A Collaborative Hardware Platform for Advancing Automatic Interpretation in ECG signals.
(The Kenshin Project)

By Luis Gordillo, Jorge Soto, Hugo Jiménez, David Hernández and Daniela Díaz

The device presented here marks the beginning of the Kenshin Project. It all started with the intention of developing an Expert System, which in Artificial Intelligence is: a computer system capable of emulating the decision-making ability of a human expert.

In the early stage of this project, a block diagram of an Expert System was initially designed. This block diagram can be observed in Figure 11 of the published article. For the original PhD proposal, it was decided to start developing the block for recognizing different waveforms in ECG signals, as this is the first step a human expert takes when interpreting an electrocardiogram. A human practitioner first recognizes a waveform and then extracts its clinical parameters. To understand the need for the aforementioned procedure, it’s necessary to acknowledge that many waveforms exist in ECG signals. From the device’s perspective, these waveforms appear at random, as it’s not possible to predict the pathology the next patient will present.

Before developing our waveform recognition algorithm, it was necessary to define the signal conditioning algorithms to be used, as the cleaner the signal, the better. In the realm of signal conditioning, thousands of algorithms have already been published. Most of these publications report high efficiency in their performance. Unfortunately, as is common in most publications, the source code for the published methods is only available if you write it yourself.

For the case of our waveform recognition method, the problem became even more complex. Since in order to have the different waveforms we required for testing, we needed to use various signal databases, all available at pyshionet.org. Each of these databases had unique characteristics, such as different sampling rates and system noise levels, which led to significant algorithm compatibility issues. Upon realizing this, we decided to take on a new challenge: to standardize ECG signals and promote academic collaboration. The result is this, the Kenshin Project.

At this point, it was clear that our first step was to develop a new electrocardiograph device and make it available to the academic community. This newly developed electrocardiograph is based on a Front-End solution from Analog Devices. Fortunately, such integrated circuits exist and are designed to comply with all the necessary standards for developing a diagnostic-grade electrocardiograph. A design description is published in the article “A Collaborative Platform for Advancing Automatic Interpretation in ECG signals”. The design documentation is available for download in a .zip file, which contains the schematic design, a bill of materials, the Gerber files for PCB fabrication, and the assembly drawings to facilitate circuit exploration and debugging.

The enclosure of the electrocardiograph was 3D printed using high-definition resin, and the step files are also included in the .zip file. It is a good looking device, and some pictures are displayed in the following gallery. Please click on the image to enter full-screen mode, or if desired, full-resolution images can be downloaded from the link below.

The Kenshin device features a USB 2.0 Type-C interface for communication with computing devices running any operating system that supports the USB protocol. A video demonstrating the device’s structure, and another showing the device enumeration, are presented below: