Difference between revisions of "Locating objects in a fluorescent microscopic image"

From Course Wiki
Jump to: navigation, search
(→‎Precision and accuracy of centroids of stationary objects)
 
(3 intermediate revisions by one user not shown)
Line 1: Line 1:
 
{{Template:20.309}}
 
{{Template:20.309}}
 
[[Category:20.309 development]]
 
[[Category:20.309 development]]
 +
 +
[[Category:20.309 development]]
 +
 +
{Template:20.309}
  
 
==Overview==
 
==Overview==
 +
 
In this exercise, you will figure out where stuff is in an image.
 
In this exercise, you will figure out where stuff is in an image.
 +
 +
===Background===
 +
 +
Before you begin this lab exercise, be certain you understand these concepts:
 +
 +
* [[Fluorescent probes]]
 +
* [[Epifluorescence microscopy]]
 +
* [[Optical detectors and noise]]
 +
* [[Image processing]]
 +
* [[Image acquisition and preprocessing]]
 +
* [[Morphological processing and segmentation of fluorescent microsphere images]]
  
 
===Goals===
 
===Goals===
−
* Develop software to find locations of fluorescent microspheres in a digital image, &sigma;<sub>x,n</sub> and &sigma;<sub>y,n</sub>.
+
 
−
* Investigate the effect of optical noise sources on centroid accuracy and precision.
+
* Develop software to precisely locate fluorescent microspheres in a digital image
 +
* Understand the limit
 +
 
 +
==Notation==
 +
<center>
 +
{| class="wikitalble"
 +
|-
 +
|<i>X<sub>i</sub></i> , <i>Y<sub>i</sub></i>
 +
|Horizontal and vertical coordinates of the <i>i</i>th particle's centroid
 +
|-
 +
|<i>X<sub>i</sub> [n]</i> , <i>Y<sub>i</sub> [n]</i>
 +
|Centroid of particle <i>i</i> in frame number <i>n</i> of an image sequence of <i>N</i> frames
 +
|-
 +
|<i>&sigma;<sub>x,i</sub></i> , <i>&sigma;<sub>y,i</sub>
 +
|Standard deviation of particle <i>i</i> centroid in multiple images
 +
|}
 +
</center>
  
 
==Procedure==
 
==Procedure==
Line 39: Line 71:
 
==Report==
 
==Report==
 
# Compare the accuracy an precision of intensity-weighted centroids at different gain settings.
 
# Compare the accuracy an precision of intensity-weighted centroids at different gain settings.
−
#* What is the trend in the uncertainty?
+
#* How does the uncertainty depend on the exposure setting?
−
#* What is the likely explanation for the trend?
+
#* What is the likely explanation of the trend?
  
 
===Error sources===
 
===Error sources===

Latest revision as of 21:54, 4 January 2013

20.309: Biological Instrumentation and Measurement

ImageBar 774.jpg

{Template:20.309}

Overview

In this exercise, you will figure out where stuff is in an image.

Background

Before you begin this lab exercise, be certain you understand these concepts:

Goals

  • Develop software to precisely locate fluorescent microspheres in a digital image
  • Understand the limit

Notation

Xi , Yi Horizontal and vertical coordinates of the ith particle's centroid
Xi [n] , Yi [n] Centroid of particle i in frame number n of an image sequence of N frames
σx,i , σy,i Standard deviation of particle <i>i centroid in multiple images

Procedure

Image 1 micron fluorescent microspheres

  1. Use imaqtool to record an image of 1 micron fluorescent microspheres at 40X magnification.
    • Minimize quantization noise by selecting 16-bit monochrome mode.
    • Adjust the camera settings to achieve good contrast with minimal noise.
  2. Plot a histogram of the pixel values to verify optimal exposure
    • Use a log axis for the pixel count.
    • Take another image if the contrast is low or some pixels are over-saturated.
  3. Write a function in Matlab or another languageto compute the centroid of each microsphere
    • The function should return an N row by 2 column matrix, where N is the number microspheres in the image. The first column should contain the X coordinate and the second column should contain Y.

Relevant Matlab commands

imaqtool imhist semilogy im2bw graythresh regionprops

Centroid precision and accuracy

  1. Image fluorescent microspheres as before, but change the settings in imaqtool to capture 5 sequential frames instead of 1.
    • Ensure that the camera gain is set to 0.
  2. Run your centroid finding code on each image.
  3. Compute the standard deviation of each particle's centroid, σx and σy.
  4. Turn the gain up to 5; re-adjust the exposure setting, and acquire another set of 5 images
  5. Turn the gain up to 10; re-adjust the exposure setting, and acquire another set of 5 images
  6. Run your centroid finder on the gain=5 and gain=10 image sets and compute the standard deviation.

Report

  1. Compare the accuracy an precision of intensity-weighted centroids at different gain settings.
    • How does the uncertainty depend on the exposure setting?
    • What is the likely explanation of the trend?

Error sources

Discuss the effect of the following error sources on the centroid values.

  1. Shot noise
  2. Readout noise
  3. Dark current noise
  4. Ambient light
  5. Mechanical vibration

<-- Todo: image representations, histogram and how to expose image page imaqtool/camera settings page Determine gain value to use -->

Retrieved from "http://engineerbiology.org/w/index.php?title=Locating_objects_in_a_fluorescent_microscopic_image&oldid=6710"