Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
January 22, 2024Open Access

Haralick descriptors and Random Forest classification achieve ~79% accuracy distinguishing medicated Parkinson's patients from controls.

View Full Paper
Ask AI
Bookmark
Share

Why the study?

The study aimed to evaluate features extracted from EEG signals through HOS, Haralick descriptors, and Fractal Features as new biomarkers for Parkinson disease identification.

Does feature extraction from EEG signals using Haralick descriptors and Fractal Features accurately detect Parkinson's Disease in individuals undergoing an attentional cognitive task?

Population

50 individuals from the Open Neuro repository who underwent an attentional cognitive task

Comparison

Parkinsons subjects vs control subjects

Key result

Haralick descriptors combined with a Random Forest classifier achieved an accuracy of 79.49% in differentiating between Parkinson's patients on medication and healthy control subjects.

Authors

CSCarolline Angela dos Santos SouzaGVGiovanni Guimarães VianaBCBruno Fonseca Oliveira Coelho

Discussion

Loading...

Member takes

Overview

EEG Haralick features may support PD biomarker exploration; leaves open prospective validation in larger cohorts.

Structured PICO

Does feature extraction from EEG signals using Haralick descriptors and Fractal Features accurately detect Parkinson's Disease in individuals undergoing an attentional cognitive task?

P
Population
50 individuals (25 with Parkinson's Disease, 25 control subjects) from the Open Neuro repository who underwent an attentional cognitive task (Oddball paradigm).
I
Intervention
Feature extraction from EEG signals using Higher-Order Spectra (HOS), Haralick descriptors, and Fractal Features, classified using Random Forest (RF) and Support Vector Machine (SVM) algorithms.
C
Comparator
Control subjects without Parkinson's disease, and Parkinson's subjects off dopaminergic medication.
O
Outcome
Accuracy in differentiating between Parkinson's subjects and control subjects.surrogate

Main Result

p-value: p=≤0.01

Textural feature extraction from EEG signals using Haralick descriptors and Random Forest classification can differentiate Parkinson's patients on dopaminergic medication from healthy controls with 79.49% accuracy.

Limitations

  • The results diverge from those usually found in papers using EEG signals for the diagnosis of Parkinson's disease.

Cite This Study

Souza et al. (2024) studied Parkinson's Disease (n=50). Haralick descriptors and Random Forest (RF) classifier vs. Support Vector Machine (SVM) and Fractal techniques was evaluated on Classification accuracy (ON medication vs Control) (p=≤0.01). Haralick descriptors combined with a Random Forest classifier achieved an accuracy of 79.49% in differentiating between Parkinson's patients on medication and healthy control subjects.

synapsesocial.com/papers/6a1c5efb4ebd09f3dfa9c24bhttps://doi.org/10.21528/cbic2023-027
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Parkinsons disease detection based on image analysis of EEG signals2024 · 1 citations
  2. 2Deep Learning Techniques for Parkinson's Detection Using EEG Signals Analysis2023 · 5 citations
  3. 3Fractal dimensions and machine learning for detection of Parkinson’s disease in resting-state electroencephalography2024 · 38 citations
  4. 4Comprehensive EEG Signal Feature Extraction for Neurological Disorder Diagnosis: Focus on Alzheimer's, Parkinson's, and Seizure Disorders2024
  5. 5Parkinsons disease diagnosis through electroencephalographic signal processing and neural network classification2024