Finite-difference time-domain (FDTD) or Yee's method (named after the Chinese American applied mathematician Kane S. Yee, born 1934) is a numerical analysis technique used for modeling computational electrodynamics (finding approximate solutions to the associated system of differential equations).
Taflove coined the FDTD acronym and published the first validated FDTD models of sinusoidal steady-state electromagnetic wave penetration into a three-dimensional metal cavity.
Has the FDTD method been applied to model the electromagnetic field of an MR scanner?
It depends on what you want to simulate exactly.
From what I could gather (not my specialty at all) the radio frequency used in MR is around 50 Mhz, so that's a wavelength of about 6 meters, which is much bigger than the system. Wavelength discretization shouldn't be an issue then.
But what are the smallest parts you want to compute?
Let's say your grid spatial step is 1 mm, then you can hold a cubic meter of field data into 48 GB of memory to give you an estimate.
Dear Loic,
Thank you for you reply.
How did you calculate that one cubic meter of field data requires 48GB of memory?
What estimate did you get?
Assessment of what?
This is assuming field components are stored as double precision floats, which take about 8 bytes.
So you have 1000 cells (1 mm dicretization) in each direction, and 6 field components per cell, which turns into
6*8*(1000)^3=48 GB of RAM.
Of course that's a low estimation, as it's only the field storage and not the extra stuff, so it can probably take up to 200 GB depending on the software you use and the analysis you require. This is just to give you a ballpark order of what to expect in terms of computer resources.
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning