Author Topic: Motion correction  (Read 2844 times)

Mlippert

  • Newbie
  • *
  • Karma: +0/-0
    • View Profile
Motion correction
« on: September 19, 2016, 08:12:08 AM »
Hi,

we are currently struggeling with the motion correction of our data. Due to the low contrast, it's fairly difficult to get consistent results. Is anybody else experiencing similar problems or already found a solution?

Best,
Michael

Daniel Aharoni

  • Administrator
  • Full Member
  • *****
  • Karma: +15/-0
    • View Profile
Re: Motion correction
« Reply #1 on: September 22, 2016, 03:32:15 AM »
Hi Michael,
The motion correction algorithm used in the Miniscope analysis code shared on GitHub came from Dario Ringash, https://scanbox.wordpress.com/2014/03/20/recursive-image-alignment-and-statistics/. I suggest reading his blog post for a bit more information on how it works. Generally this approach handles low contrast pretty well but it does need a spatial landmark such as a blood vessel or cell that always stays on to function well. In our implementation of his approach we suggest using a small ROI roughly 1/6 the size of the entire frame and to place the ROI away from the edges of the frame.