# Assessing the Carasau Bread Doughs Microwave Spectra

## TL;DR

This study evaluates microwave dielectric spectroscopy for monitoring Carasau bread dough quality, finding variability in measurements that could impact industrial use.

## Contribution

The study provides a statistical evaluation of microwave dielectric spectroscopy repeatability and reliability in Carasau bread doughs.

## Key findings

- Microwave dielectric spectroscopy measurements showed high correlation but heterogeneous data distribution.
- Variability in dielectric properties was more pronounced after leavening.
- Environmental factors like pressure, temperature, and humidity did not significantly affect measurements.

## Abstract

Carasau bread (CB) is a traditional Sardinian flatbread with significant market potential, driving the need for advanced quality monitoring solutions in its production. Recent advancements in automation and engineering have enhanced process control, but a comprehensive understanding of CB dough properties remains essential. Dielectric spectroscopy (DS), particularly in the microwave (MW) range, has emerged as a non-destructive, cost-effective tool for food characterization, providing insights into microstructure and composition. MW DS has been applied to assess fermentation dynamics and ingredient influence in CB doughs, with previous studies modeling dielectric properties using a third-order Cole–Cole model up to 8.5 GHz and later extending to 20 GHz. Despite these advancements, the repeatability, reliability, and consistency of MW DS measurements on CB doughs have not been systematically assessed. This study aims to fill this gap by analyzing MW DS measurements on ten CB dough samples with standard composition (water 50%, yeast 1.5%, salt 1.5%) in the 0.5–6 GHz range, both before and after leavening, for 10 different samples and a total of 100 measurements. Even though the correlation between spectra is high, and even if the coefficient of variation is below 5% before leavening, the z-score analysis and the kernel density estimation highlighted that the distribution of dielectric data is heterogeneous, showing that variability across samples exists, especially after leavening. Finally, the influence of pressure, temperature, and relative humidity was excluded. This statistical evaluation of MW DS measurement provided critical insights into the robustness of MW DS for industrial applications.