Smart Stent to warn of restenosis

Restenosis is the re-narrowing of blood vessels that can occur following surgery to insert a stent to relieve conditions such as atherosclerosis, which itself is thickening or hardening of the arteries.

Methods to diagnose this condition are invasive, have limited clinical outcomes and are done only when the patient is already feeling symptoms.

Now, a team from Northumbria University, Bristol University and their partners are embarking on an EPSRC-funded project to develop a Smart Stent that can detect and inform the patient and medical practitioner about stent thrombosis at the stent site, which can develop within 24 hours of implant or after months or several years.

“The proposed Smart Stent differentiates the gradual reduction in blood flow from the more frequent variations due to regular blood-pressure fluctuations, even at a very early stage,” said Dr Yu-Sheng Lin, an Assistant Professor in mechanical engineering at the Southern Taiwan University of Science and Technology.

Dr Kevin Chung-Che Huang, a senior research fellow at Southampton University, explained that the piezoelectric sensor is made using polyvinylidene fluoride (PVDF) and barium titanate (BaTiO₃), which are biocompatible materials with good piezoelectric properties.

The sensor, along with a miniature thin-film circuit board, is embedded inside a cylindrical flexible biocompatible polymer material and attached to one end of a bare-metal stent made of cobalt-chromium alloy, added Dr Daniel Ho, Associate Professor of nanophotonics and electrical & electronic engineering at Northumbria University.

The circuit board consists of an amplifier, filter, analogue-to-digital converter, ultra-low power RF transmitter (with a frequency band of 402MHz to 405 MHz), and a power section. The sensor and RF circuitry are ultra-low powered, consuming power in the range of 300μW to 500μW (0.3 to 0.5mW). The RF transmitter is set to send data at intervals of one second, to save power.

The circuit can be powered either from the emf generated from the piezo sensor itself, which the team said needs further research, or wirelessly from an external portable device that should be ‘worn’ by the patient, which receives the RF signal from the Smart Stent, processes it and sends to a cloud service through Wi-Fi.

Google TensorFlow Lite machine-learning model, running in a Coral Edge Tensor-Processing-Unit (TPU) based Application Specific Integrated Circuit (ASIC) is used in the external device to analyse the pulse waveform coming from the Smart Stent and sends it to a secure cloud service to store and transmit to other authorised devices like smartphones, when any abnormalities are sensed.

Ensuring data security, all information collected is encrypted locally before being processed by AI at the edge, which handles essential computations.

“For more complex processing, the system leverages cloud-based computing resources,” said Dr. Shuangyi Yan, senior lecturer in high performance networking & optical Networking at Bristol University. “It employs advanced, privacy-focused machine learning algorithms, applicable during both the training and inference stages. This ensures the user's privacy is maintained throughout the process.”

The Smart Stent project runs for two years and includes Tampere University (Finland), Vietnam Academy of Science & Technology, Southern Taiwan University of Science and Technology, and YMECO as partners.