SpikeDeep-classifier: a deep-learning based fully automatic offline spike sorting algorithm

摘要

Objective. Advancements in electrode design have resulted in micro-electrode arrays with hundredsof channels for single cell recordings. In the resulting electrophysiological recordings, eachimplanted electrode can record spike activity (SA) of one or more neurons along with backgroundactivity (BA). The aim of this study is to isolate SA of each neural source. This process is calledspike sorting or spike classification. Advanced spike sorting algorithms are time consuming becauseof the human intervention at various stages of the pipeline. Current approaches lack generalizationbecause the values of hyperparameters are not fixed, even for multiple recording sessions of thesame subject. In this study, a fully automatic spike sorting algorithm called ‘SpikeDeep-Classifier’ isproposed. The values of hyperparameters remain fixed for all the evaluation data. Approach. Theproposed approach is based on our previous study (SpikeDeeptector) and a novel backgroundactivity rejector (BAR), which are both supervised learning algorithms and an unsupervisedlearning algorithm (K-means). SpikeDeeptector and BAR are used to extract meaningful channelsand remove BA from the extracted meaningful channels, respectively. The process of clusteringbecomes straight-forward once the BA is completely removed from the data. Then, K-means with apredefined maximum number of clusters is applied on the remaining data originating from neuralsources only. Lastly, a similarity-based criterion and a threshold are used to keep distinct clustersand merge similar looking clusters. The proposed approach is called cluster accept or merge(CAOM) and it has only two hyperparameters (maximum number of clusters and similaritythreshold) which are kept fixed for all the evaluation data after tuning. Main results. We comparedthe results of our algorithm with ground-truth labels. The algorithm is evaluated on data of humanpatients and publicly available labeled non-human primates (NHPs) datasets. The average accuracyof BAR on datasets of human patients is 92.3% which is further reduced to 88.03% after(K-means + CAOM). In addition, the average accuracy of BAR on a publicly available labeleddataset of NHPs is 95.40% which reduces to 86.95% after (K-mean + CAOM). Lastly, wecompared the performance of the SpikeDeep-Classifier with two human experts, whereSpikeDeep-Classifier has produced comparable results. Significance. The SpikeDeep-Classifier isevaluated on the datasets of multiple recording sessions of different species, different brain areas and different electrode types without further retraining. The results demonstrate that‘SpikeDeep-Classifier’ possesses the ability to generalize well on a versatile dataset and henceforthprovides a generalized and fully automated solution to offline spike sorting.