REPORT FOR MAJOR PROJECT

ON

FETAL HEART DISEASES DETECTION USING MACHINE

LEARNING ALGORITHM

Under the supervision

of

Ms. Prachi Kaushik

Assistant Professor Department of E.C.E

SUBMITTED BY

Shobhit Gupta (08020802820)

Tushar Verma (08920802820)
Rahul Patel (06920802820)
Prince (06720802820)

Electronics & Communication Engineering Department,

Bhagwan Parshuram Institute of Technology,
Rohini, New Delhi
 Table of Content

Introduction ............................................................................................................................. 1

Proposed Work ........................................................................................................................ 2

Proposed Timeline of Project .................................................................................................. 3

Work done till date .................................................................................................................. 4

Reference................................................................................................................................. 5

i
 List of Figures

Fig.1. PERT Chart of Project Timeline ................................................................................ 3

ii
Introduction

Fetal Electrocardiograms (FECGs) provide essential insights into the heart's electrical activity,
aiding in the identification of potential arrhythmias. Real-time monitoring of FECG signals beyond
hospital settings is crucial for the prompt detection and treatment of cardiovascular diseases. The
proliferation of portable FECG devices, such as chest straps and wristbands, has led to a substantial
increase in the volume of collected FECG data. Accurate and swift arrhythmia detection is vital,
necessitating effective feature extraction and classification algorithms.

Traditional methods rely on artificial feature extraction techniques, such as discrete wavelet
transform (DWT) and support vector machines (SVM). However, these approaches present
challenges, including dependence on artificial features and complex calculation processes. In recent
years, the emergence of deep learning has transformed one-dimensional data processing, prompting
exploration into neural networks for FECG signal analysis.

This study introduces an arrhythmia classification method based on an improved deep residual
network. The proposed approach segments FECG signals into 5-second intervals using an
overlapping segmentation method to address class imbalance. Employing discrete wavelet
transform (DWT) for denoising, the improved deep residual convolutional neural network serves
for arrhythmia classification. Additionally, the focal loss function is utilized to overcome
challenges associated with imbalanced classifications between different arrhythmia classes. The
enhanced deep residual network not only deepens the network without compromising performance
but also enhances feature diversity through the use of multiple convolution kernels of varying sizes.

1
Proposed work

The proposed initiative aims to develop a sophisticated system for classifying a range of heart
diseases, utilizing an enhanced deep residual network. The key components of the system include:
1. Data Acquisition:
Collect diverse cardiac signals, such as electrocardiograms (FECG), from portable devices like
chest straps and wristbands.
2. Data Preprocessing:
Employ an overlapping segmentation method to divide FECG signals into 5-second intervals,
addressing class imbalance.
Apply discrete wavelet transform (DWT) to denoise the segmented FECG signals.
3. Feature Extraction:
Extract relevant features from pre-processed FECG segments.
Leverage a deep residual network to autonomously capture discriminative features crucial for heart
disease classification.
4. Heart Disease Classification Model:
Implement an advanced deep residual convolutional neural network, incorporating diverse
convolution kernels for enhanced feature perception.
Integrate the focal loss function to address imbalanced classifications across different heart disease
categories.
5. Training and Validation:
Split the dataset into training and validation sets for iterative model training and optimization.
6. Evaluation Metrics:
Utilize standard metrics, including accuracy, precision, recall, and F1 score, to assess the
classification model's performance.
7. Result Analysis:
Analyse classification outcomes, evaluating the model's ability to accurately detect and categorize
various heart diseases.
8. Deployment:
Deploy the heart disease classification system for real-time monitoring of diverse cardiac signals.
This system, with its enhanced deep residual network and robust feature extraction, aspires to offer
an accurate and efficient solution for the automatic classification of various heart diseases. The
focus extends beyond arrhythmia, encompassing a broader range of cardiac conditions to contribute
to proactive and precise heart health monitoring and diagnosis.

2
Proposed Timeline of Project

Fig.1. PERT Chart of Project Timeline

3
Work done till date

Right now, our project is doing really well, and we're at Step 3 of our plan. To explain what we've
been up to: first, we gathered a bunch of different heart signals using devices like chest straps and
wristbands. Then, we took a close look at the data and made it cleaner and more balanced through
a careful process. Now, we're in the thick of it, working on something called "feature extraction"
in Step 3.

Feature extraction is like finding the most important parts in the heart signals that can help us tell
if there's a problem. We're using a smart system to do this, and what we discover in this step will
be the building blocks for the next part of our project.

Our plan for the next steps is to create a really advanced system (we call it a deep residual
convolutional neural network) in Step 4. This system will be super smart and will help us see even
more details from the heart signals. Once we're done, we want to use this system to keep an eye on
different heart signals in real-time. Our big goal is to make a system that can automatically tell if
there's a problem with the heart, helping doctors and patients stay on top of things for better heart
health.

4
References

[1] Hasan, Tabreer T., Manal H. Jasim, and Ivan A. Hashim. "Heart Disease Diagnosis System
based on Multi-Layer Perceptron neural network and Support Vector Machine." (2022).
[2] Zhang, Yang, and Zhidong Zhao. "Fetal state assessment based on cardiotocography parameters
using PCA and AdaBoost." 2021 10th International Congress on Image and Signal Processing,
BioMedical Engineering and Informatics (CISP-BMEI). IEEE, 2021.
[3] Srinivasarao Kakumanu, D. L. P. M. R. D. . (2021). Recognition of Fetal heart diseases through
machine learning techniques.
[4] Dr. P. Ratna Babu, P. Lokaiah, An effective noise reduction technique for class imbalance
classification, International Journal of PsychosocialRehabilitation, vol24, Issue 04, 2020.
[5] P. Lokaiah, S. Nyamathulla, M. Kiran kumar, KBV Rama Narasimham. Performance
Evaluation of Different Machine Learning Techniques for Prediction of Diabetes., Journal of
Critical Reviews, vol7, Issue 18, 2020.
[6] Khanna, Dishant, and Arunima Sharma. "Kernel-Based Naive Bayes Classifier for Medical
Predictions." Intelligent Engineering Informatics. Springer, Singapore, 2018. 91-101.
[7] Warmerdam, G. J. J., et al. "Detection rate of fetal distress using contraction-dependent fetal
heart rate variability analysis." Physiological measurement 39.2 (2018): 025008.
[8] Fergus, Paul, et al. "Classification of caesarean section and normal vaginal deliveries using fetal
heart rate signals and advanced machine learning algorithms." Biomedical engineering online 16.1
(2017): 89.
[9] Nagendra, Vinayaka, et al. "Evaluation of support vector machines and random forest classifiers
in a real-time fetal monitoring system based on cardiotocography data." 2017 IEEE Conference on
Computational Intelligence in Bioinformatics and Computational Biology (CIBCB). IEEE, 2017.
[10] Georgoulas, George, et al. "An ordinal classification approach for CTG categorization." 2017
39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society
(EMBC). IEEE, 2017.