BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Limited Angle Tomography:  Inpanting in Phase Space by Deep Learni
 ng - Tatiana Bubba\, University of Helsinki
DTSTART:20200221T160000Z
DTEND:20200221T170000Z
UID:TALK140008@talks.cam.ac.uk
CONTACT:AI Aviles-Rivero
DESCRIPTION:Limited angle geometry is still a rather challenging modality 
 in computed tomography (CT). Compared to the standard filtered back-projec
 tion (FBP)\, regularization-based methods\, combined  with  iterative  sch
 emes\,  help  in  removing  artifacts  but  still  cannot  deliver  satisf
 actory reconstructions.   Based  on  the  result  that  limited  tomograph
 ic  datasets  reveal  parts  of  the wavefront (WF) set in a stable way an
 d artifacts from limited angle CT have some directional property\, we prop
 ose a method that combines\, in the phase space\, the information coming f
 rom the visible part of the WF set and ”inpaints” the invisible one by
  learning it with a convolutional neural network (CNN) architecture.  The 
 WF set information is accessed by using the directional features of shearl
 ets combined with a compressed sensing formulation\, which is well suited 
 to derive  visible  and  invisible  coefficients.   Compared  to  other  r
 ecently  proposed  deep  learning strategies for (limited data) CT\, our m
 ethod provides a superior performance\, an (heuristic) understanding of wh
 y the method works\, providing a more reliable approach especially for med
 ical applications.  This is a joint work with G. Kutyniok\, M. Lassas\, M.
  Marz\, W. Samek\, S. Siltanenand V. Srinivasan
LOCATION:MR 11\, Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR