Drift removal in plant electrical signals via IIR filtering using wavelet energy

Saptarshi Das, Barry Juans Ajiwibawa, Shre Kumar Chatterjee, Sanmitra Ghosh, Koushik Maharatna, Srinandan Dasmahapatra, Andrea Vitaletti, Elisa Masi, Stefano Mancuso

Introduction

Plant electrophysiology is a burgeoning field that seeks to understand how plants communicate and respond to their environment through electrical signals. Unlike human bio-signals such as ECGs or EEGs, which benefit from decades of standardized processing techniques, plant signals are notoriously difficult to analyze. They are characterized by extremely low amplitudes and are frequently obscured by "drift"—low-frequency fluctuations caused by environmental factors like humidity, leaf movement, or hardware-specific gain.

This research introduces a standardized pre-processing framework designed to "clean" these signals. By utilizing an optimized digital filter tuned through wavelet energy criteria, the researchers have developed a method to distinguish a plant's genuine biological response from background noise and unpredictable drift.

Methodology: The Wavelet Energy Criterion

The core of this approach is the optimization of digital Infinite Impulse Response (IIR) filters. The study evaluated four traditional filter types: Butterworth, Chebyshev Type I, Chebyshev Type II, and Elliptic. To determine the optimal parameters for these filters, the authors employed a Wavelet Energy Criterion.

This process involves three primary steps:

  1. Decomposition: A Wavelet Packet Transform (WPT) is used to break the raw plant signal into various frequency "nodes."
  2. Feature Space Mapping: The energy at each node is calculated, creating a multi-dimensional feature space that characterizes the signal's properties.
  3. Optimization: Using the Nelder-Mead Simplex algorithm, the researchers tune filter parameters (such as cut-off frequency and ripple) to minimize the variance between "pre-stimulus" baseline signals across different experiments.

By standardizing the background noise, any subsequent deviation following a stimulus can be confidently identified as a biological reaction rather than an artifact of the recording environment.

Experimental Setup and Datasets

The study utilized a custom dataset derived from 11 tomato plants (Solanum lycopersicum). To ensure high-quality data, recordings were conducted in a controlled environment designed to minimize external interference.

Experimental setup featuring a Faraday cage, LED lighting, and electrode placement in the tomato plant stem.

As shown in the setup, plants were housed within a Faraday cage to block electromagnetic interference. Signals were captured at a 10 Hz sampling rate using stainless steel needle electrodes inserted into the upper and middle sections of the plant stem. The plants were subjected to three distinct chemical stressors: Ozone ($O_3$), Sulphuric acid ($H_2SO_4$), and Sodium Chloride ($NaCl$).

Key Results and Insights

The optimization process revealed that the Chebyshev Type-II filter was the most effective at preserving signal integrity while minimizing background discrimination. The optimal cut-off frequency ($ω_c$) for this filter was determined to be approximately 0.77 Hz.

Electrical signal responses of tomato plants to various chemical stressors including Ozone, H2SO4, and NaCl.

Key findings from the analysis include:

  • Baseline Standardization: The filter successfully forced the pre-stimulus energy distributions of disparate experiments to overlap, effectively "detrending" the data and removing initial bias.
  • Superior Performance: The Chebyshev Type-II filter achieved a significantly lower cost function ($J_{min} = 11.64$) compared to the standard Butterworth filter ($J_{min} = 17.48$).
  • Visualizing Response: Once the optimal filter is applied, the energy shift in the post-stimulus signal becomes clearly visible, providing a robust statistical indicator of the plant's reaction to the stressor.

Conclusion and Future Work

This study provides a critical "common platform" for plant signal analysis, moving the field toward the level of standardization seen in human medical diagnostics. While the results are promising for tomato plants, future research is needed to determine if the 0.77 Hz cutoff is universal across different species with varying vascular structures. Additionally, the researchers aim to move from signal cleaning to automated classification, potentially allowing computers to identify specific environmental stressors based solely on the plant's electrical "signature."

Summary updated

Abstract

Plant electrical signals often contains low frequency drifts with or without the application of external stimuli. Quantification of the randomness in plant signals in a stimulus-specific way is hindered because the knowledge of vital frequency information in the actual biological response is not known yet. Here we design an optimum Infinite Impulse Response (IIR) filter which removes the low frequency drifts and preserves the frequency spectrum corresponding to the random component of the unstimulated plant signals by bringing the bias due to unknown artifacts and drifts to a minimum. We use energy criteria of wavelet packet transform (WPT) for optimization based tuning of the IIR filter parameters. Such an optimum filter enforces that the energy distribution of the pre-stimulus parts in different experiments are almost overlapped but under different stimuli the distributions of the energy get changed. The reported research may popularize plant signal processing, as a separate field, besides other conventional bioelectrical signal processing paradigms.
BibTeX
@article{das2015drift,
  author = {Saptarshi Das and Barry Juans Ajiwibawa and Shre Kumar Chatterjee and Sanmitra Ghosh and Koushik Maharatna and Srinandan Dasmahapatra and Andrea Vitaletti and Elisa Masi and Stefano Mancuso},
  title = {{Drift removal in plant electrical signals via IIR filtering using wavelet energy}},
  journal = {Comput. Electron. Agric.},
  year = {2015},
  doi = {10.1016/j.compag.2015.08.013},
}
Figure