Kevin Dejbod
A Passionate Biomedical & ML Engineer that loves viewing the world with a pair of maths glasses. This is my portfolio where I share my projects, tutorials, and experiences in the field.
Home
Projects
Professional Experience
Research
Tutorials
Design Portfolio
These are my staple projects that I show! I have tutorials, research, professional projects and open source projects (click on sidebar). For each project if you click on the title you will be taken to the project page with more details.
Projects
Dynamic-ECG : Algorithms for ECG Signal Analysis
My Python-based library for ECG analysis, including R,P,T wave detection, Poincare analysis, wavelets-based analysis and several Visualisation features! Portability with both short ECG & Long Form Holter data. Formats such as NumPy, EDF, H5 and even Apple Watch ECG data (CSV export).
CatchAF: Multi-Modal Atrial Fibrillation Detection Model
An AF detection model that uses Dynamic-ECG for Poincare-Plot generation as the data input for a Computer Vision model. Trained on the IRIDIA-AF dataset, the model achieved a 98% accuracy in detecting AF from ECG data. The model is available for download and use in the CatchAF repository.
Reinforce-CCA-Register: Deep Reinforcement Learning for Cardiac CT-Fluoroscopy Registration
Implemented a deep reinforcement learning model using PyTorch and deep Q network to register 6DOF cardiac CT images with fluoroscopy images. Achieved a 90% success rate in image registration, improving the accuracy of cardiac procedures. Collaborated with CHU de Bordeaux, France, and the UWA Medical Imaging Physics Group, using CARTO EP files.
ML-Urgency System: Real-time Arrhythmia Sagemaker Detection System
Developed a real-time arrhythmia detection system using AWS Sagemaker and Lambda. The system uses a deep learning model trained on the MIT-BIH dataset to detect arrhythmias in ECG data. The model is deployed on AWS Sagemaker and invoked using AWS Lambda functions. The system achieved an accuracy of 98% in detecting arrhythmias in real-time ECG data.
SMART BOOT: Assistive Smart Orthopedic Sensor Device
Constructed Gait-Force frequency Algorithm Development using non-linear differential equation modeling, implemented with Python. Engaged in signal engineering & sensor design. Focused on high bandwidth data optimization. Developed a biosensor area monitoring system. CoLed the clinical prototype Development, including the regulatory & patent application process.