Science - USA (2021-12-10)

axial resolution of ~10mm. Light is delivered
through a single mode fiber, and many de-
signs feature bespoke micro-optics at the
distal facet, giving a side view of the sam-
ple. Images are constructed pixel by pixel by
mechanically retracting the endoscope in a
spiral motion, which enables 3D tomogra-
phy of the tissue surrounding the endoscope.
Rather than imaging the cross section of ves-
sels and similar structures, our system is de-
signed to operate in a different regime, namely
to image more-distant surfaces in the far
field of the fiber output at near–video frame
rates. As such, our system has a greater depth
range, limited only by collection efficiency
( 18 ), and yet lower axial resolution, which is
constrained by the shortest coherence length
of the light it is possible to transmit through
MMFs without spatiotemporal compensation
(see Eq. 1). The use of a MMF that supports
many spatial modes also enables high-speed
DMD-based point scanning rather than
mechanical scanning, allowing for high–frame
rate operation while sacrificing flexibility of
the endoscope itself. Emerging techniques to
monitor the TM of flexible MMFs in real
time with access to only the proximal end
have potential to overcome this restriction
( 25 , 26 ).

There are also several methods to image in

three dimensions through scattering media

under development. Time-gated LIDAR is a

well-established technique to see through ob-

scurants by filtering out scattered light (by

photon arrival time) to recover a signal domi-

nated by ballistic photons that have traveled

straight through the occluding object ( 27 ).

More recently, non–line-of-sight 3D imaging

around corners has been demonstrated, rely-

ing on streak or single-photon avalanche diode

(SPAD) cameras to overcome signal mixing

caused by several diffuse reflections from

opaque walls ( 28 ).

Our approach is fundamentally different

from these methods. Our measurements are

formed purely from light that has forward

scattered many times through a MMF, with

no ballistic component remaining. We extract

image data from this nonballistic light by pre-

characterization of a TM that enables the

spatial control of the incident field to guide

it through the MMF to its target. Our work

also complements recent demonstrations of

3D imaging through thick randomly scat-

tering media, in which diffuse scattering effects

are inverted using ultrafast pulse detection

( 29 ) and optical sectioning through larger

form factor multicore fibers using in situ

distal holography ( 30 ). Our prototype TOF-

based 3D imaging system provides an exciting

new modality to MMF-based microendoscopy,

with many potential applications to remote

inspection and biomedical imaging in the life

sciences.

REFERENCESANDNOTES

1. Y. Choiet al.,Phys. Rev. Lett. 109 , 203901 (2012).


2. S. Ohayon, A. Caravaca-Aguirre, R. Piestun, J. J. DiCarlo,
   Biomed. Opt. Express 9 , 1492–1509 (2018).


3. S. Turtaevet al.,Light Sci. Appl. 7 , 92 (2018).


4. S. A. Vasquez-Lopezet al.,Light Sci. Appl. 7 , 110
   (2018).


5. J. W. Goodman,J. Opt. Soc. Am. 66 , 1145 (1976).


6. M. Plöschner, T. Tyc, T.Čižmár,Nat. Photonics 9 , 529– 535
   (2015).


7. M. C. Velsink, Z. Lyu, P. W. H. Pinkse, L. V. Amitonova,
   Opt. Express 29 , 6523–6531 (2021).


8. S. M. Popoffet al.,Phys. Rev. Lett. 104 , 100601 (2010).


9. T. Cižmár, K. Dholakia,Nat. Commun. 3 , 1027 (2012).


10. S. Popoff, G. Lerosey, M. Fink, A. C. Boccara, S. Gigan,
    Nat. Commun. 1 , 81 (2010).


11. I. N. Papadopoulos, S. Farahi, C. Moser, D. Psaltis,Opt. Express
    20 , 10583–10590 (2012).


12. R. Di Leonardo, S. Bianchi,Opt. Express 19 , 247– 254
    (2011).


13. S. Liet al.,Light Sci. Appl. 10 , 88 (2021).


14. I. M. Vellekoop, A. P. Mosk,Opt. Lett. 32 , 2309– 2311
    (2007).


15. A. P. Mosk, A. Lagendijk, G. Lerosey, M. Fink,Nat. Photonics 6 ,
    283 – 292 (2012).


16. I. Gusachenko, M. Chen, K. Dholakia,Opt. Express 25 ,
    13782 – 13798 (2017).


17. J. Trägårdhet al.,Opt. Express 27 , 30055– 30066
    (2019).


18. I. T. Leite, S. Turtaev, D. E. Boonzajer Flaes, T.Čižmár,
    APL Photonics 6 , 036112 (2021).


19. B. Schwarz,Nat. Photonics 4 , 429–430 (2010).


20. J. Carpenter, B. J. Eggleton, J. Schröder,Laser Photonics Rev.
    11 , 1600259 (2017).


21. W. Xiong, C. W. Hsu, H. Cao,Nat. Commun. 10 , 2973 (2019).


22. R. N. Mahalati, R. Y. Gu, J. M. Kahn,Opt. Express 21 ,
    1656 – 1668 (2013).


23. J. G. Fujimotoet al.,Nat. Med. 1 , 970–972 (1995).


24. J. Liet al.,Light Sci. Appl. 9 , 124 (2020).


25. S. Li, S. A. R. Horsley, T. Tyc, T.Čižmár, D. B. Phillips,
    Nat. Commun. 12 , 3751 (2021).


26. G. S. Gordonet al.,Phys. Rev. X 9 , 041050 (2019).


27. L. Wang, P. P. Ho, C. Liu, G. Zhang, R. R. Alfano,Science 253 ,
    769 – 771 (1991).


28. A. Veltenet al.,Nat. Commun. 3 , 745 (2012).

1398 10 DECEMBER 2021•VOL 374 ISSUE 6573 science.orgSCIENCE

0

0.5

1

1.5

Mean Depth Precision (cm)

0.05 0.1 0.15 0.2

Angular Position in FOV (rad.)

0

20

40

60

80

Reflectivity Signal to Noise Ratio

40cm

80cm

120cm

160cm

200cm

A

B

Fig. 3. System characterization.(A) Depth
precision given by the standard deviation of
repeated depth measurements and plotted as a
function of scene depth and radial position in the
field of view. Here, we used a movable flat white
screen as a target object. (B) Signal-to-noise ratio of
the associated reflectivities. FOV, field of view.

Fig. 4. 3D imaging with enhanced fidelity.A series of views of a 3D reconstruction of a mannequin head

located at a range of 30 cm. Here, depth is represented as the surface profile, and reflectivity is rendered

as the surface shading. The data for this reconstruction were acquired in 2 s.

RESEARCH | REPORTS