Transcript Automated Counting of Microbial Colonies on Rehydratable Film

Automated Counting of
Microbial Colonies on
Rehydratable Film Media
Mehrdad Saadat
Biology 4400
Spring Semester 1999
Automated Counting of
Microbial Colonies on
Rehydratable Film Media
I. Background
II. Methods and Results
III. Conclusions
Background
• Use of Petri dishes
– Every day, biologists use petri dishes filled with
media to culture bacteria
– One use for culturing is enumeration
– A simple formula can be applied to a set of cells
grown on media to enumerate the number
of cells in the original sample being tested
Background
• BioCount 
– Developed by
Apogee
Systems
Background
• BioCount 
– This is a scanner based system used for
detecting and enumerating bacterial
colonies in petri dishes
– The system uses raster imaging to detect
curvatures on a given media
– The program then accepts a set of
curvatures as a colony
Background
• Previous Results
– The BioCount system has proven to be
reliable in counting colonies on mFC agar
– Paula Tennant-Clegg showed in her
research that the BioCount system could
accurately count coliform colonies on
mFC agar when properly cultured
Background
• Rehydratable Film Media
– Recently the 3M company has developed
the Petrifilm®
– This is a much easier and faster way of
plating for several reasons:
 no need to spend hours making a selective
media
 no accidents in “cutting into” the agar while
streak plating
Background
• Can the BioCount system be adapted
for use with Petrifilms?
• If so, it provides a possibility for
– More rapid analysis
– Greater objectivity: Elimination of
variability associated with manual counts
– Automated documentation
Methods and Results
• Objectives of this Project
– Determine the Biocount® parameters for
optimal counts of total aerobic Petrifilms
– Compare manual and automated colony
counts
– Identify possible causes of false positive
and false negative errors in the automated
counts
Methods and Results
• Procedure
– First: Prepare a fresh culture of aerobic
bacteria
Bacillus subtilis, Enterococcus fecalis,
Staphylococcus aureus
– Once a turbid broth was obtained, it could
be diluted down to give a desired
concentration of bacterial cells (10-14)
Methods and Results
• Procedure (cont.)
– One ml of the dilution was then plated on
the Petrifilm aseptically using a pipette
– Proper procedure was followed to cover
the film and incubate it at the temperature
given by 3M™ (approximately 33º C)
– After 24 hours the films were removed
from the incubator
– At this time colonies were manually
counted
Methods and Results
• Procedure (cont.)
– The films were then scanned by the
desktop scanner, four films at a time
– Once plated and counted by hand, the
colonies were scanned using a typical
desktop scanner by HP at 150 dpi
Methods and Results
• Hypothesis:
There is not a significant difference in the
count of bacterial colonies on a Petrifilm
between a manual count and that of one
done by an automated system
Methods and Results
• Parameters
– Using the parameters tool, I attempted to
optimize the parameters so that the
program would detect an almost exact
number of colonies on a given film — if
this were to work I would then test the
same parameters on another film.
Methods and Results
• Parameters (cont.)
– Optimal parameters should pass two
tests:
• They must only allow for detection of few, if
any false positives nor any false negatives.
• They must accurately count the present
colonies
Methods and Results
• Parameters (cont.)
– Minimum Peak Size: 6
– Radius Step: 1
– Vector Radius Maximum: 3
– Gray Threshhold: 135
– Radius Power: 1.01
– RGB color settings: 240, 169, 25
– Intensity settings: 2.15, 1.25, .08
Methods and Results
• We already knew that the program was set
to utilize a gray scale in order to detect
colonies
• We also know that if we pointed to a given
colony with the computer’s mouse, and
used the right button of the mouse to click
on the colony, we would set the gray-scale
parameters to that colony’s colors
• An area of the film
may be enlarged to
give more distinct
colonies
• When enlarged the
colonies were not all
the same color and
there was a gradient of
colors within a given
colony.
• This appeared to be a
major source of error
in optimizing the
system
Methods and Results
Film #
---------
Manual
Count
-------------
Automated
Count
---------------
Difference
--------------
1
745
707
38
2
286
323
17
3
525
509
16
4
498
475
23
5
442
504
62
6
387
433
46
7
240
332
92
8
413
500
87
Methods and Results
• Statistical Analysis: t-test
t = 1.582
– At a 95% confidence level, the deviation
between the automated and manual count is not
significant
– Therefore, the data support the hypothesis
Conclusions
• Overall, there is good agreement between manual
and automated colony counts on total aerobic
Petrifilms
• If we were to choose a dark colony as our basis
for a gray-scale, the program would detect
anything below this level of darkness
• If we were to choose a light colony the program
would detect, light colonies, as well as dark
colonies, but it would also detect many false
positives and background interference.
Conclusions
• Areas for future work
– Microbiology research: Find ways to promote
colony growth with greater uniformity of color
– Computer research: Enhance the colony
detection method to handle color gradients
within colonies