Cytodeep

PROJECT MEMBERS
Bekir Açıkça
Oğuzhan Gür

2023
ISTANBUL
 ● TABLE OF CONTENTS
● TABLE OF CONTENTS ............................................................................................................. 1
● LIST OF TABLES ....................................................................................................................... 1
● LIST OF FIGURES ..................................................................................................................... 2
1 Executive Summary ..................................................................................................................... 3
2 Business Project Details ............................................................................................................... 4
3 Key Personnel .............................................................................................................................. 6
4 Business/Project Goals ................................................................................................................. 7
5 What makes the business different? ............................................................................................. 8
6 Market research ............................................................................................................................ 9
7 Profiling customers .................................................................................................................... 11
8 Algorithms, Flow Chart and Programming Details ................................................................... 11
9 Potential Project Risks and Backup Plans.................................................................................. 12
10 Mathematical Model and Foundations ....................................................................................... 13
11 Design and Technical Drawings ................................................................................................ 15
12 BOM Bill of Materials List ........................................................................................................ 21
13 Milestones, Work Packages and Intermediate Deliveries .......................................................... 23
14 Project Time Plan and Schedule ................................................................................................ 23
15 Literature Survey, Patent Search ................................................................................................ 24

● LIST OF TABLES
Tablo 1 Competitor Analysis ............................................................................................................... 8
Tablo 2 Risk and Backup Plan ........................................................................................................... 12
Tablo 3 Sample Counting .................................................................................................................. 14
Tablo 4 Bill of Materials .................................................................................................................... 22

1
 ● LIST OF FIGURES
Şekil 1 Cytodeep Colony Counter ....................................................................................................... 3
Şekil 2 CytoDeep Working Block Diagram ........................................................................................ 5
Şekil 3 Sample Mediums ..................................................................................................................... 5
Şekil 4 Market Research .................................................................................................................... 10
Şekil 5 Dual CNN MobileNet Model Architecture ........................................................................... 11
Şekil 6 Exploded Design of Cytodeep ............................................................................................... 15
Şekil 7 Circuit Design ........................................................................................................................ 16
Şekil 8 Technical Drawing of Cytodeep ........................................................................................... 16
Şekil 9 Technical Drawing of Cover................................................................................................. 17
Şekil 10 Technical Drawing of Lower Body ................................................................................... 18
Şekil 11 Technical Drawing of Upper Body ..................................................................................... 18
Şekil 12 Technical Drawing of Petri Part ......................................................................................... 19
Şekil 13 Technical Drawing of Middle Body ................................................................................... 20
Şekil 14 Trello Summarize ................................................................................................................ 23
Şekil 15 Time Plan and Schedule ...................................................................................................... 23

2
1 Executive Summary
A microbial colony is a visible cluster of microorganisms formed by the continuous proliferation of a single
microorganism cell on a solid medium or surface. Colonies can be counted by simple methods since they
can be seen with the naked eye. Counting is used extensively in the food industry, where competition is
very high. In the production sector, doing this manually with the naked eye is not very efficient. Therefore,
accurate and rapid counting of the number of microorganisms in foods is a problem. These problems are
even greater with counting methods, such as serial dilution. Cytodeep is a product created to facilitate
this counting process. Cytodeep uses image processing techniques in counting. In this way, it accelerates
the counting process and increases the accuracy rate. Raspberry Pi mini computer will be used to create
a compact model. Appropriate lighting and camera will be used to get a clear view of the Petri dish. These
systems will be placed in our uniquely designed device, which is printed from 3D printing.

Şekil 1 Cytodeep Colony Counter

3
2 Business Project Details
In microbiological analysis processes, the number of microorganisms in the culture medium and the
type of microorganism is important. The count results are of great importance in the evaluation of the
examined sample in terms of microbiological quality. Rapid microbiological food and water analysis
are very important in terms of detecting emerging bacterial infections and preventing these infections.
The problems that caused the prolongation of the analyzes were examined in the form of 5 items:
● Increasing Number of Samples
Increasing yield rate in the food and water industry increases the speed of microbiological
analysis and prolongs the analysis time. Increased analysis and the number of samples
examined, the limitations of bacterial colonies on the medium, and the manual colony
counting process performed in microbiological tasks are the causes of prolonged periods and
fatigue.
● Similarities of Bacterial Colonies on Medium
K. pneumoniae and E. coli colonies formed on MacConkey medium and Salmonella and
Shigella colonies formed on Hektoen Enteric medium can be given as examples of the
similarities of bacterial colonies. In MacConkey medium, E. coli colonies have a slimy
structure but not K. pneumoniae. In Hektoen Enteric medium, Salmonella colonies have a
dark and black structure, while Shigella colonies are lighter and transparent (Waltman, 2000).
It is very difficult and time consuming to notice these differences with the naked eye.
● Manual Count Process
Manuel sistemlerin insan kaynaklı hatalara açık olması ve zaman kaybı olumsuz yönleri
olarak öne çıkmaktadır (Goss vd., 1974). Mikrobiyolojik süreçlerde yapılan koloni manuel
sayım işlemi, kolonileri tek tek kalemle sayarak yapılmaktadır, bu da sürecin oldukça
uzamasına ve yorucu olmasına sebep olmaktadır ve hata payı oranını yükselmektedir. Ayrıca
yapılan çalışmalar sayma işleminin uzmandan uzmana farklılık gösterebilen,
standardizasyonun gerekli olduğu bir alan olduğunu göstermektedir (Biston vd., 2003).
● Repeated Counting
The colony forming unit (CFU) used to indicate the contamination of the sample indicates
bacterial colonies formed on the medium. In microbiological processes, the CFU should be
less than 300, if the CFU is more than 300, the sample is diluted and the colonies are counted
again (Boukouvalas et al., 2019). The serial dilution (Dilution) method is time consuming,
error prone and causes eye strain (Vongmanee et al., 2018). Manual counting, which takes 3-
5 minutes per sample on average, prolongs the analysis process as it is repeated and increases
the margin of error.
● Inadequacy of Existing Solutions
There are different commercial companies that produce solutions for colony counting. Current
solutions consist of manual colony counters with low accuracy and partially autonomous
colony counters that are expensive. However, existing solutions have become expensive due
to their import from abroad and the exchange rate effect. Existing solutions provide accurate
results for certain media types and certain conditions, as they use conventional image
processing techniques. Error margins are quite high in counting operations performed with
classical colony counting aids or completely manually.

The CytoDeep device to be developed takes the most ideal medium image and identifies and counts
the colonies on those mediums together with the medium type. The microbiological analysis process
needs to be carried out quickly and with an error margin of less than 6%, so it will be very useful to
use high-accuracy techniques such as deep learning in the analysis process. Using the dual-based
Faster R-CNN deep learning model will accelerate bacteria detection and colony counting processes.

4
Şekil 2 CytoDeep Working Block Diagram

Şekil 3 Sample Mediums

5
3 Key Personnel

Name : Bekir Açıkça

Skills: Solidworks, PTC Windchill, LabVIEW, C, C#
Experience: System Engineering at Femsan Electric Motors

Name : Oğuzhan Gür

Skills: Solidworks, Matlab, C, SAP, Python
Experience: Internship at IGA Istanbul Airport

6
4 Business/Project Goals
The CytoDeep project is a microbiological study using artificial intelligence and image processing
technique.
It aims to speed up high-accuracy and time-consuming processes such as bacterial identification and
colony counting in analysis.

Product/service Features Benefits

● Camera ● Visualization ● Clearer image
quality
● LED ● Lighting ● Clearer image
quality
● Rasberry Pi ● It transfers the ● Speeding up data
data from the usage
device to the
computer.
● Arduino ● It will provide ● Facilitates engine
servo motor control
control and
adjustment of
angles.
● Servo Motor ● It will be used for ● Blocking outside
opening and light during
closing the device viewing
cover.

7
5 What makes the business different?
There are different commercial companies that produce solutions for colony counting Tablo 1.
However, it has become expensive due to its import from abroad and the effect of the exchange rate.
In this project, it was tried to develop a domestic production solution for a need that can be solved
with high costs from the outside. Existing solutions provide accurate results under certain media types
and certain conditions, as they use conventional image processing techniques. We are developing a
new approach as deep learning techniques such as Double-Based Faster R-CNN will be used in our
project. It also has what we call multiple counting. Multiple counting is the separation of colonies of
different bacteria on the medium and counting them simultaneously in real time. Since the SMD
LEDs used work over 5 volts, they are suitable for energy saving. Autonomously adjusting the light
according to the medium type, our multi-count and double CNN-based model and energy saving are
among the features that make us innovative in this field.

Tablo 1 Competitor Analysis

8
6 Market research
The industrial microbiology market is large and growing due to the increasing preference for
automation. The increased emphasis on industrial food and safety is fueling market growth. Industrial
microbiology has applications in agriculture, cosmetics, food and beverage, and pharmaceuticals. As
the demand for consumer products increases with the increasing population around the world, the
need for product quality and safety testing is also increasing. Various public concerns and government
regulations regarding the safety of products are expected to continue to provide an opportunity for
market growth.

Industrial microbiology deals with the screening, remediation and management of microorganisms
for the production of various end-use products.

Testing for pathogens quickly and accurately can protect brand image, reduce industry costs, and
reduce epidemic research and can expedite the intervention. Increasing product recalls and growing
consumer concerns are driving the global industrial microbiology market. The 2020 size of the global
market is 4.36 Billion Dollars and the sector is expected to reach 9.46 Billion Dollars in 2026 with a
CAGR of 10.97%. On the basis of geography, the global industrial microorganisms market is
segmented into North America, Europe, Latin America, Asia Pacific and Rest of the World. North
America is expected to be an important regional segment in the industrial microbiology market.

9
Şekil 4 Market Research

10
7 Profiling customers
With this structure, the device to be developed in the project; It can be used in hospital laboratories,
university laboratories, food production and inspection institutions, private analysis laboratories
(about 150 laboratories registered in Turkish labs), control laboratories of the Ministry of Agriculture
and Forestry, public health laboratories and the pharmaceutical industry.

8 Algorithms, Flow Chart and Programming Details

Şekil 5 Dual CNN MobileNet Model Architecture

11
9 Potential Project Risks and Backup Plans

Tablo 2 Risk and Backup Plan

12
10 Mathematical Model and Foundations

No matter which method is used, counting in petri dishes should be done without delay at the
end of the incubation. While seedings in dilutions containing 30-300 colonies were taken into
account before, the number in the sample is calculated by using the weighted arithmetic
average and based on the results of sowing made from two consecutive dilutions recently. The
formula used in this calculation is; N= C / [V(n1 + 0.1 X n2) X d] and colony counts between
15-300 are taken into account.

N = Number of microorganisms in 1 g or 1 ml of food sample

C = Total number of colonies in all counted petri dishes

V = Volume transferred into counted petri dishes (ml)

n1= Number of petri dishes counted from the first dilution

n2= Number of petri dishes counted from the second dilution

d = The dilution ratio of the more concentrated of the 2 consecutive dilutions counted.

In the new terminology, the number obtained as a result of counting is given as cfu/g, not as
"pieces/g". Microorganisms that cannot grow and form colonies can be found in the medium
used, although they are alive. It is more accurate to use (Colony Forming Unit ; cfu) to indicate
that these are not counted and only colony-formers are counted.

Example 1: The bulk cultural count method was applied and each dilution was cultivated in 2
petri dishes. 190 and 163 colonies at 10-2 dilutions, 25 and 16 colonies at 10-3 dilutions were
obtained.

C = 190+163+25+16= 394

V = 1 (1 ml was pipetted into petri dishes for bulk cultural enumeration)

n1= 2 (2 petri dishes planted from 10-2 dilutions were evaluated)

n2= 2 (two petri dishes planted from 10-3 dilutions were evaluated)

d = 10-2 (which is the dilution ratio of 2 consecutive dilutions to the more concentrated one)

N = C / [V(n1 + 0,1 X n2) X d]= 394 / [(2 + 0,1 X 2) X 10-2]= 394 / (2,2 X 0,01)= 17909

This value should be shown with 1 decimal after the comma and the result should be given as
1.8X104 cfu/g (ml if liquid).

13
Example 2: If the above sample smear was made by cultural counting method, and the same
number of colonies were obtained in petri dishes, only V=0.1 ml and the count result would be
calculated as 1.8X105 cfu/g.

Example 3 : The smear method was used in a liquid food and 280, 310 colonies were obtained
from the original sample (10o dilution), 12 and 16 colonies were obtained from the 10-1
dilution.

C = 280+16= 296 (more than 300 and less than 15 colonies are not taken into account)

V = 0.1

n1= 1 (1 of 2 petri dishes was evaluated for counting)

n2= 1 (1 of 2 petri dishes was evaluated for counting)

d = 10o = 1 (original sample was used).

N = 296 / [0.1(1 + 0.1 X 1) X 1] = 296/0.11 = 2690 = 2.7X103 cfu/ml

Tablo 3 Sample Counting

14
11 Design and Technical Drawings
While designing the Cytodeep device to be able to count with high accuracy, the primary goal
was to get high quality images. In order to achieve this goal, we first determined which
parameters were important for us with a mechanism we established. Determined parameters
and calculations;

1. The distance between the camera and the agar should be 170mm,

2. The required distance between the LEDs and the agar is 168.5mm,

3. It was calculated that the angle at which the light should reach the agar should be 72.5
degrees.

After these calculations, the components to be included in the device and the need for
protection from external light during counting were foreseen and the device was designed using
SolidWorks software Şekil 6. The design consists of 8 parts in order to facilitate the work on
the electronics. After the design is completed, the device will be three-dimensionally printed.
PLA (Polylactic Acid) type filament will be preferred in printing because it is not harmful to
human health and is durable.

Şekil 6 Exploded Design of Cytodeep

15
 Şekil 7 Circuit Design

Şekil 8 Technical Drawing of Cytodeep

16
Şekil 9 Technical Drawing of Cover

17
Şekil 10 Technical Drawing of Lower Body

Şekil 11 Technical Drawing of Upper Body

18
Şekil 12 Technical Drawing of Petri Part

19
 Şekil 13 Technical Drawing of Middle Body

Cytodeep Drawings

20
12 BOM Bill of Materials List
N Model and Unit price Quantity Total Reason
o Name and Unit amount
Raspberry Pi It will be used to
1 Original Camera 1 Pieces image the agar
Module V2 703,56 TL 703,56 TL surface.

It will be used for

2 Servo Motor 1 Pieces opening and closing
MG90S
63,01 TL 63,01 TL the device cover.

3 It will provide servo

Original Arduino 1 Pieces motor control and
Uno R3
504,33 TL 504,33 TL adjustment of angles

4 It processes the data

Raspberry Pi 4 8GB from the device and
- Model 4B
5096,89 1 Pieces
5096,89 transfers it to the
TL TL
computer.

5 Raspberry Pi High 1 Pieces It will provide a

Quality Camera 8- better quality agar
50mm Zoom Lens 1218,89 1218,89 image.
TL TL

6 White LED Strip 5 Meters It will be used for

20,00 TL 100,00 TL interior lighting.

7 Esun Filament 5 Pieces For 3D print

293,54 TL 1467,70
TL

8 Motip Plastic 1 Pieces It will be used for

Surface Primer 400
92,00 TL 92,00 TL easier dyeing of the
ml prototype print.

9 Akçalı Spray Paint It will be used for

400 ml
85,00 TL 3 Pieces
255,00 TL painting the
prototype print.

10 Politek Polyester 1 Pieces Fix printing errors

Steel Putty 1 KG
100,00 TL 100,00 TL

11 Adapter It will be used for

185,00 TL 1 Pieces
185,00 TL power and
integration
connection of the
device.

21
N Model and Unit price Quantity Total Reason
o Name and Unit amount
12 Rechargeable 9 Volt For the energy needs
Lithium-ion battery
135,00 TL 1 Pieces
135,00 TL of the device

13 Baird-Parker It will be used to train

medium
12,14 TL 1 Package
85,00 TL the Test and Deep
Learning model.

14 Macconkey medium It will be used to train

8,00 TL 1 Package
80,00 TL the Test and Deep
Learning model.

15 Tryptic Soybean It will be used to train

medium
8,00 TL 1 Package
80,00 TL the Test and Deep
Learning model.

16 Hektoen Enteric It will be used to train

medium
8,00 TL 1 Package
80,00 TL the Test and Deep
Learning model.

17 Cetrimide medium It will be used to train

8,00 TL 1 Package
80,00 TL the Test and Deep
Learning model.

18 Petri dish It will be used to

4,00 TL 1 Package
60,00 TL prepare the dataset.

19 Membrane filter 10 Pieces It will be used in the

42,00 TL 420,00 TL testing phase.

Total : 10806,38
TL

Tablo 4 Bill of Materials

22
13 Milestones, Work Packages and Intermediate Deliveries

Şekil 14 Trello Summarize

14 Project Time Plan and Schedule

Şekil 15 Time Plan and Schedule

23
15 Literature Survey, Patent Search

1) Hogekamp L, Hogekamp SH, Stahl MR (2020) Experimental setup and image processing

method for automatic enumeration of bacterial colonies on agar plates. PLoS ONE 15(6):

e0232869. https://doi.org/10.1371/journal. pone.023286

2) Ayanoğlu, K. (2011). Gıda Teknolojisi Koloni Sayımı. Ankara: MEB.

3) Waltman, W. D. (2000). Methods for the cultural isolation of Salmonella. Salmonella in

domestic animals, 355-372.

4) Goss, W. A., Michaud, R. N., & McGrath, M. B. (1974). Evaluation of an automated colony

counter. Applied Microbiology, 27(1), 264-267.

5) Biston, M.-C., Corde, S., Camus, E., Marti-Battle, R., Esteve, F., & Balosso, J. (2003). An

objective method to measure cell survival by computer-assisted image processing of numeric

images of Petri Dishes. Physics In Medicine And Biology, 1551-1563.

6) Boukouvalas, D. T., Prates, R. A., Leal, C. R. L., & de Araújo, S. A. (2019). Automatic

segmentation method for CFU counting in single plate-serial dilution. Chemometrics and

Intelligent Laboratory Systems, 195, 103889.

7) Vongmanee, N., Bunmak, A., & Visitsattapongse, S. (2018, May). An Automated Colony

Counter for Serial-Dilution Culture Method. In Proceedings of the 2018 10th International

Conference on Bioinformatics and Biomedical Technology (pp. 42-45).

8) Peeler, J. T., Leslie, J. E., Danielson, J. W., & Messer, J. W. (1982). Replicate Counting Errors

by Analysts and Bacterial Colony Counters. Journal of Food Protection, 45(3), 238-240. doi:

9) Goutte, C., & Gaussier, E. (2005, March). A probabilistic interpretation of precision, recall

and F-score, with implication for evaluation. In European conference on information retrieval

(pp. 345-359). Springer, Berlin, Heidelberg.

24