Detection of BABESIA BIGEMINA in Cattle Blood: AI and Impedance

Methods

Jacob Varghese

1,2 a

, Allen George Thomas

1,2

, Khadeeja Nilofer Saleem Mukkunnoth

1,2

Sivan Murali

1,2

and Lakeshmy G B

1,2

TKM Institute of Technology, APJ Abdul Kalam Technological University, Thiruvananthapuram, Kerala, India

Department of Biomedical Engineering, TKM Institute of Technology, Kollam, Kerala, India

Keywords:

Anaplasma, Babesia Bigemina, Haemoprotozoans, IoU, Mean Average Precision, Theileria, YOLO V8.

Abstract:

Haemoprotozoans are a diverse group of blood-borne parasites that cause signiﬁcant economic losses in the

veterinary ﬁeld. In cattle, the three most common haemoprotozoans are Babesia, Theileria, and Anaplasma.

Detection and treatment of these parasites are currently time-consuming and require laboratory facilities,

which can delay treatment and lead to poorer outcomes, including increased anaemia and death. Infections

also lead to diminished productivity, compromised reproductive performance, and increased vulnerability to

secondary infections. To address this challenge, the project aims to develop novel technologies that will help

in early detection of the parasite. The paper presents the design of an embedded AI software to detect the

presence of Babesia bigemina protozoan within the cattle blood. The software used YOLO V8 model to train

the system, and the software was integrated into a 3D printed open ﬂexure microscope. The model yields a

mean average precision of 66.2 percent for an IoU threshold of 0.5 and 34.7 percent for an IoU threshold of 0.5

to 0.9. The project also proposed research on the change in conductivity and impedance of the infected cattle

blood and concluded that the presence of foreign particles, such as protozoans, in the blood samples resulted

in a decrease in conductivity by values ranging between 2.2 to 3 milli siemens and an increase in impedance

by a value within a range of approximately 330 to 450 milli ohm compared to normal blood samples.

1 INTRODUCTION

Haemoprotozoans are a diverse group of single-celled

eukaryotic organisms transmitted by blood-feeding

invertebrates, causing diseases in various animals

and humans. They include genera like Babesia,

Hepatozoon, Theileria, and Trypanosoma, with dif-

ferent species identiﬁed in wildlife such as cattle,

dog and rats(A. S. Nair, 2011). Haemoprotozoan

species affecting cattle include Theileria annulata,

Trypanosoma evansi, and Babesia bovis. These par-

asites cause signiﬁcant economic losses globally due

to their impact on livestock health and productivity.

Theileriosis, trypanosomosis, and babesiosis are ma-

jor haemoprotozoan diseases, with difﬁculties in di-

agnosis due to low parasitemia and concurrent in-

fections. Tick-borne diseases, such as babesiosis,

anaplasmosis, and theileriosis, are major constraints

in the dairy industry, leading to decreased produc-

tivity and increased control costs. The hot and hu-

https://orcid.org/0009-0001-2862-681X

mid climate in tropical regions like India provides a

favorable environment for haemoprotozoan parasites

transmitted by vectors like ticks, posing a constant

threat to susceptible animals(A. Tlili and Jaffrezic-

Renault, 2006). Haemoprotozoa species detected in

cattle in Northern Kerala were Theileria like piro-

plasms and Babesia bigemina, with PCR revealing

Trypanosoma evansi, Theileria sp., and B. bigem-

ina(A. S. Nair, 2011). Detection and diagnosis of

these diseases can be challenging as they require lab-

oratory facilities and time consuming that even in-

creases the rate of mortality. There is an alarming

need of an on-ﬁeld device that will accurately and pre-

cisely detect the presence of haemoprotozan within

the animal blood. This concept is the main build-

ing block of our device(C. S. Bhatnagar and Meena,

2015). A Non-invasive portable device can do the vet-

erinarians and the farmers a good turn as they give out

rapid results in the ﬁeld detection fundamentally. The

results would not only be rapid but also with tower-

ing accuracy. In particular this device facilitates to

characterize to redline the different types of proto-

Varghese, J., Thomas, A. G., Mukkunnoth, K. N. S., Murali, S. and G B, L.

Detection of BABESIA BIGEMINA in Cattle Blood: AI and Impedance Methods.

DOI: 10.5220/0013584700004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 1, pages 729-734

ISBN: 978-989-758-763-4

729

zoans and its variations. Finding tick-borne diseases

in cattle while out in the ﬁeld poses signiﬁcant chal-

lenges. Typically, diagnosing these diseases requires

access to laboratory facilities equipped with special-

ized tools like microscopes for examining blood sam-

ples and PCR (Polymerase Chain Reaction) tests for

identifying speciﬁc pathogens(B. R. Maharana and

Sudhakar, 2016). However, such facilities may not

always be readily available, especially in remote or

rural areas where these diseases are prevalent. Fur-

thermore, the cost associated with conducting these

tests can vary depending on factors such as the type

of test required and the location of the testing facility.

In some cases, farmers may need to travel long dis-

tances to access testing centres, incurring additional

expenses and time delays. Unfortunately, the delay

in initiating treatment procedures due to the time-

consuming nature of laboratory testing can have se-

vere consequences. Which may even lead to the death

of the animal, causing a devastative economic loss for

the farmer. The primary outreach of this project is

to concisely overcome the current challenges in de-

tecting the presence of haemoprotozoan within the

cattle blood, especially during the on-ﬁeld examina-

tion. Thereby narrowing down the constraints which

leads to the lagging of treatment initiation(B. R. Ma-

harana and Sudhakar, 2016). Hence, the main aim of

this project is to develop an on ﬁeld low-cost portable

device that could accurately and precisely detect the

presence of protozoans from the blood sample under

study. Additionally, the device may also be able to

differentiate the genera of protozoans, since the drug

administration for each protozoan is speciﬁc. The

main objectives of the project are Analysing the re-

lationship between protozoan infection and change

in the electrical conductivity and impedance property

of the blood due to the infection(I. Szyma

nska and

Kaliszan, 2007). Developing a novel methodology

based on the observation from the ﬁrst objective to de-

sign a circuit to detect and measure the corresponding

change in blood properties(G. C. McConnell, 2009).

Designing and developing a software to accurately de-

tect each genus of protozoan and concluding whether

the blood sample under study is protozoan infected or

not(B. K. Yap, 2018). For the time being the study

only concentrates on one species. To successfully im-

plement the designed software to a low-cost portable

3D printed microscope with high resolution.

2 METHODOLOGY

The proposed solution for the particular problem that

this project addresses, can be done mainly in two

Figure 1: Circuit diagram of Microcontroller based Human

Blood Conductivity Measurement System

ways, that is, one complete hardware and the other

an AI software system embedded with 3D printed mi-

croscope. In this chapter the design of the hardware,

components used to develop the circuit, role of each

component in the circuit etc will be discussed ﬁrst.

Following session of the same chapter will introduce

about the software design as well as the design and

structure of the 3D printed microscope.

2.1 System Hardware: Detection of

Haemoprotozoan in Cattle Blood by

Analyzing the Conductivity and

Impedance of the Blood Sample

This project aims to designs a circuit to measure

the conductivity and impedance of liquids(B. S.

R. Bharati and Bhaskar, 2013), particularly blood

samples, using important components like the LM741

op-amp, Arduino ATmega328, and a display as men-

tioned in ﬁgure 1. The system consists of three parts,

namely:

• Signal Conditioning Circuit

• Conductivity Cell (Electrodes)

• Microcontroller and Display

Working Principle:

An AC signal is imposed on electrodes immersed in

the sample. The ions in the liquid facilitate the ﬂow of

current, and the voltage across the electrodes reﬂects

the liquid’s impedance. The resulting signal is am-

pliﬁed, rectiﬁed, and converted to a readable format,

from which conductivity is calculated(McAdams and

Jossinet, 1995).

Signal Conditioning:

The signal conditioning unit consists of

• Wein Bridge Oscillator: It provides a 1kHz sine

wave at an amplitude of 10Vpp by using the

LM741 op-amp and an RC feedback network.

The frequency is given by

f =

2πRC

(1)

INCOFT 2025 - International Conference on Futuristic Technology

730

• Impedance Bridge: An AC signal from the oscil-

lator excites the bridge, one arm of which con-

tains the sample. Sample conductivity changes

the impedance of the bridge, hence the voltage

distribution.

The last component includes the Ampliﬁcation and

Processing which is followed by rectiﬁcation and ﬁl-

tering to get a DC voltage representative of conduc-

tivity. This DC voltage is digitized through an ADC

for processing.

2.2 System Software: Real Time

Monitoring and Detection of the

Haemoprotozoan from the Cattle

Blood Sample Using Low-Cost

Portable Microscope Embedded

with AI Software

2.2.1 Blood Cell Classiﬁcation Software

The developed software to classify the blood cells in-

fected with haemoprotozoan will be architected to ef-

ﬁciently handle various tasks involved in processing

the data, training and evaluation of models, and mak-

ing inference(J. Knapper, 2022). The followings are

included:

1. Modular Components: Dedicated modules for

augmentation, model creation, training, evalua-

tion, and inference of data. Reusability in func-

tions and classes promotes maintainability and

reusability of code.

2. Flexibility: Allows customization of model ar-

chitectures, hyperparameters, pre-trained models,

loss functions, and optimization techniques.

3. Performance Optimization: Batch processing,

parallelization, and GPU acceleration are em-

ployed to ensure efﬁciency(C. Honrado, 2018).

4. Error Handling and Platform Independence: Ro-

bust error handling ensures smooth operation,

while platform independence allows compatibil-

ity across operating systems.

2.2.2 Camera Module

The camera module includes a 4-megapixel CMOS

sensor that can capture images at 1920×1920 and

record videos at 640×480 pixels as shown in ﬁg-

ure 2. This high resolution enables detailed analysis

and live demonstrations. The CMOS sensors convert

light into a digital signal, providing high-quality im-

ages suitable for microscopy applications(Berney and

O’Riordan, 2008).

Figure 2: CMOS sensor obtained from a 1600x digital USB

microscope.

2.2.3 Parasite Detection Methodology

1. Data Preparation: Blood sample images contain-

ing Babesia parasites are formatted in the YOLO

structure, with annotation ﬁles marking bounding

boxes around parasites. The dataset is split into

training and validation sets.

2. Model Selection: YOLOv8, a state-of-the-art ob-

ject detection model, is chosen for its high accu-

racy and real-time detection capabilities.

3. Model Training: The model is trained for 100

epochs using the AdamW optimizer. Data aug-

mentation techniques involved include blur, me-

dian blur, grayscale conversion, and CLAHE to

enhance robustness.

4. Evaluation: Model performance was measured in

terms of precision, recall, and mean average pre-

cision.

5. Deployment: The developed model is deployed

after validation in the identiﬁcation of haemopro-

tazoon infections from blood sample images.,

6. Improvement: To perform well continuously,

make iterations in ﬁne-tunes, additional data, or

advance the object detection algorithm.

2.2.4 Affordable Portable Microscope Design

The microscope utilizes a 3D-printed Open Flexure

Microscope which resolves samples 2–5 µm in size.

This open-sourced created, by Dr. Richard Bowman

at the University of Bath, is a cost effective, portable

solution, well suited for education and research. Im-

ages are acquired live and analyzed by the software to

identify in real time infections caused by haemoproto-

zoa(J. Knapper, 2022), (C. Honrado, 2018), (Berney

and O’Riordan, 2008).

Detection of BABESIA BIGEMINA in Cattle Blood: AI and Impedance Methods

731

Figure 3: Prototype of the hardware

Figure 4: PCB and circuit components

3 RESULT

3.1 Blood Conductivity and Impedance

The objective of this project is to measure the conduc-

tivity and impedance of blood samples using a con-

ductivity measuring circuit with a Wein bridge oscil-

lator, LM741 IC, copper electrode arrangement, and

an Arduino microcontroller as shown in the ﬁgure 3.

A conductivity measuring circuit was constructed

comprising a Wein bridge oscillator with an LM741

IC as the signal source generator. A copper electrode

arrangement was used as the conductivity cell, im-

mersed in the blood samples. An Arduino microcon-

troller was employed to process the signals and cal-

culate conductivity and impedance values. Figure 4

shows the PCB fabrication.

3.1.1 Observations

The conductivity measurements ranged from 2.2 ms

to 5.8 ms. A total of 20 readings were obtained within

this range, with incremental changes in conductivity

values. Impedance values were calculated as the re-

ciprocal of conductivity.

The presence of foreign particles, such as proto-

zoans, in the blood samples resulted in a decrease in

conductivity to a range of 2.2 to 3ms and an increase

in impedance within a range of 330 to 450 mohms

compared to normal blood samples. This suggests

that the electrical properties of blood are inﬂuenced

by the presence of contaminants, highlighting the im-

Table 1: Observations of conductivity and impedance mea-

surements of blood samples.

Reading Normal Blood Sample Infected Blood Sample

Conductivity (mS) Impedance (mω) Conductivity (mS) Impedance (mω)

1 4.4 227.27 2.8 357.14

2 4.35 229.89 3.2 312.50

3 4.3 232.56 3.0 333.33

4 4.25 235.29 2.4 416.67

5 4.2 238.10 2.9 344.83

6 4.1 243.90 3.4 294.12

7 4.15 240.96 2.3 434.78

8 4.3 232.56 3.3 303.30

9 3.95 253.60 2.6 384.62

10 3.8 263.16 2.2 454.44

portance of monitoring conductivity and impedance

for detecting abnormalities.

3.2 Result of the Software System

The project integrated YOLOv8 into a 3D-printed in-

verted microscope to develop an automated system

for detecting Babesia Bigemina in blood samples.

Positive and negative samples were tested, and the

software effectively analyzed, classiﬁed, and detected

the parasite. Due to dataset availability, the study fo-

cused on Babesia Bigemina.

3.2.1 Dataset and Training

The dataset consisted of blood sample images with

bounding boxes for Babesia parasites. Training and

validation sets were made.

• Model Training: YOLOv8 was trained for 100

epochs using the AdamW optimizer with auto-

matic learning rate and momentum.

• Data Augmentation: Techniques include blur,

median blur, grayscale conversion, and contrast-

limited adaptive histogram equalization.

3.2.2 Model Evaluation and Results

The model’s performance was evaluated on the val-

idation set based on metrics such as precision, re-

call, mAP50, and mAP50-95 computed at various IoU

thresholds. Observations can be seen in Table 4.2.

Table 2: Results of the Model.

SL. No Performance Metrics Babesia Extracellular

1 Precision 0.685 0.462

2 Recall 0.736 1.000

3 mAP50 0.892 0.995

4 mAP50-95 0.621 0.895

3.2.3 Metric Deﬁnitions

• Precision: Ratio of true positives to total predicted

positives, measuring prediction accuracy.

INCOFT 2025 - International Conference on Futuristic Technology

732

Figure 5: Precision-Recall Curve

Figure 6: Normalized confusion matrix

• Recall: Ratio of true positives to total actual pos-

itives, assessing the model’s ability to detect all

instances.

• mAP50: Mean average precision at IoU 0.5, pro-

viding a balance between precision and recall.

• mAP50-95: Mean average precision across IoU

thresholds (0.5–0.95), offering a comprehensive

evaluation.

• Class-Speciﬁc Metrics: Precision, recall, and

mAP for individual classes, enabling detailed per-

formance analysis.

3.2.4 Observations

For YOLOv8, satisfactory metrics were obtained with

more room for improvement in detecting Babesia par-

asites. Future optimizations can be done for better re-

sults with increased accuracy and generalization.

4 DISCUSSION AND

CONCLUSION

4.1 Conductivity and Impedance of

Protozoan Affected Blood

The project developed hardware and software tech-

nologies for the detection of haemoprotozoan in cattle

Figure 7: F1- Conﬁdence Curve

blood. The hardware consists of a conductivity mea-

surement circuit consisting of a Wein bridge oscil-

lator, LM741 IC, copper electrodes, and an Arduino

microcontroller. This setup facilitates portable blood

conductivity and impedance measurement, which is

of great value during the identiﬁcation of changes in

blood properties.

While blood impedance measurement is preferred

for the detection of blood-borne parasites, results may

be affected by factors such as temperature and for-

eign particles that limit accuracy. These disadvan-

tages could be overcome by incorporating advanced

sensors and electrodes.

In conclusion, as much as the LM 741-based con-

ductivity circuit is not fully accurate for the detection

of haemoprotozoa, it effectively detects changes in

impedance and conductivity of blood.

4.2 Real Time Monitoring and

Detection of Haemoprotozoan Using

Embedded Ai System

The proposed study is to design a portable, non-

invasive device for the detection of haemoprotozoan

parasites in cattle blood, keeping in mind the limita-

tions of existing methods of diagnosis that are time-

consuming, expensive, and require laboratory facil-

ities. The focus is on the development of an on-

ﬁeld, low-cost device for identifying protozoan gen-

era so that appropriate drugs can be administered.

The study focuses on the relationship between pro-

tozoan infection and changes in blood conductivity

and impedance properties. A software integrated

with a low-cost 3D-printed high-resolution micro-

scope is developed to analyze blood samples and

detect infections. The project emphasizes the po-

tential of biomedical engineering in enhancing vet-

erinary healthcare, supporting cattle farming-a key

contributor to Indian GDP and milk/meat produc-

tion(E. O. Adekanmbi and Srivastava, 2023).

Haemoprotozoan diseases have a major impact on

livestock, resulting in ﬁnancial losses from infections

and mortalities(Garcia and Sabuncu, 2019). Facili-

ties for diagnosis are generally insufﬁcient, relying on

conventional techniques of microscopic examination

and serological tests like ELISA, which are laborious

and time-consuming(Technologies, 2017). Modern

molecular techniques such as PCR detect the disease

during its latent phase more effectively(E. O. Adekan-

mbi and Srivastava, 2023). Electrochemical

Impedance Spectroscopy and immunosensor-based

methods show potential in the diagnosis of diseases

such as Babesia bovis through the analysis of electri-

cal properties(Cole, 1941)(Macdonald and Johnson,

Detection of BABESIA BIGEMINA in Cattle Blood: AI and Impedance Methods

733

2005)(Garcia and Sabuncu, 2019).

Although the novel diagnostic approach was in-

troduced, certain constraints occurred. The presence

of other pathogens, such as viruses, bacteria, or other

protozoan genera, was not considered; temperature

dependencies during experiments were not consid-

ered, either. Low-cost components reduced accuracy,

and software analysis was restricted to Babesia, ex-

cluding complex genera like Theileria and Anaplasma

due to the limitation of the dataset. These factors con-

strained the model’s total mean average precision. Fu-

ture research can address these limitations by replac-

ing low-cost components with precise ICs or sensors

like AD5933(H. Cho and Baek, 2021), designing spe-

ciﬁc electrodes for impedance-based detection, and

expanding datasets through more blood sample col-

lection and annotation.

Enhancing training models to differentiate haemo-

protozoan genera accurately and employing advanced

microscopic technologies like lensless microscopy or

muscope can broaden the device’s applications.

These methodologies could also be applied to the

diagnosis of human parasites such as Plasmodium and

Trypanosoma, laying the foundation for innovative

veterinary diagnostics with signiﬁcant societal and

economic beneﬁts.

REFERENCES

A. S. Nair, R. Ravindran, B. L. e. a. (2011). Haemopro-

tozoa of cattle in northern kerala, india. Tropical

Biomedicine.

A. Tlili, A. Abdelghani, S. A. and Jaffrezic-Renault, N.

(2006). Impedance spectroscopy and afﬁnity measure-

ment of speciﬁc antibody–antigen interaction. Mate-

rials Science and Engineering: C, 26(2):546–550.

B. K. Yap, S. N. A. M. Soair, e. a. (2018). Potential point-of-

care microﬂuidic devices to diagnose iron deﬁciency

anemia. Sensors, 18(8):2625.

B. R. Maharana, A. K. Tewari, B. C. S. and Sudhakar, N. R.

(2016). Important hemoprotozoan diseases of live-

stock: challenges in current diagnostics and therapeu-

tics: an update. Veterinary World, 9(5):487–495.

B. S. R. Bharati, C. S. P. and Bhaskar, P. (2013). Atmel

microcontroller based human blood conductivity mea-

surement system. IJEE.

Berney, H. and O’Riordan, J. J. (2008). Impedance mea-

surement monitors blood coagulation. Analog De-

vices.

C. Honrado, L. Ciuffreda, D. S. e. a. (2018). Dielectric char-

acterization of plasmodium falciparum-infected red

blood cells. Journal of the Royal Society Interface,

15(147):20180416.

C. S. Bhatnagar, B. Bhardawaj, D. S. and Meena, S. K.

(2015). Incidence of haemoprotozoan diseases in cat-

tle in southern rajasthan, india. International Journal

of Current Microbiology and Applied Sciences.

Cole, K. S. (1941). Dispersions and absorption in di-

electrics. Journal of Chemical Physics, 9:341–351.

E. O. Adekanmbi, M. W. U. and Srivastava, S. K. (2023).

Dielectric characterization of babesia bovis using the

dielectrophoretic crossover frequency. Electrophore-

sis, 44(11-12):1001–988.

G. C. McConnell, R. J. Butera, e. a. (2009). Bioimpedance

modeling to monitor astrocytic response to chroni-

cally implanted electrodes. J. Neural Eng., 6.

Garcia, A. and Sabuncu, A. C. (2019). Electrical system for

bioelectric impedance using ad5933 impedance con-

verter.

H. Cho, S. R. L. and Baek, Y. (2021). Anemia diagnostic

system based on impedance measurement of red blood

cells. Sensors (Basel), 21(23):8043.

I. Szyma

nska, H. Radecka, J. R. and Kaliszan, R. (2007).

Electrochemical impedance spectroscopy for study of

amyloid β-peptide interactions. Biosensors and Bio-

electronics, 22(9):1955–1960.

J. Knapper, J. Stirling, D. G. R. e. a. (2022). Smart feedback

for reliable scanning: developing the openﬂexure mi-

croscope.

Macdonald, J. R. and Johnson, W. B. (2005). Fundamentals

of Impedance Spectroscopy, pages 1–26.

McAdams, E. T. and Jossinet, J. (1995). Tissue impedance:

a historical overview. Physiol. Meas., 16:A1–A13.

Technologies, K. (2017). Impedance Measurement Hand-

book, 6th Edition. Keysight Technologies, Santa Rosa,

CA.

INCOFT 2025 - International Conference on Futuristic Technology

734