P07-Automatic pallidal neurons recognition based on the detection of the number of clusters from microrecordings in dystonia. Issue 4 (April 2018)

• Record Type: Journal Article
• Authors: Klempíř, O.; Krupička, R.; Sieger, T.; Jech, R.
• Is Part Of: Clinical neurophysiology. Volume 129:Issue 4(2018:Apr.)
• Page Start: e15
• Page End: e16
• Publication Date: 2018-04
• Subjects: Neurophysiology -- Periodicals; Electroencephalography -- Periodicals; Electromyography -- Periodicals; Neurology -- Periodicals; 612.8
• DOI: 10.1016/j.clinph.2018.01.052
• Languages: English
• ISSNs: 1388-2457

Abstract

Abstract : Intraoperative microelectrode records (MER) are considered as the standard electrophysiological method for the precise positioning of the deep brain stimulation (DBS) electrode into the Globus pallidus interna (GPi). The final GPi position is chosen based on the firing patterns of individual neurons. Finding the number of neurons is usually done manually during the spike sorting. We propose methodology for neurons recognition based on the unsupervised learning. Thirty MERs (24 kHz 10s) of the basal ganglia from 10 patients (43.3(±14), 5F) with dystonia were recorded during DBS implantation. MERs were filtered in the 300–3000 Hz band and the amplitude detector (3× std of the background noise) was used to detect spikes. The WaveClus features were computed and its 2 PCA components were extracted for every spike. The optimal number of clusters evaluated by an expert rater, K-means + gap criterion (alg. 1) and the GMM + BIC (alg. 2) were analyzed. The total Intraclass correlation showed a significant inter-rater agreement for all 3 rater procedures (ICC = 0.62, p < 0.001), when post hoc weighted Cohen's Kappas were 0.79 (expert vs. alg. 1; p < 0.001) and 0.53 (expert vs. alg. 2; p < 0.001). It can contribute to better description of type of the GPi neurons involved in (non)motor functions. Supported by the GA Č R No. 16-13323S and AZV No. 16-28119a.