Heart rate detection with CMOS preprocessor based on neural network

Table of Contents

• About
• Getting Started
  • Prerequisites
• Installing the requirements
• Running the code
• Todo
• License

About

Heart rate detection with neural network-based CMOS preprocessor. It involves developing a heart rate detection model using TinyML techniques on the PhysioNet dataset and parsing it on a CMOS preprocessor written in Spice.

Getting Started

These instructions will get you a copy of the project up and running on your local machine.

Prerequisites

Python libraries that you need to use this software:

• WFDB
• Tensorflow

CMOS simulation tools, that are required:

• Eldo
• EZwave

Installing

A step by step series of examples that tell you how to get a development env running. All the installation steps are being handled by the requirements. You can use venv by typing:

$ python -m venv venv

Activate venv and install requierments.

$ pip install -r requirements.txt

Running the code

In order to run the code, you will only need to prepare two files, multiplication coefficients grid and model file.

prepare_dataset_pipeline.py

$ python prepare_dataset.py -h
usage: prepare_dataset.py [-h] --ds-dir DS_DIR --output-file OUTPUT_FILE --sample-length SAMPLE_LENGTH --sample-freq SAMPLE_FREQ [--download]

options:
  -h, --help            show this help message and exit
  --ds-dir DS_DIR       Directory with ECG data from physionet (default: None)
  --output-file OUTPUT_FILE
                        File in which the dataset will be saved (default: None)
  --sample-length SAMPLE_LENGTH
                        The length of a sample's window (default: None)
  --sample-freq SAMPLE_FREQ
                        Number of sampled points from a window (default: None)
  --download            Whether to download database or not (default: False)

parse_model.py

Now, in order to run the code you can call the parse_model.py directly.

$ python parse_model.py -h
usage: parse_model.py [-h] --model MODEL --grid GRID

options:
  -h, --help     show this help message and exit
  --model MODEL  Path to file, that contains TensorFlow model (default: None)
  --grid GRID    Path to file, that contains weight grid (default: None)

Authors

• Piotr Baryczkowski - ANN to CMOS preprocessor parser implementation - Piotr45
• Sebastian Szczepaniak - Implementation of dataset pipeline and training ANN - D3nz13

See also the list of contributors who participated in this project.

TODO

• clean parser code

License

This project is licensed under the MIT License - see the LICENSE file for details