Repository logo

ISEL - Instituto Superior de Engenharia de Lisboa

Permanent URI for this community

http://hdl.handle.net/10400.21/8

Browse

Browsing ISEL - Instituto Superior de Engenharia de Lisboa by Author "Abalde, Sara Fernández"

Now showing 1 - 1 of 1
  • Development of an analysis pipeline for human microelectrode recordings in Parkinson’s disease
    Publication . Abalde, Sara Fernández; Marques, Gonçalo Caetano; Mendonça, Marcelo
    Background: Deep brain stimulation is a common treatment for advanced Parkinson’s Disease (PD). Intraoperative microelectrode recordings (MER) along preplanned trajectories are often used for accurate identification of subthalamic nucleus (STN), a common target for deep brain stimulation (DBS) in PD. However, this identification is performed manually and can be difficult in regions of transition. Misidentification may lead to suboptimal location of the DBS lead and inadequate clinical outcomes. Methods: A tool for unsupervised analysis and spike-sorting of human MER signals with feature extraction was developed. We also trained and tested a hybrid unsupervised/supervised machine learning approach that uses extracted MER time, frequency and noise properties for high-accuracy identification of STN. Lastly, we compared neurophysiological characteristics of different STN functional segments. Results: We obtained a classification accuracy of 96:28 3:15 % (30 trajectories, 5 patients) for individual STN-DBS surgery MER using an approach of "leave one subject out" validation with support vector machine classifier, all features based on time and frequency domain and human expert labels. The unsupervised sorting approach allowed us to sort a total of 357 STN neurons in 5 subjects. Dividing the STN in a dorsal, probably motor region, and a ventral, probably non-motor portion, we’ve found a higher burst rate (median (interquartile range) of 1.8 (1.5) vs 1.15 (0.05) bursts/s, p=0.001) and firing rate (median (interquartile range) of 21.4 (16.85) vs. 15.3 (14.33), p=0.013) of dorsal STN neurons among other features. Ongoing work will refine these results using anatomical gold standard through lead trajectory reconstruction, fused with an STN functional subdivision atlas. Conclusions: We’ve developed a tool for human MER analysis and extraction of related features, that provided good preliminary results in STN classification. In line with the literature, we were able to find preliminary activity differences in functionally segregated STN segments. This tool is fast and generalizable for other brain regions. Ongoing work using patient’s anatomy can further validate its’ usefulness in optimizing electrode placement and research purposes.
    2018-12-07Master thesis Open access Show more