Compression-Based Complexity Analysis of Thalamic EEG Using Multiscale Preprocessing Techniques

Mohan, Poornima and Balasubramanian, Karthi and Nagaraj, Nithin and Prabhu, Gayathri R and Pati, Sandipan (2025) Compression-Based Complexity Analysis of Thalamic EEG Using Multiscale Preprocessing Techniques. IEEE Access, 13. pp. 209602-209619.

Text
Compression-Based_Complexity_Analysis_of_Thalamic_EEG_Using_Multiscale_Preprocessing_Techniques.pdf - Published Version
Download (1MB)

Contribution	Name	Email

Abstract:	Quantifying the complexity of biomedical signals offers critical insight into underlying physiological and pathological dynamics. This study systematically evaluates compression-based complexity measures—Effort-to-Compress (ETC) and Lempel-Ziv Complexity (LZC)—and compares them with classical entropy-based metrics, including Shannon Entropy, Approximate Entropy (ApEn), Sample Entropy (SampEn), and Permutation Entropy (PermEn). Analyses were performed on both synthetic benchmark signals and clinical thalamic EEG recordings acquired during seizure and non-seizure states. Robustness was assessed under Gaussian, Laplacian, and powerline noise, with and without preprocessing via Discrete Wavelet Transform (DWT) and Differential Pulse Code Modulation (DPCM). ETC consistently demonstrated the highest discriminative power and noise resilience, achieving large effect sizes and classification accuracies exceeding 85% when combined with full-scale DWT preprocessing. In contrast, LZC performed reliably in raw data but degraded following multiscale transformations. Entropy-based measures such as SampEn and ApEn remained competitive under clean conditions yet were more sensitive to noise and preprocessing variability. These findings establish that no single complexity metric is universally optimal; rather, performance depends on signal modality, noise structure, and preprocessing design. For thalamic EEG-based seizure detection, ETC with DWT preprocessing provides a robust, interpretable, and parameter-free framework suitable for clinical and real-time neurophysiological applications.
Item Type:	Journal Paper
Subjects:	School of Natural and Engineering Sciences > Complex Systems
Divisions:	Schools > Natural Sciences and Engineering
Date Deposited:	23 Dec 2025 09:01
Last Modified:	23 Dec 2025 09:02
Official URL:	https://ieeexplore.ieee.org/document/11282434
Related URLs:
Funders:	*
Projects:	*
DOI:	10.1109/ACCESS.2025.3641434
URI:	http://eprints.nias.res.in/id/eprint/3028

Actions (login required)

View Item