Reducing depth induced spherical aberration in 3D ... · Reducing depth induced spherical aberration in 3D widefield fluorescence microscopy by wavefron t ... engineer the point spread
Post on 26-Aug-2018
222 Views
Preview:
Transcript
Reducing depth induced spherical aberration in 3D widefield fluorescence microscopy by wavefront coding using the SQUBIC phase
mask
Nurmohammed Patwary1, Ana Doblas2, Sharon V. King1, and Chrysanthe Preza1♦, 1Dept. of Electrical and Computer Engineering, The Univ. of Memphis, Memphis, TN, USA
2Dept. of Optics, Univ. of Valencia, E-46100, Burjassot, Spain
ABSTRACT
Imaging thick biological samples introduces spherical aberration (SA) due to refractive index (RI) mismatch between specimen and imaging lens immersion medium. SA increases with the increase of either depth or RI mismatch. Therefore, it is difficult to find a static compensator for SA1. Different wavefront coding methods2,3 have been studied to find an optimal way of static wavefront correction to reduce depth-induced SA. Inspired by a recent design of a radially symmetric squared cubic (SQUBIC) phase mask that was tested for scanning confocal microscopy1 we have modified the pupil using the SQUBIC mask to engineer the point spread function (PSF) of a wide field fluorescence microscope. In this study, simulated images of a thick test object were generated using a wavefront encoded engineered PSF (WFE-PSF) and were restored using space-invariant (SI) and depth-variant (DV) expectation maximization (EM) algorithms implemented in the COSMOS software4. Quantitative comparisons between restorations obtained with both the conventional and WFE PSFs are presented. Simulations show that, in the presence of SA, the use of the SIEM algorithm and a single SQUBIC encoded WFE-PSF can yield adequate image restoration. In addition, in the presence of a large amount of SA, it is possible to get adequate results using the DVEM with fewer DV-PSFs than would typically be required for processing images acquired with a clear circular aperture (CCA) PSF. This result implies that modification of a widefield system with the SQUBIC mask renders the system less sensitive to depth-induced SA and suitable for imaging samples at larger optical depths. Keywords: Three-dimensional microscopy, wavefront encoding, point-spread function engineering, image restoration, phase mask.
1. INTRODUCTION The wavefront encoding technique has been implemented successfully in point-spread function (PSF) engineering3 by placing a phase mask at the back focal plane of the microscope objective. The PSF is aberrated due to the refractive index (RI) mismatch between the immersion medium of the microscope objective lens and the object mounting medium. Spherical aberration (SA) increases with increased focal depth and due to this depth dependence, spherical aberration is a dynamic process. Spherical aberration reduces image resolution and increases the complexity of computational optical sectioning microscopy (COSM). Different depth variant restoration methods have been developed to address the depth variability of the PSF that add computational burden. To address this problem we test a wavefront modification designed to decrease the depth dependency of PSFs. Pupil modification by a phase mask has been studied in extended depth-of-field (EDOF) microscopy3 where a cubic phase mask (CPM) was designed to gain EDOF. In EDOF imaging an intermediate encoded image is taken using a modified pupil and a computational process decodes this image. EDOF with a generalized cubic phase mask (GCPM) has been successfully studied in 3D computational microscopy2. In addition, adaptive optics and depth-variant deconvolution algorithms6 have also been used to counteract spherical aberration7-9. A strata based model in the restoration process5 represents the volume using multiple PSFs computed at specific depths to mitigate the depth variability of the PSFs. Performance of this approach depends on the thickness of the object. The number of PSFs required increases with the increase of thickness and so does the computational burden and memory requirement. Principal component analysis (PCA) is another approach to solve the depth-variant imaging problem10, 11
♦cpreza@memphis.edu; phone 1 901 678-4369; fax 1 901 678-5469; cirl.memphis.edu
Three-Dimensional and Multidimensional Microscopy: Image Acquisition and Processing XXI, edited by Thomas G. Brown, Carol J. Cogswell, Tony Wilson, Proc. of SPIE Vol. 8949, 894911
© 2014 SPIE · CCC code: 1605-7422/14/$18 · doi: 10.1117/12.2040191
Proc. of SPIE Vol. 8949 894911-1
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
which minimizes the computational burden in restoration process but at the same time requires a lot of memory for an accurate PCA component calculation. These approaches work well in that they produce results in which artifacts due to SA are minimized, but they require a high performance computation facility. Hence, a simpler solution would be to reduce the PSF depth variability of the PSFs.
Recently, a simple but effective phase mask (SQUBIC) has been reported1 to reduce the SA impact over a large range of depth. This phase mask works effectively in reducing wavefront distortion, and hence makes the PSF invariable over a large range of depth. This depth invariability makes object restoration from the blurred image using a single PSF over a wide range feasible. In this paper, we show the performance of the SQUBIC phase mask in 3D fluorescence microscopy in restoring a 3D object. Both qualitative and quantitative comparison is presented to show the improvement gained by using SQUBIC phase mask in the back focal plane in lieu of the conventional CCA system. This paper is organized as follows. Section 2 describes the background and theory behind SQUBIC WFE-PSF. Section 3 and 4 include the simulation methods and results, respectively. Summary of conclusions and future possible improvements are discussed in Section 5.
2. BACKGROUND 2.1 SQUBIC encoded WFE-PSF
Depth induced spherical aberration comes from the wavefront distortion due to the propagation of light within layers with RI mismatch, which depends on light wavelength, object RI, and numerical aperture (NA) of the objective lens. All these factors introduce different degrees of freedom and make SA a dynamic process. The idea of the SQUBIC phase mask is to apply a fixed dominant wavefront distortion via a phase mask 2 ( , )Aφ π ϕ ρ α= 1 at the back focal plane of the objective lens. To address different imaging conditions and design parameters, variation of the value of A is considered in order to vary the amount of wavefront distortion applied. Efficiency of the SQUBIC phase mask increases with the increase of A. In practice, the microscope objective’s back focal plane is projected on a spatial light modulator (SLM) through a 4F optical system and the SLM resolution limits the value of A16. The phase function ( , )ϕ ρ α is defined as1 :
32 21 sin ( ) 1 1( , )
1 cos( ) 2ρ α
ϕ ρ αα
⎡ ⎤− −= +⎢ ⎥
−⎢ ⎥⎣ ⎦ (1)
where ρ is a normalized pupil radius, ( )1sin NAnα −= and n is the refractive index of the lens’ immersion medium.
2.2 3D DV WFE-PSF formation
In optical sectioning microscopy, a 3D volume is acquired by optically slicing the sample through the axis of light propagation (Z-axis) and capturing 2D images at multiple depths. When the specimen emitted wavefront passes through the objective lens, its Fourier transform is generated at the back focal plane12 . To calculate WFE-PSFs from CCA PSFs, a generalized pupil function is calculated by taking the Fourier transform of the CCA PSF slices and adding the SQUBIC phase function ( , )x yf fφ to the CCA pupil phase of each slice. This process is described by2 :
(2)
where, { }1F − • is 2-D inverse Fourier transform, ( , )x yH f f is clear circular aperture pupil function , λ is the emission
wavelength, and ( , ; , )x y i oW f f z z is the optical path length error due to defocus and SA as a function of the normalized
spatial frequencies xf and yf .
{ } 2(2 / ) ( , ; , ) ( , )1, ( , ) ( , ) x y i o x y
i o
j W f f z z j f fz z x yh x y F H f f e eπ λ φ−=
Proc. of SPIE Vol. 8949 894911-2
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
2.3 Forward model formation
The forward image is simulated by computing 3D DV-PSFs at each layer using the superposition integral2 :
( ) ( , , , ) ( )WFE i o i o i o oO
g h x x y y z z s d= − −∫∫∫i 0x x x (3)
where, ( , , )i i ix y z=ix is a point in the image space, ( , , )o o o ox y z=x is a point in the object space O and ( )s ox is the object of interest. 2.4 Image restoration
Intermediate images, simulated using the forward model, are restored using the space invariant expectation maximization (SIEM)13 and the depth variant expectation maximization (DVEM)5 algorithms. Both methods are iterative. In SIEM process, only one single PSF is used to restore the whole volume where in the DVEM method multiple DV-PSFs are used to restore the total volume. In the restoration process the region of interest (ROI) is divided in the axial direction into multiple strata and a PSF is computed at the edge of each stratum. Hence, the number of PSFs required in DVEM is equal to the number of strata plus one. The number of 3D deblurring computations needed is equal to the number of strata. As 3D deblurring is a computationally costly process, complexity increases with the increase in number of strata. A higher number of strata is required to achieve a more accurate restoration result. Both these algorithms are implemented in CosmEstimation module COSMOS4 .
3. SIMULATION METHODS 3.1 Depth-variant (DV) PSF
Two hundred CCA DV-PSFs were calculated in MATLAB using the Gibson-Lanni optical path distance (OPD) model14 and vectorial field approximation15 at an interval of 0.3 µm to cover 60 µm depth. PSFs were calculated on a 128×128×1024 grid with a voxel size 0.1 µm × 0.1 µm × 0.1 µm with the following parameters: (a) the light point source is located at varying depth (starting from 0 µm to 60 µm with 0.3 µm spacing) in water (RI, nwater = 1.33) below the coverslip (b) a 20X/0.8 NA objective lens and (c) the emission wavelength λemmision = 515 µm. The voxel size is chosen to satisfy the Nyquist criteria. To compute WFE-PSFs, CCA PSF data were oversampled and over-ranged i.e., the number of pixels in the x-y plane was increased by the oversampling factor (we note that if over-ranging and oversampling is done by the same factor, the frequency domain sampling remains the same). WFE-PSFs were calculated from the CCA PSFs using the methods described in section 2.2. We used A = 88 to compute the SQUBIC phase mask.
3.2 Test object and forward image formation
To demonstrate the performance of the engineered PSF with the SQUBIC phase mask, three different types of object were studied. Test object 1 (Fig. 3a) consists of three small spheres (3 μm in diameter). Coordinates of the bead centers are (16.8 µm, 12.5 µm, 20 µm), (16.8 µm, 16.8 µm, 30 µm), (16.8 µm, 21.1 µm, 40 µm) respectively. The RI of the mounting medium is assumed to be 1.33, while the immersion medium of the lens is air (i.e., RI = 1.00). A second synthetic object (Fig. 3e) has only one 3-µm in diameter bead and depth position of that bead was varied from 20 µm to 40 µm. Test object 3 has five spherical beads of 3-µm diameter (Fig. 2a). The beads are centered at different (x,y,z) coordinates: (16.8 µm, 8.3 µm, 10 µm), (16.8 µm, 12.5 µm, 20 µm), (16.8 µm, 16.8 µm, 30 µm), (16.8 µm, 21.1 µm, 40 µm) and (16.8 µm, 25.3 µm, 50 µm) respectively. All synthetic objects are simulated on a 336×336×600 grid where each voxel size is 0.1 µm × 0.1 µm × 0.1 µm. The object is set within a larger grid (336×336×1250) to allow enough empty space to completely capture the intensity of the blurred images. A 7×7×7 Gaussian kernel with standard deviation 2.5 is used to smooth the object. This test object is convolved with all 200 PSFs using the variant tab of the CosmTools4 to compute the simulated image, referred to here as the forward model image, (see Fig. 2b & c) and restored using CosmEstimation module of COSMOS software.
Proc. of SPIE Vol. 8949 894911-3
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
4.1 WFE-PS
Fig. 1 showsview of CCAthe CCA PSwith the PSFalong the marespectively qualitatively further quanterror (NMSE
where, 0 (h xthe axial shicomputation.CCA imaginsystem.
4.2 Forward
Simulated imspherical abeshows the ob
Fig. 1:SQUBIat 0 µmnormalLens us
SFs using SQU
s CCA and SQA PSF at 0 μm F location shif
F at increased daximum intensand Fig. 1f shseen that the
tify the WFE-PE) was calculat
, , )x y z is the Pft of the PSF . Fig. 1g shows
ng compared to
d image
mages of a test erration. Simubject with five
: PSF engineeIC-PSF at 0 µmm and 50 µm,lized mean sqused: 20x/0.8 N
4
UBIC phase m
QUBIC WFE Pand at 50 μm rfts along the adepth is asymmity point. Fig. hows the corrPSF only shifPSF’s variabiled:
NM
PSF at zero depis compensates the NMSE3D vo SQUBIC WF
object demonsulated forwardsimilar spheric
ering with SQUm, (d) SQUBI (f) axial profuare error betw
NA air waveleng
4. RESULT
mask:
PSFs at differerespectively. T
axial direction.metric and exten
1c and Fig. 1dresponding XZfts along the axity compared t
03
(D
hMSE =
pth and ( ,ih x yed by registerinvs depth curveFE imaging wh
strate that SQUd images werecal beads (obje
UBIC phase mIC-PSF at 50 µfile through thween PSF comgth 515 nm.
TS AND DI
ent depths beloTwo large diffe Second, the Cnds along z-axd show XZ vieZ profiles. Froxial direction to the PSF of
0
( , , ) (
( , , )ix y z h x
h x y z
−
, )y z is the PSFng the maximu
e. It is seen thathich infers the
UBIC WFE ime computed usect 3 as describ
mask. (a) CCAµm, (e) axial pe center of SQ
mputed at 0 µm
ISCUSSION
ow the coverslierences betweeCCA PSF at 0xis. Fig. 1e shews of SQUBICm the XZ viewithout changthe CCA syste
2
2
, , )
)
x y z
F at some otherum intensity pt, the NMSE3Dreduced depth
maging suffers ing the metho
bed in section 3
A-PSF at 0 µmprofile though
QUBIC-PSFs cm and other PS
N
ip. Fig. 1a andn these two PS
0 µm is symmeows the XZ prC WFE-PSFs
ew of the SQUging its shape em, the 3D no
r depth. To coplanes between
increases morh sensitivity of
less variabilityod described is3.2). Fig. 2b is
m, (b) CCA-the center of C
computed at 0SFs computed
d Fig. 1b showSFs are observeetric along z- arofile of the CCat 0 μm and atUBIC WFE-PSat increased d
ormalized mean
mpute the NMn the PSFs usee rapidly in the
f the modified
y due to depth-s section 3.2. the simulated
-PSF at 50 µmCCA-PSFs com µm and 50 µat different lo
w the XZ ed. First, axis, but CA PSFs t 50 μm, SFs it is epth. To n square
(4)
MSE3D, ed in the e case of imaging
-induced Fig. 2a forward
m, (c)mputedµm, (g)ocation.
Proc. of SPIE Vol. 8949 894911-4
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
image in casWFE imagininvariant to d
4.3 Image re
We studied r1 as describebeads at diffshows the qua large objecBoth SIEM ausing the WF
4.3.1 RestorSI EM restorsystem with forward simu(middle of Frestored imagthat, both thimaging systphase mask (intersecting suggests thatsignificant re To show theposition of a case using a and WFE SQPSFs) to quaNMSE vs. dapproximatel
Fig. 2: Simul60 µm deep simulated imof the object 20x/0.8 NA.
se of CCA imang. Although thdepth in contra
estoration from
restoration perfed in Section 3ferent depths. Suantitative comct (object 3) toand DVEM resFE system.
ation in a framration is done tSQUBIC pha
ulated image isFOV along z-axges for both Ce location andtem (Fig. 3b) w(Fig. 3c). To sheach bead as t, we can use thestoration perfo
e decreased de single bead (oPSF computed
QUBIC imaginantify the restodepth of bead ly 36% NMSE
lated forward Cobject. Empty
mage. (b) A simin (a). Images
aging system ahe image of eacast to the CCA
m intermediat
formance of SQ3.2. The purposSecond we con
mparison of depo show the perstoration result
me work of spto show a com
ase mask. Thrs calculated usxis) was used
CCA and SQUBd shape of the whereas all thrhow a quantitatshown in Fig.he WFE imagiormance.
epth sensitivityobject 2) alongd at 30 µm. Thng system was oration results. (or z-axis pos
E variation whe
CCA image any slices (ES) armulated CCA i are shown usi
and Fig. 2c is tch bead appearimage in which
te images
QUBIC phase mse of this simunsider another pth sensitivity brformance of Wts are shown in
pace invariant mparison study ree identical bing the methodin the SI restoBIC imaging smiddle bead
ee beads are retive compariso. 3b. The inteing model to m
y of WFE imag the z axis frohe NMSE betw
also computedFig. 3e shows
sition) curve. Tereas the conve
nd SQUBIC intre added to allimage of the oing different co
the correspondrs more distorth images of be
mask for both ulation is to shSIEM simulatbetween CCA
WFE imaging n this case for
(SIEM) modebetween the r
beads are consids described inoration processsystem. Fromis restored betestored almoston, we plotted aensity profile
minimize comp
aging in termsm 20 µm to 40
ween the correspd (based on Eqs one of theseThe WFE imaentional CCA
termediate imalow enough spobject in (a). (color scales to a
ding simulated ed in the SQUB
eads at deeper d
SI and DV EMow the feasibiion with only and SQUBIC system over w
r comparison in
el restoration perfidered at an in
n Section 3.2. Ts. Fig. 3a-c sho
m the restorationtter than the ot equally in casan intensity prois consistent w
putational effor
s of a quantita0 µm and restoponding objec
q. 4 but using tobjects from a
aging system wimaging gives
ages. (a) A synace for the ext
c) A simulatedallow visualiza
forward imagBIC image thadepth show mo
M algorithms Flity of using oone bead (objeimaging system
wide range of dn order to dem
formance of Cnterval of 10 µThe PSF calcuow the xz viewn results, it ca
other two beadse of the WFE ofile (Fig. 3d) with the qualitrt over a 20 µm
ative metric mored a simulatets and restoredtrue and estim
a set of 11 objewith the SQUBs almost 73% v
nthetic 33.6 µmtended intensit
d SQUBIC WFation of small i
es in case of San the CCA imore axial asymm
First we considenly one PSF toect 2). This simm. Lastly, we depth i.e. up to
monstrate the b
CA and WFE µm (object 1) lated at a 30-µ
w of the objectan be said qualds in case of t system with Salong the diagotative observam range withou
measure, we vaed WFE imaged images for bo
mated images inects. Fig. 3f shBIC phase mavariation over
m x 33.6 µm wty spread in th
FE intermediatentensity values
SQUBIC mage, it is
metry.
er object o restore mulation consider o 60µm. enefit of
imaging and the
µm depth t and the litatively the CCA SQUBIC onal line tion and ut losing
aried the e in each oth CCA nstead of hows the sk gives a 10-µm
wide andhe WFEe images. Lens:
Proc. of SPIE Vol. 8949 894911-5
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
depth range. when the beasystem suggeusing the comSQUBIC WF
4.3.2 DepthTo show therange of dept4a-i show theSIEM and Drestored imagfive strata Dsystem. FromPSF is the wrestored imagDVEM restoSIEM restora
For a better uThis time wparallel to thcenter of nex
Fig. 3: Peimage of t(b) Convdepth; (cprofile thr30-µm deThe minimthe locatiodetails in e
Though the Nad location exests that, SIEMmputationally FE system thes
variant restoe comparison sth, we simulate XZ views ofVEM restoratiges with the PS
DVEM restoratm the restoratioworst whereas fges of WFE im
ored images ofation.
understanding,e did not draw
he z axis. Eachxt bead and he
erformance evthe object withventional imag) similar restorrough the diagoepth below the mum NMSE3D on of the singeach image. A
NMSE for the Cxactly matches M restoration in
costly DVEMse artifacts are
ration study betweened the 60 µm d
f the object andion. Fig. 4b-e aSF computed ation model reson results it cafor the WFE s
maging system f CCA imaging
, we show a quw a single line time after pas
ence plotted the
valuation of Wh three beads loing restorationration as in (bonal line showcoverslip; anderror computele PSF. Restor20x/0.8 NA ai
CCA imaging swith the PSF
n the presenceM restoration al
minimized as e
n the restoratiodeep object wid the restored iare xz views ofat 0 µm and 30spectively. Figan be said thatsystem signific
using the PSFg system which
uantitative restoe though the cssing though the intensity vs.
WFE system usocated at 20 µmn using the SIE) for the mod
wn in figure (b)d (f) NMSE3Ded for the SIEMred images areir lens is simul
system is less tlocation, the r
e of SA leads tlgorithm for thevident by the
on performanceith five identicimages for difff the restored i0 µm respectivg. 4f-i are corrt, for CCA im
cant amount ofF computed at 3h involves five
oration performcenter of each he center of ondepth curve. F
ing the SQUBm, 30 µm and 4EM algorithm
dified imaging ); (e) XZ-sectiorestoration an
M restoration oe displayed at lated.
than the one obrapid variationto restoration ahe conventiona
slow variation
e of CCA andal beads as desferent kind of rmages for CCA
vely and Fig. 4dresponding res
maging system f information c30 µm can sufe times the co
mance by plottbead, rather w
ne bead we shiFig. 4j shows t
BIC PSF and S40µm depth, re
and a CCA Psystem using on image of an
nalysis of theoccurs at a 30-separated non
btained for then of the NMSEartifacts which al system. On tn of the NMSE
d WFE imaginscribed in sectrestoration techA system of wd-e are restoredstored images the SIEM rest
can be restoredfficiently compmputational bu
ting intensity pwe took the prfted along x axthe intensity v
SI computationespectively, belPSF that is comthe SQUBIC Pn object with single bead atµm depth whic
nlinear color sc
e WFE imagingE in the CCA
can only be mthe other handcurve (Fig. 3f
ng system overtion 3.2 (objecthniques includ
which b and c ard images with for the WFE
toration with thd with this PSFpete with the fiurden compare
profile along throfile through xis to pass thros. depth profil
n. (a) XZ-sectlow the coversmputed at 30 PSF; (d) intensa single bead at a varying depch correspondscales to show
g system imaging
mitigated d, for the f ).
r a wide t 3). Fig.
ding both re SIEM two and imaging he 0-µm F. SIEM ve strata ed to the
he z axis. the line
ough the e of the
tion lip. µmsity at a pth. s to the
Proc. of SPIE Vol. 8949 894911-6
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
SIEM restoramask can reaintensity profneeded for thphase mask wrecover all tburden for th
Fig. 4: Perfosection imagcomputed at (CCA); (e) DSQUBIC-SIEwith five stratwo-strata DVto show deta20x/0.8 NA a
In this paper,results show back focal pldepth and redepth-induceCCA system The image reencoding thrThrough the phase mask lithographica
ation with the Pasonably reducfile of two stra
he WFE imaginwhich is to makthe necessary he DVEM appr
ormance compae of the object 0 µm, (c) CC
DVEM with fivEM restorationata restoration VEM. Each imails and figure air lens is simu
, we have presethat, the effect
lane of the imastored these us
ed SA and bearequires a DV
estoration procrough the SQUproper selectiocan easily b
ally. With the
PSF computedce the depth seata DVEM restng system in thke the imaginginformation fr
roach.
arison of WFE(beads are loc
CA-SIEM restove strata restorn using PSF co
(SQUBIC); (jmage is shown t
(j) and (k) areulated.
ented performat of depth-indu
aging system. Wsing a single Pds can be reas
VEM restorationcess with a SQUBIC phase mon of the mask
be implementeincrease of de
d at 30 µm. Theensitivity of thetoration (Fig. 4he case simulatg system less serom the blurre
E systems basecated axially at oration using Pration (CCA);mputed at 30 µj) restoration pto its own scale plotted by co
5. Cance of a new Wuced spherical We simulated iPSF. Simulatiosonably restoren method using
QUBIC WFE-Pmask is therefk parameter A,ed in an existesign paramete
e intensity profe imaging syst4k) it can be inted here. This iensitive to SA ed microscope
ed on the SQU10, 20, 30, 40
PSF computed(f) SQUBIC-Sµm; (h) DVEMprofile for SIEMle to show theironcatenating li
CONCLUSIWFE imaging aberration canimaging of 3Dn results show
ed using a spacg a larger num
PSFS is therefofore a solutionits performanc
ting microscoper A, the wrapp
file shows thatem over a wid
nferred that, theis consistent wso that SIEM r
e images and
UBIC PSF and 0 and 50 µm); (d at 30 µm; (dSIEM restoratioM with two straM (with PSF ar details. Figurines through th
ION approach using
n be reduced by sparse beads d
w that the SQUce-invariant (S
mber of PSFs reore faster and n to decrease ce can be imprpe setup16 or ped phase rapi
t, WFE imaginde range (i.e. ue DVEM restorith the purposerestoration algthereby avoid
either SI or D(b) CCA-SIEM) DVEM withon using PSF cata restoration at 30 µm); (k)re (a)-(i) are drhe object to ge
g a SQUBIC py using a SQUBdistributed in a
UBIC WFE-PSFSIEM) restoratisulting in longrequires fewerdepth induced
roved. By usinalternatively,
idly changes b
ng with SQUBIup to 60 µm). Fration technique of using the Sorithm can sufing the compu
DV restoration. M restoration ush two strata rescomputed at 0 (SQUBIC); (i) restoration pr
rawn to their owet the depth pr
phase mask. SimBIC phase maa volume overFs are less senion process wh
ger computationr resources. Wd spherical abng a SLM the S
it can be fabetween π− t
IC phase From the ue, is not SQUBIC fficiently utational
(a) XZ-
sing PSF storation µm; (g)
) DVEM rofile for wn scale rofile. A
mulation sk in the a 60 µm
nsitive to hereas, a nal time.
Wavefront berration. SQUBIC abricated o π and
Proc. of SPIE Vol. 8949 894911-7
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
may become difficult to implement the mask practically and fine sampling will be needed in computation which will increase the computational burden2 . In this study, our concentration was confined to low numerical aperture objective lens. Further study is being performed using a high numerical aperture oil-immersed objective lens.
6. ACKNOWLEDGEMENT This work is supported by the National Science Foundation (NSF CAREER award DBI-0844682, PI: C. Preza; NSF IDBR award DBI-0852847, PI: C. Preza) and The University of Memphis (through a Herff fellowship to N. Patwary).
7. REFERENCES [1] G. Saavedra, I. Escobar, R. Martínez-Cuenca et al., “Reduction of spherical-aberration impact in microscopy by
wavefront coding,” Optics Express, 17(16), 13810-13818 (2009). [2] S. Yuan, and C. Preza, “Point-spread function engineering to reduce the impact of spherical aberration on 3D
computational fluorescence microscopy imaging,” Optics Express, 19(23), 23298-23314 (2011). [3] E. R. Dowski, and W. T. Cathey, “Extended depth of field through wave-front coding,” Applied Optics, 34(11),
1859-1866 (1995). [4] [Computational Imaging Research Laboratory, Computational Optical Sectioning Microscopy Open Source
(COSMOS) software package]. [5] C. Preza, and J.-A. Conchello, “Depth-variant maximum-likelihood restoration for three-dimensional
fluorescence microscopy,” Journal of the Optical Society of America A-Optics Image Science and Vision, 21, 1593-1601 (2004).
[6] L. S. M. M. Rahman, D. Schuster and C. Preza, "Comparison of computational methods developed to address depth-variant imaging in fluorescence microscopy", SPIE 8589, id. 85890A 7 pp., (2013).
[7] M. J. Booth, M. A. Neil, R. Juškaitis et al., “Adaptive aberration correction in a confocal microscope,” Proceedings of the National Academy of Sciences, 99(9), 5788-5792 (2002).
[8] M. J. Booth, M. Schwertner, T. Wilson et al., “Predictive aberration correction for multilayer optical data storage,” Applied physics letters, 88(3), 031109-031109-3 (2006).
[9] C. Mauclair, A. Mermillod-Blondin, N. Huot et al., “Ultrafast laser writing of homogeneous longitudinal waveguides in glasses using dynamic wavefront correction,” Optics express, 16(8), 5481-5492 (2008).
[10] M. Arigovindan, J. Shaevitz, J. McGowan et al., “A parallel product-convolution approach for representing the depth varying point spread functions in 3D widefield microscopy based on principal component analysis,” Optics express, 18(7), 6461-6476 (2010).
[11] S. Yuan, and C. Preza, "3D fluorescence microscopy imaging accounting for depth-varying point-spread functions predicted by a strata interpolation method and a principal component analysis method." 79040M-79040M-7.
[12] J. W. Goodman, [Introduction to Fourier Optics] McGraw-Hill Book Company, New York(1968). [13] J. A. Conchello, “Superresolution and convergence properties of the expectation-maximization algorithm for
maximum-likelihood deconvolution of incoherent images,” JOSA A, 15(10), 2609-2619 (1998). [14] S. F. Gibson, and F. Lanni, “Experimental test of an analytical model of aberration in an oil-immersion
objective lens used in three-dimensional light microscopy,” Journal of the Optical Society of America A-Optics Image Science and Vision, 9, 54–66 (1992).
[15] O. Haeberlé, “Focusing of light through a stratified medium: a practical approach for computing microscope point spread functions. Part I: Conventional microscopy,” Optics communications, 216(1), 55-63 (2003).
[16] S. V. King, A. Doblas, N. Patwary et al., "Implementation of PSF engineering in high-resolution 3D microscopy imaging with a LCOS (reflective) SLM", “Implementation of PSF Engineering in High-Resolution 3D Microscopy Imaging with a LCoS (Reflective) SLM,” SPIE 8949-39, (2014).
Proc. of SPIE Vol. 8949 894911-8
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/04/2014 Terms of Use: http://spiedl.org/terms
top related