Fractal Dimension as a discriminative feature for high accuracy classification in motor imagery EEG-based brain-computer interface

Moaveninejad, Sadaf; D'Onofrio, Valentina; Tecchio, Franca; Ferracuti, Francesco; Iarlori, Sabrina; Monteriù, Andrea; Porcaro, Camillo

doi:10.1016/j.cmpb.2023.107944

Cited by 8 publications

(2 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This “hidden information” captured by non-linear methods such as fractal dimension analysis may be additional and complementary to linear methods and could shed light on the physiological neural communication, computation, and cognition in healthy as well as patients with neuropathological conditions ( Goldberger, 2001 ; Goldberger et al, 2002 ; Zhang and Raichle, 2010 ; Rodríguez-Bermúdez and García-Laencina, 2015 ; Porcaro et al, 2017 , 2019 , 2020a , b , 2022 ). This is the reason why time-series fractal analysis is more and more used in different research fields ranging from basic neuroscience ( Di Ieva et al, 2014 , 2015 ; Moaveninejad et al, 2024 ), neurophysiology ( Adeli et al, 2008 ; Ahmadlou and Adeli, 2012 ; Ahmadlou et al, 2012a , b ), translational neuroscience ( Smits et al, 2016 ; Porcaro et al, 2020b , 2021 , 2022 ; Fiorenzato et al, 2024 ; Olejarczyk et al, 2024 ) to genetic variability in human phenotypes ( Cattani and Pierro, 2013 ; Lee, 2020 ; Borri et al, 2022 ).…”

Section: Introductionmentioning

confidence: 99%

Characterization of antiseizure medications effects on the EEG neurodynamic by fractal dimension

Porcaro,

Seppi,

Pellegrino

et al. 2024

Front. Neurosci.

Self Cite

View full text Add to dashboard Cite

ObjectivesAn important challenge in epilepsy is to define biomarkers of response to treatment. Many electroencephalography (EEG) methods and indices have been developed mainly using linear methods, e.g., spectral power and individual alpha frequency peak (IAF). However, brain activity is complex and non-linear, hence there is a need to explore EEG neurodynamics using nonlinear approaches. Here, we use the Fractal Dimension (FD), a measure of whole brain signal complexity, to measure the response to anti-seizure therapy in patients with Focal Epilepsy (FE) and compare it with linear methods.MaterialsTwenty-five drug-responder (DR) patients with focal epilepsy were studied before (t1, named DR-t1) and after (t2, named DR-t2) the introduction of the anti-seizure medications (ASMs). DR-t1 and DR-t2 EEG results were compared against 40 age-matched healthy controls (HC).MethodsEEG data were investigated from two different angles: frequency domain—spectral properties in δ, θ, α, β, and γ bands and the IAF peak, and time-domain—FD as a signature of the nonlinear complexity of the EEG signals. Those features were compared among the three groups.ResultsThe δ power differed between DR patients pre and post-ASM and HC (DR-t1 vs. HC, p < 0.01 and DR-t2 vs. HC, p < 0.01). The θ power differed between DR-t1 and DR-t2 (p = 0.015) and between DR-t1 and HC (p = 0.01). The α power, similar to the δ, differed between DR patients pre and post-ASM and HC (DR-t1 vs. HC, p < 0.01 and DR-t2 vs. HC, p < 0.01). The IAF value was lower for DR-t1 than DR-t2 (p = 0.048) and HC (p = 0.042). The FD value was lower in DR-t1 than in DR-t2 (p = 0.015) and HC (p = 0.011). Finally, Bayes Factor analysis showed that FD was 195 times more likely to separate DR-t1 from DR-t2 than IAF and 231 times than θ.DiscussionFD measured in baseline EEG signals is a non-linear brain measure of complexity more sensitive than EEG power or IAF in detecting a response to ASMs. This likely reflects the non-oscillatory nature of neural activity, which FD better describes.ConclusionOur work suggests that FD is a promising measure to monitor the response to ASMs in FE.

show abstract

Section: Introductionmentioning

confidence: 99%

Characterization of antiseizure medications effects on the EEG neurodynamic by fractal dimension

Porcaro,

Seppi,

Pellegrino

et al. 2024

Front. Neurosci.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Its applications span across several disciplines, with notable implications in the realm of neuroscience (Di Ieva, 2016 , 2024 ), where traditional analyses cannot sufficiently describe the morphological and functional complexity of the human brain. Indeed, fractal analysis, and especially its main index, the fractal dimension (FD), has evolved from describing brain architecture focusing on healthy (Madan and Kensinger, 2016 , 2017 , 2018 ; Marzi et al, 2020 , 2021 , 2024 ; Pani et al, 2022 ) and pathological (Marzi et al, 2018 ; Pantoni et al, 2019 ) conditions, to a comprehensive tool for analyzing brain functions under physiological (Cottone et al, 2017 ; Marino et al, 2019 ; Collantoni et al, 2020 ; Porcaro et al, 2020b , 2024 ), pathological conditions (Di Ieva et al, 2014 ; Smits et al, 2016 ; Porcaro et al, 2019 , 2020a , 2021 , 2022a , b , c ; Varley et al, 2020 ; Fiorenzato et al, 2024 )—and see Meregalli et al ( 2022 ) and Díaz Beltrán et al ( 2024 ) for in-depth reviews—and brain computer interfaces (BCIs) (Moaveninejad et al, 2024 ).…”

mentioning

confidence: 99%