GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY

Department of Computer Science and Engineering

Cardiovascular Disease Detection From ECG

Guide Name PRESENTED BY

1
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

ABSTRACT
The electrocardiogram (ECG) is one of the most important tools for identifying problems with the heart. Most ECG records
were earlier kept on paper. It might be challenging and time-consuming to manually review ECG paper records. We can
automate analysis and diagnosis if we digitize these ECG paper records. In order to display the electrical activity of the heart
using various methods, the project aims to apply machine learning to transform paper ECG records into an I-D signal and
extract P, QRS, and T waves that are present in ECG signals. After feature extraction, it can help with the diagnosis of the
majority of heart diseases. The process involves a series of steps, including breaking down the original ECG report into 13
leads, transforming it into signal data, applying smoothing techniques, converting it into binary images through thresholding and
scaling. After extracting features from these images, dimension-reduction methods such as Principal Component Analysis are
utilized to gain insights into the data. Subsequently, multiple classifiers such as k-nearest neighbours (KNN), Logistic
Regression, Support Vector Machine (SVM),xgboost and a Voting-Based Ensemble Classifier are put to use. The model's
finalization depends on meeting specific criteria regarding accuracy, precision, recall, F1-score, and support. Ultimately, this
finalized model plays a vital role in diagnosing cardiac conditions, determining whether a patient has experienced or is
experiencing Myocardial Infarction, Abnormal Heartbeat, Normal, History of Myocardial infarction or enjoys good cardiac
health by analyzing ECG reports. Without any delay, the results are instant

2
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

Proposed Approach:
Data Collection and Pre-processing:ECG data, including images and associated clinical information, were
collected and pre-processed. This data often included ECG images and corresponding patient health records.
Feature Extraction:Relevant features were extracted from ECG images using various techniques. In the case of
ECG images, features might include amplitude, duration, intervals, and shape characteristics of specific ECG
waveforms
Training and Validation:Splitting the dataset into training and validation sets. Training the model on the
training set and fine-tuning hyperparameters.
Model Selection: Choose an appropriate machine learning model architecture for ECG image classification,
such as k-Nearest Neighbour(KNN),SVM,logistic regression,xgboost by ensembling classifcation method
Evaluation: Assessing the model's performance using metrics like accuracy, precision, recall, and F1-score.
Analyzing any biases or limitations in the model's predictions..
Deployment: Create a user-friendly interface for healthcare professionals to input ECG images and view
diagnostic results.
3
Architecture:

4
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

Design Methodology:
Data Collection: Gather a different and agent dataset of ECG recordings, counting both normal and abnormal
cases. Ensure that the information is labeled with diagnostic information, indicating the sort of cardiovascular
illness (e.g., Abnormal Heartbeat, myocardial infarction, heart failure, Normal, History of Myocardial
infarction).
Data Pre-processing: Clean and pre-process the ECG images to ensure data quality and consistency. This may
involve noise reduction, resizing, and standardization of format.
Feature Extraction: Extract relevant features from the pre-processed ECG images. These features should
capture important patterns related to cardiovascular health. Feature extraction methods can include wavelet
transforms, signal processing techniques.
Dataset Splitting : Divide the dataset into training, validation, and test sets. The training set is used to train the
machine learning model, the validation set helps in tuning hyperparameters, and the test set is for evaluating the
model's performance.

5
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Design Methodology:
Model Selection: Choose an appropriate machine learning model architecture for ECG image classification.
From ensembling classification
Model Training: The data is divided into training set, validation set and testing set. Demonstrate model
selection using training data using appropriate loss functions and optimization techniques. Normalize the model
to avoid over fitting. Monitor training progress and adjust hyperparameters to adjust as needed..
Model Evaluation: Evaluate the model's performance on validation and testing data using appropriate metrics
(e.g., accuracy, sensitivity, specificity, F1 score, AUC-ROC).
Deployment: Create a user-friendly interface or integrate the model into healthcare systems to allow healthcare
professionals to input ECG images for automated disease detection.
Validation and Testing: Collaborate with healthcare experts to validate the model's predictions against clinical
diagnoses. Ensure that the model performs reliably and consistently in real-world scenarios.

6
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Module Description:
• Data Collection and Pre-processing Module:
• Data Acquisition: Collect and compile a diverse dataset of ECG images from various sources.
• Data Cleaning: Pre-process the dataset by removing noise, standardizing formats, and ensuring data
quality.
• Feature Extraction Module:
• Image Pre-processing: Prepare ECG images for feature extraction, including resizing, normalization,
and noise reduction.
• Feature Extraction Algorithms: Implement algorithms to extract relevant features from ECG
images. Common techniques include wavelet transforms, histogram analysis, or deep learning-based
feature extraction.
• Model Development Module:
• Model Selection: Choose an appropriate machine learning model architecture for ECG image
classification, such as k-Nearest Neighbour(KNN),SVM,logistic regression,xgboost by ensembling
classifcation method
• Model Training: Train the selected model using the preprocessed dataset, defining the loss function
and optimization strategy.
• Hyper parameter Tuning: Optimize model hyper parameters, including learning rates, batch sizes,
7
and network architecture, to improve performance.
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Module Description:
• User Interface and Deployment Module:
• User Interface Development: Create a user-friendly interface for healthcare professionals to input
ECG images and view diagnostic results.
• Integration: Integrate the machine learning model into healthcare systems, making it accessible to
users.
• Validation and Testing Module:
• Clinical Validation: Collaborate with healthcare experts to validate the model's predictions against
clinical diagnoses, ensuring real-world applicability.
• Testing: Conduct thorough testing, including unit testing, integration testing, and system testing, to
identify and resolve issues.

8
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Implementation:

9
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Implementation:
k-nearest neighbors (KNN): - Logistic Regression: -

10
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Implementation
Support Vector Machine (SVM): - XGBoost: -

11
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

Implementation
Voting Based Ensemble Classifier : -

12
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

UML Diagrams:

13
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

Scope Limitations:

1. Early Detection:The model aims to contribute to the early detection of cardiovascular diseases, enabling
timely intervention and improving patient outcomes.
2. Diagnostic Support:It serves as a diagnostic support tool for healthcare professionals, assisting them in
the interpretation of ECG images and providing additional insights.
3. Image quality: The performance of ECG-based tests largely depends on the quality of the ECG image.
Low-quality images may reveal important details, making it difficult to detect suspicious items.
4. Limited data: ECG images generally represent the electrical activity of the heart. Although they can be
useful in some cases, they may not have the important information needed to diagnose heart diseases
associated with abnormal processes or other physiological factors.
5. Continuous Model Maintenance:Regular updates and continuous monitoring are required to keep the
model accurate and up-to-date with advancements in the field and changes in the patient population.

14
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
References:
• Alty, S. R., Millasseau, S. C., Chowienczyk, P. J., & Jakobsson, A. (2003). Cardiovascular disease prediction
using support Vector Machines. 2003 46th Midwest Symposium on Circuits and Systems.
https://doi.org/10.1109/mwscas.2003.156229
• 7 Asadi, S., Roshan, S. E., & Kattan, M. W. (2021). Random forest swarm optimization-based for heart
diseases diagnosis. Journal of Biomedical Informatics, 115, 103690.
https://doi.org/10.1016/j.jbi.2021.103690
• Bhardwaj, R., Nambiar, A. R., & Dutta, D. (2017). A study of machine learning in Healthcare. 2017 IEEE 41st
Annual Computer Software and Applications Conference (COMPSAC).
https://doi.org/10.1109/compsac.2017.164
• Jan, M., Awan, A. A., Khalid, M. S., & Nisar, S. (2018). Ensemble approach for developing a smart heart
disease prediction system using classification algorithms. Research Reports in Clinical Cardiology, Volume
9, 33–45. https://doi.org/10.2147/rrcc.s172035
• H. Bleijendaal et al., "Computer versus cardiologist: Is a machine learning algorithm able to outperform an
expert in diagnosing phospholamban (PLN) p.Arg14del mutation on ECG?", Heart Rhythm, vol. 18, no. 1,
pp. 79-87, 2020, [online] Available: https://doi.org/10.1016/j.hrthm.2020.08.021.
15
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Review Paper:

16
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3
Review Paper:

17
 GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

Conclusion:
In conclusion, this Cardiovascular Disease Detection from ECG Images using Machine Learning holds immense
potential for transforming healthcare by automating the diagnosis and early detection of cardiovascular conditions.
And represents a powerful tool for improving cardiac care and has the potential to make a significant positive
impact on public health. It is a dynamic and evolving field with the potential to save lives, reduce healthcare costs,
and enhance the quality of cardiovascular care. The practical outcomes demonstrate that by employing the
provided predictive model on ECG images of new patients, we can achieve quicker and more precise predictions
for individuals with heart conditions. Furthermore, if the image-based feature extraction is carried out effectively
and optimally, along with enhancing the accuracy of our model, this research can be expanded to encompass
various other cardiac diseases.

18
GOKARAJU RANGARAJU INSTITUTE OF ENGINEERING AND TECHNOLOGY
Department of Computer Science and Engineering
Major Project Review-3

THANK YOU

19