996116 1
71 76
7 Eliceiri K WBerthold M RGoldberg I Get al Biologi-

312

cal imaging software toolsJ Nature Methods20129


697 710
8

Collins T ImageJ for microscopy J Bioloty Techniques200743 1 S25 S30

9 Abramoff MMagalhaes PSunanda R Image processing

2012

17

mensional analysis of particle coarsening in high volume


fraction solid-liquid mixtures J

Acta Materialia

200654 2027 2039


22 Rowenhorst D JLewis A CSpanos G Three-dimensional analysis of grain topology and interface curvature in a

with ImageJ J Biophotonics International200411

-titanium alloy J Acta Materialia201058 16

36 42

5511 5519

10 Schneider C ARasband W SEliceiri K W NIH Image

23 Bolte SCordeliRes F P A guided tour into subcellular

to ImageJ 25 years of image analysis J Nature Meth-

colocalization analysis in light microscopy J Journal of

ods20129 671 675

Microscopy
2006224 3 213 232

11 ImageJ EB / OL 2012 11 20 http / / rsb info


nih gov / ij / index html In
12 Baldissera M RRios P RHein L R Oet al Three-dimensional characterization of pores in Ti-6Al-4V alloy
J Materials Research
201114 102 106
13 Sintay S D Statistical microstructure generation and 3D

24 Meijering E HNiessen W JViergever M A Quantitative evaluation of convolution-based methods for medical


image interpolation J Medical Image Analysis2001
5 2 111 126
25 Iannuccelli EMompart FGellin Jet al NEMO a tool
for analyzing gene and chromosome territory distributions

microstructure geometry extraction D Carnegie Mellon

from 3D-FISH experiments J Bioinformatics2010

University2010

26 5 696 697

14 Vanderesse NMaire Chabod Aet al Microtomo-

26 Doube MKosowski M MArganda-Carreras Iet al

graphic study and finite element analysis of the porosity

BoneJ free and extensible bone image analysis in ImageJ

harmfulness in a cast aluminium alloy J International

J Bone2010
47 6 1076 1079

Journal of Fatigue201133 1514 1525


15 Wang M YWilliams J JJiang Let al Three dimensional 3D microstructural characterization and quantita-

27 William EHarvey E C Marching cubes A high resolution 3d surface construction algorithm J Computer
Graphics198721 4 163 169

tive analysis of solidified microstructures in magnesium-

28 Sheets K GZhou YErtel M Ket al Neuroprotectin D1

based alloys J MetallographyMicrostructureand A-

attenuates laser-induced choroidal neovascularization in

nalysis20121 1 7 13
16 Harimkar S PDahotre N B Characterization of microstructure in laser surface modified alumina ceramic J
Materials Characterization
200859 6 700 707

mouse J Molecular Vision201016 320 329


29 Walter TShattuck D WBaldock Ret al Visualization
of image data from cells to organisms J Nature Methods20107 479 479

17 Parker J Stack Alignment Align3 _ TP EB / OL

30 Schmid BSchindelin JCardona Aet al A high-level

2012 11 15 http / / www med harvard edu /

3D visualization API for Java and ImageJ J BMC

jpnm / ij / plugins / Align3TP html

Bioinformatics201011 274

18 Thevenaz PRuttimann U EUnser M A pyramid ap-

31 Sabella P A rendering algorithm for visualizing 3D scalar

proach to subpixel registration based on intensity J

fields J Computer Graphics198822 51 58

IEEE Transactions on Image Processing19987 1

32 Lewiner TLopes HVieira A Wet al Efficient imple-

27 41
19 Russ J C The image processing hand book boca raton

mentation of marching cubes cases with topological guarantees J Journal of Graphics Tools20038 1 15

M CRC Press Taylor and Francis GroupLLC2011

33 Schindelin JArganda-Carreras IFrise Eet al Fiji an

20 Soille P Morphological image analysis principles and

open-source platform for biological-image analysis J

applications M New York Springer-Verlag2004


21 Rowenhorst D JKuang J PThornton Ket al Three-di-

Nature Methods
20129 676 682

Vous aimerez peut-être aussi