Heart Attack Detection Device: Project Details
3 min read
Introduction
Our startup involved around 12 person team aiming to detect heart attacks in under 30 minutes by measuring a biomarker called troponin in the blood. We sought to integrate our detection module with a device called Aria, produced by Pooyandegan Rah Saadat. Our design used a microfluidic cartridge that moves blood samples through multiple layers and chambers, then reads certain signals to identify elevated troponin levels.
We patented our method of linking our microfluidic module to the Aria device in the United States. We also pursued another patent covering the unique way the fluid moves through the cartridge, controlled by a PID-based approach on a microcontroller. This approach gave us precise control, reacting in microseconds to manage fluid flow and ensure accurate measurements.
Core Device Operation
A small sample of blood enters the upper layer of the cartridge, moving through different chambers and waste outlets. Additional layers measure the fluid’s connectivity (using an AD5933 IC) and run it through a 24-bit A/D converter for chemical analysis. By reading troponin concentrations, doctors can decide if a patient is likely experiencing a heart attack.
The core detection was handled by a chemistry and biomedical team. On the electronics side, we had three members: two focused on hardware, while I managed the software on an X-Mega microcontroller. My tasks included:
- Implementing the AD5933 interface for conductivity readings.
- Deploying a custom PID controller to handle microfluidic movements.
- Reading high-precision data through the 24-bit A/D converter.
- Integrating SPI, I2C, and UART protocols with RFID and other peripherals.
Additional Prototypes: Smart Scale & Smart Belt
Our electronics team also developed two side projects:
- Smart Belt: A wearable pedometer that attached to a belt and measured steps. It collected data and sent it via a simple wireless connection to an Android app.
- Smart Scale: Used the AD5933 IC with electrodes to measure body fat and total body water. This data was then shown to users through an Android application. It gave a broader picture of body composition beyond just weight.
These prototypes built on some of the same ideas around reading signals accurately with minimal noise, then passing results to a user-friendly interface.
Key Challenges & Contributions
One major challenge was precision in fluid movement. Our microfluidic cartridge relied on exact timing—sometimes in the microsecond range—to stop fluid where needed. I implemented the microcontroller logic that paused, advanced, or reversed fluid flow as directed by the PID control loop.
Another challenge was ensuring stable readings from the AD5933 and the 24-bit A/D converter, since even minor fluctuations could alter the final troponin measurement. We tested various calibration techniques and noise reduction approaches before finalizing the board design.
Tech Stack
- Microcontroller: X-Mega (Atmel)
- Sensing & Data Acquisition: AD5933 IC for conductivity, 24-bit ADC for chemical readings
- Communication Protocols: SPI, I2C, UART
- Control & Logic: PID-based fluid control, microsecond-level timing