Scribd
Upload
69 views

D3

This document provides an outline and summary of key topics from a medical imaging signals and systems discussion, including: 1) Providing feedback on homework 2 assignments and common mistakes. 2) Reviewing Fourier transforms and how to calculate them using properties like shifting and scaling. Worked examples of Fourier transform calculations are shown. 3) Discussing sampling and the Nyquist limit to prevent aliasing when taking discrete samples of a continuous function. 4) Explaining deconvolution and its goal of amplifying high frequency information attenuated by an imaging system to achieve better resolution, with the tradeoff of potentially amplifying noise.

Uploaded by

kinglbf
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
69 views

D3

This document provides an outline and summary of key topics from a medical imaging signals and systems discussion, including: 1) Providing feedback on homework 2 assignments and common mistakes. 2) Reviewing Fourier transforms and how to calculate them using properties like shifting and scaling. Worked examples of Fourier transform calculations are shown. 3) Discussing sampling and the Nyquist limit to prevent aliasing when taking discrete samples of a continuous function. 4) Explaining deconvolution and its goal of amplifying high frequency information attenuated by an imaging system to achieve better resolution, with the tradeoff of potentially amplifying noise.

Uploaded by

kinglbf
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 12

Medical Imaging Signals and Systems

Discussion 3

September 24/25, 2019


BioE C165/EE C145B/BioE/EE C261

1
Outline
• HW 2 Feedback
• Review and Worked Problems:
– Fourier Transforms
– Sampling Review
– Deconvolution Demo

2
HW 2 Feedback
• Mean of ~55/70, std ~ 17
• Common Mistakes:
– Not handing in homework/attempting problems
– Properties of Deltas (there are two!)
– Impulse object size (FWHM/10) vs resolution (FWHM)
– Point Spread Function/Impulse Response (h(x)) versus Transfer Function
(H(k))
– Resolution versus detection
• Read over solutions, come to office hours/Discussion otherwise

3
Fourier Transforms and how we calculate them
• Fourier Transforms:
∞
–𝐹 𝑘 = ‫׬‬−∞ 𝑓 𝑥 𝑒 −𝑖 2𝜋𝑘𝑥 𝑑𝑥
• 99% of the time we don’t compute this integral!
– Instead, we use known Fourier pairs and properties.
• Today:
– Go through one GSI-led example and two student examples

4
Fourier Transforms
𝑥−𝑥0
• Ex: Compute the Fourier transform of 𝑓 𝑥 = ⊓
𝑎
• Solution: Use the Shifting and Scaling properties:
– Given ℱ{𝑓(𝑥)} = 𝐹(𝑘)
• ℱ{𝑓(𝑥 − 𝑥0 )} = 𝐹(𝑘) 𝑒 −𝑖2𝜋𝑘𝑥0
𝑥
• ℱ{𝑓 } = 𝑎 𝐹(𝑎𝑘)
𝑎
– So we combine these together, knowing ℱ{⊓ (𝑥)} = 𝑠𝑖𝑛𝑐(𝑘)
– Your final answer should be 𝐹 𝑘 = 𝑎 𝑠𝑖𝑛𝑐(𝑎𝑘) 𝑒 −𝑖2𝜋𝑘𝑥0

5
Fourier Transform Practice
• Compute the Fourier Transform of these problems:
– 𝑓 𝑥 = 𝑔𝑎𝑢𝑠𝑠 𝑥 ∗⊓ (𝑥)
• Hint: What is the Fourier transform of convolutions?
– 𝑓 𝑥 = cos(2𝜋𝑥) + sin 2𝜋 𝑥

7
Solutions
• ℱ 𝑓 𝑥 = ℱ 𝑔𝑎𝑢𝑠𝑠 𝑥 ∗⊓ 𝑥 = ℱ 𝑔𝑎𝑢𝑠𝑠 𝑥 ℱ ⊓ 𝑥
= 𝑔𝑎𝑢𝑠𝑠 𝑘 𝑠𝑖𝑛𝑐 𝑘
– Key property – Fourier transform of convolutions is the multiplication
between the two
• ℱ 𝑓 𝑥 = ℱ cos 2𝜋𝑥 + sin 2𝜋 𝑥
= ℱ cos 2𝜋𝑥 + ℱ sin 2𝜋 𝑥
= 0.5 𝛿 𝑘 − 1 + 𝛿 𝑘 + 1 − 0.50𝑖 𝛿 𝑘 − 1 − 𝛿 𝑘 + 1
1−𝑖 1+𝑖
= 𝛿 𝑘−1 + 𝛿(𝑘 + 1)
2 2
• Key property: Fourier Transforms are linear!

8
Sampling – How do we model it?
• Sampling is the process of taking points from a continuous function,
typically equally spaced, as a discrete representation of the original
function.
– What is the Nyquist limit and why do we need to calculate and obey it?
1
• Δ𝑥 ≤ – this prevents aliasing
2𝑘𝑚𝑎𝑥

– How is the limit determined mathematically?


• Calculate the minimum Δ𝑥 = 1/𝑘𝑠 such that the replicates in k-space don’t
alias. See the next page for visualization.

9
F(k) FS (𝑘)
𝑘𝑠

−𝑘𝑚 𝑘𝑚 k
𝑘 −2𝑘𝑠 −𝑘𝑠 𝑘𝑠 2𝑘𝑠 3𝑘𝑠 𝑘
(𝑘𝑠 − 𝑘𝑚 )
S(𝑘)
FS (𝑘)
𝑘𝑠 𝑘𝑠

−2𝑘𝑠 −𝑘𝑠 𝑘𝑠 2𝑘𝑠 3𝑘𝑠 𝑘 (𝑘𝑠 − 𝑘𝑚 ) 𝑘𝑠 2𝑘𝑠 𝑘

Oppenheim and Schafer: Discrete-Time


Signal Processing (2010)

10
Examples of Aliasing

11
Deconvolution
• Deconvolution is the reversal of the convolution performed by a given
(imaging system), typically with some sort of high pass/boost filter
– What is the goal of deconvolution?
• To amplify attenuated high frequency information for better/ideal resolution
– What trade-offs are there?
• Typically you amplify noise i.e. SNR loss for higher resolution

12
Swap to Matlab Here

13

You might also like