2022 journal article

Investigating dielectric spectroscopy and soft sensing for nondestructive quality assessment of engineered tissues

BIOSENSORS & BIOELECTRONICS, 216.

By: S. Shohan n , Y. Zeng*, X. Chen *, R. Jin * & R. Shirwaiker n 

co-author countries: United States of America πŸ‡ΊπŸ‡Έ
author keywords: Biofabrication; Dielectric spectroscopy; Non-destructive quality monitoring; Soft sensing; Tissue engineering
MeSH headings : Biosensing Techniques; Dielectric Spectroscopy / methods; Gelatin; Humans; Methacrylates; Tissue Engineering / methods
Source: Web Of Science
Added: October 3, 2022

Non-destructive, inline quality monitoring techniques that can overcome the limitations of traditional, offline assays are essential to support the scale-up production of tissue engineered medical products (TEMP). In this work, we investigate a new soft-sensing approach with non-destructive dielectric spectroscopy (DS) that synergistically utilizes inline sensor data and predictive analytics to estimate unmeasured TEMP quality profiles. First, the performance of DS during the assessment of gelatin methacrylate (GelMA) constructs containing human adipose-derived stem cells was investigated in comparison to a traditional biochemical assay. The effects of two critical biofabrication parameters (photocrosslinking duration and volume of growth media) on a key scalar metric (Δϡ) were determined over 11 days of in vitro culture, where the metric was associated with the permittivity response of cells to alternating electric fields during DS and corresponding cellular metabolic activity. To enable accurate quality prediction while minimizing direct data collection to reduce the risk of cytotoxicity from prolonged exposure to the DS sensor electrodes and electric fields, we then developed a bilinear basis mixed model (BBMM) as a soft sensor. With comprehensive consideration of different variation sources, this model was designed to estimate missing permittivity profiles of constructs based on the measured DS dataset and biofabrication parameters. Results of benchmarking showed that BBMM outperformed state-of-the-art vector-prediction methods from literature in two different missing data estimation mechanisms. The high-accuracy BBMM provides a novel DS-driven soft sensing system as an inline monitoring tool suitable for scaled-up or scaled-out TEMP production systems.