Fundamentals of Light Microscopy II:

Fluorescence, Deconvolution, Confocal,

Multiphoton, Spectral microscopy

Integrated Microscopy Course

 Review
• Lecture 1: Microscopy Basics
– Light train
– Kohler illumination*
– Getting an image
• Diffraction and refraction
• Objectives
– Special techniques
• Dark field
• Phase Contrast
• Differential Interference Contrast
 Lecture Objectives

1. Understand the Principles of Florescence Microscopy

2. Understand how Deconvolution Improves Resolution

3. Understand How Laser Scanning Confocals work

4. Understand the Principles of Spectral Unmixing

 Light Spectrum

Visible Spectrum = 400nm to 750nm

Planck’s Law: E=hc/l where h is Planck’s constant, c is the
speed of light, and l is the wavelength of a photon

Shorter Wavelength Longer Wavelength

1. Higher energy Lower energy
2. More Harmful to cells Less harmful to cells
3. More easily scattered Less easily scattered
 Fluorescence

• First described in the 1800’s by Sir George Stokes

(fluorspar)
• Light emission that continues only during the
absorption of the excitation light
• Stokes Shift: the phenomenon in which the emitted
light has longer wavelength than the excitation light
 Jablonski Diagram
3
2
S2 Vibrational relaxation
1
0
Internal Excited Singlet
States
conversion
3

2
S1 Vibrational relaxation
1

Emission light
Heat
Quenching
3
2
S0 1 Ground state energy
levels
0
Excitation light
 Characteristics of Fluorescence

1. Because of the differing rotational, vibrational and

even electronic states a molecule can have, peaks
are BROAD.
2. Each spectrum is characteristic of a given molecule.

3. Before returning to its normal state, the electron remain at

the excited level for approximately 10 – 15 seconds.
4. Because the excited state is HIGHLY REACTIVE it can
“bleach” (react with other molecules).

5. Everything is a point of light. Light goes in all directions.

 Fluorescent Microscope
(Epi-fluorescence)
Emission
Filter
Emission Wavelength
Excitation Wavelength

Dichroic
Beam splitter

Excitation Filter Objective =

Condenser
3 Major Parts
1. Excitation Filter (Ex)
2. Dichroic Beam splitter
3. Emission Filter (Em)

Sample
 Filter Spectrum

FITC TxRed

FITC Filter Set Texas Red Filter Set

Ex Filter Dichroic Em Filter Ex Filter Dichroic Em Filter

 Problem: Fluorophore Spectral Overlap
• With broad fluorescent peaks and broad collection parameters,
fluorescence from one fluorophore can appear in a second channel

Alexa Dye Excitations with FITC Filter Alexa Dye Emissions with FITC Filter

You can mistake yellow or orange for green fluorophore!

How to correct for this? One way is to do image subtraction –
Label with 1 fluorophore and see how bad the bleed through is in second channel.
Get intensity ratio then subtract out contribution from other fluorophore
Example:
Corrected Intensity Green = Measured Intensity Green – X% Intensity Orange
Confocal allows better ways!
 Resolution of
Fluorescence Microscopy
Bright field Epi-Fluorescence
Max NA of a Oil Objective
Max NA of a Oil Objective
n (oil) = 1.515
n (oil) = 1.515
NA (oil) = (1.515) (1.00) = 1.5
NA (oil) = (1.515) (1.00) = 1.5

Max resolution for a oil objective under Bright
Field
d = 1.22 / (NA ob + NA cond)
d = 1.22 (546nm) / (1.5 + 1)
d = 266nm
Max resolution for a oil objective under
Epi-Florescence
 = 546nm d = 1.22 / (2 * NA obj)  = 546nm (green)
(green)
d = 1.22 (546nm) / (2* 1.5)
d = 222nm
* It is possible to detect fluorescence
from smaller structures but it will be
impossible to resolve them*
 Fluorescence Microscopy

Pros and Cons of Fluorescence Microscopy

Pros Cons
Allows for multiple labels Out of focus light blurs image.
Techniques: Immunofluorescence, Must have fluorescent reporter
FISH
Sensitive Fluorescent bleed through
Relatively inexpensive Difficult with thick sections
 Challenges
• Spectral Overlap: Broad peaks cross over into
one another. One fluorophore comes off in 2
channels
• Blur: Out of focus light decreases resolution
• Bleaching: Excited fluorophores react to
become non-fluorescent

• Phototoxicity: Light can harm cells

• Background/Autofluorescence: Cells have
fluorophores too. May look like the ones you want to
examine.
Deconvolution Microscopy

Computationally Increases Image

Resolution
 Deconvolution Microscopy and 3D
Reconstruction
Key points
X-Z
•All fluorescent molecules emit light, both in
focus and out of focus
•an image always blurred by the contribution of
X-Y
light from structures which are out of focus
•phenomenon can be relatively well defined by
the optical properties of the image formation in
the microscope imaging system
Definitions
1. PSF – (Point spread function) how
Y
a point of light reacts when it is in X
focus and out of focus, determined
experimentally or back calculated
Z voxel
from the collected images.
2. Voxel – (volume element) 3-
dimensional pixel
 What does Deconvolution Do?
Deconvolution looks at each voxel (3D pixel) and
determines its relationship with the voxels around it.

Then it either subtracts out what it calculates is blur from

nearby bright voxels (easy way) or it moves out of focus
light back to its voxel of origin (hard way).

Raw Data Deconvolved Data

 Methods for Deconvolution
1. Non-Restorative Algorithms – (unsharp mask, nearest
neighbors, no neighbors, multiple neighbors) Subtract a given
amount of light from nearby voxels to account for blur effect.
A. Relatively cheap, very simple to do, quick,
computationally undemanding. Doable with Metamorph.
B. Subtractive – Trade higher contrast for decreased
signal. UNUSABLE for quantitation.
2. Restorative Algorithms – (inverse filter, constrained iterative
deconvolution, maximum likelihood estimation, blind deconvolution
and many others)
A mathematical method for applying the point-spread function (PSF)
to remove and/or reassign the "out-of-focus" blur in an image
specimen. - NOT subtractive

Premises
a. Since the image has been acted on by the PSF, a set of
weighted sines and cosines will represent this. Use Fourier
Transformation to calculate these.
c. Divide fourier transformation of measured image by fourier
transformation of the PSF (the OTF) and you should get closer to
true image. Perfect if all values were right.
d. Reblur with original PSF and you should get back to observed
image.
 Types of Restorative Algorithm

Inverse Filter does this once and stops. Can intensify noise and
produce artifacts. Quicker.
Constrained iterative deconvolution keeps trying to improve
image by decreasing difference between reblurred and original.
Certain constraints applied to solution. Slower and computationally
intense.
Maximum likelihood estimation further applies statistical
calculations to noise and is still more computationally intense. May
be better for noisy data
Blind deconvolution can refer to either using an empirically
determined PSF or determining the exact PSF as you go.
Typical Constrained Iterative Deconvolution
Output

Standard Avg Counts

Iteration Residual (R) Residual Normalized R
--------- ------------ -------- ------------
1 0.02446074 6.2 1.000
2 0.02190482 5.6 0.896
3 0.02070552 5.3 0.846
4 0.01977012 5.0 0.808
5 0.01897621 4.8 0.776
6 0.02169737 5.5 0.887
7 0.02062611 5.2 0.843
8 0.01981340 5.0 0.810
9 0.01908517 4.8 0.780
10 0.01842614 4.7 0.753

The Normalized R is the

best way to see if the
image is still improving
 Differences Between Methods:
There is a reason to go through iterative solutions

CID

Original
Nearest
Neighbor
 Deconvolution Microscopy
Raw Deconvolved

Pros and Cons of Deconvolution Microscopy

Pros
Allows for higher Resolution
Allows collection of image plane stacks
Hg Bulb allows for variety of filter
combinations
Can get superior sensitivity
Cheaper than confocal
Cons
Computationally intense if iterative
Minimal penetration of thick sections
Sensitive to spherical aberrations
Can amplify noise, produce artifacts
Requires 3D data set
Confocal Microscopy

Physically Increases Image

Resolution
 Laser Confocal Scanning Microscopy
Laser 1980’s

Dichroic Photo-
Beam Multiplier
Splitter Tube
(PMT)

Detector

Objective Confocal Pin Hole

Out of Focus

Sample In Focus

Out Of Focus
 Laser Light Sources

Lasers

Bulb
 Laser Confocal Scanning Microscopy

• The laser beam excites a point on the specimen. It also

inadvertently excites other points on the specimen.
• Only the In-Focus emission light is allowed to be detected by the
PMT.
• The Light detected by the PMT is associated to a pixel (picture
element) on the monitor.
• The laser beam then moves to the the next point and another pixel
is collected.

Confocal Pinhole
Result on the Intensity of the Result on the Resolution of
Size Image the Image
Increase opening Increase Intensity decrease resolution

Decrease Decrease intensity increase resolution

opening
 Laser Confocal Scanning Microscopy

No Pin Hole 50um Pin Hole

Pros and Cons of Confocal Microscopy

Pros Cons

Allows for higher resolution Limited EX peaks on lasers

Allow collection of stacks of image planes and Phototoxicity
3D reconstruction (up to a 40º C temp jump at focal point)
Laser penetrates somewhat thick sections Loss of image intensity
Better control of bleed through/autofluorescence Fairly expensive
Faster than deconvolution Prone to Photobleaching
Precise Laser Positioning (FRAP)
Multiphoton Microscopy

Optically increases resolution

 Multiphoton Microscopy
•Based on phenomena
described by Maria Goppert-
Mayer (1931)
•Two photons of low energy
(700-1000 nM)
(long wavelength) can
excite a fluorophore
resulting in emission of a
photon of higher energy
(short wavelength)
•Requires near-
simultaneous absorption of
the two photons

(400-500 nM)
 Multi-Photon Confocal Microscopy
Characteristics
1. Attosecond pulses only properly strike at the In-Focus plane
of the specimen.
2. Multiple short pulses of longer wavelength light have the
same amount of energy as one long pulse of shorter
wavelength.
Pros and Cons of Multi-Photon Microscopy
Pros Cons

More penetrating – up to hundreds of Best excitation wavelength is not always

micron
No out of focus light
Laser is tunable. Can get range of
wavelengths including UV dyes.
Better for Living cells if you’re careful.
Can also kill off specific cells/organelles.
obvious
Cost (Very Expensive)
Laser is POWERFUL. Can damage cells,
slides and even walls.
Difficult to align and use (Less so now)
 Multi-Photon Images

Mouse Liver Tissue Mouse Cardiac Tissue

Grey = hepatocytes
Green = collagen Green = collagen
Red = myocytes
Spectral Unmixing

Computationally Resolves
Spectral Overlap
 Spectral Unmixing
Allows one to use and resolve multiple
fluorophores with highly overlapping emission
spectra

Specific Cases:

• You need to use fluorophores with highly

overlapping signal (GFP and YFP; Mitotracker
red and dsRed)
 Emission Fingerprinting - the Method
1.
• Instead of using an emission filter, spectral
*contains spectral information for each pixel
unmixing relies on obtaining intensity values
for each pixel at a number of wavelengths,
and then creating a spectrum. Next, the user
compares this spectrum to defined spectra
for different fluorophores. After mathematic
comparison, you can determine the true
value of each fluorophore present.

2.

3.
 Linear Unmixing – How it works!
Unmixing separates the total emission signal into weighted contributions of each dye based on
the knowledge of their “emission fingerprints” – pixel by pixel!

Combined spectra = a x GFP + b x YFP

A final problem – data management
(Or: “Where is my file?!!”)

Answer – save files with a fixed filename such

as: Initial for scope
1ss1022ad

Slide number Field of view

Your initials
The date
Second answer – Save in a fixed place on server
References:

1. Molecular Probes www.probes.com

2. Chroma Filters www.chroma.com

3. Omega Optical www.omegafilters.com

4. Zeiss Microscopes www.zeiss.de/us/micro/home.nsf

5. Applied Precision Deconvolution www.api.com

6. Fluorescent Protein Spectra www.bdbiosciences.com/spectra/

7. Free Image Processing Program ` www.gimp.org

8. Confocal List serve listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal

9. Microscopy List serve www.msa.microscopy.com/MicroscopyListserver

10. Olympus www.olympusmicro.com/primer/