Limits...
Automatic online spike sorting with singular value decomposition and fuzzy C-mean clustering.

Oliynyk A, Bonifazzi C, Montani F, Fadiga L - BMC Neurosci (2012)

Bottom Line: The results of these tests showed an excellent accuracy in discriminating low-amplitude and overlapping spikes under strong background noise.This new software provides neuroscience laboratories with a new tool for fast and robust online classification of single neuron activity.This feature could become crucial in situations when online spike detection from multiple electrodes is paramount, such as in human clinical recordings or in brain-computer interfaces.

View Article: PubMed Central - HTML - PubMed

Affiliation: Section of Human Physiology, Department of Biomedical Sciences and Advanced Therapies, Faculty of Medicine, University of Ferrara, Via Fossato di Mortara 17/19, 44121, Ferrara, Italy. lynnry@unife.it

ABSTRACT

Background: Understanding how neurons contribute to perception, motor functions and cognition requires the reliable detection of spiking activity of individual neurons during a number of different experimental conditions. An important problem in computational neuroscience is thus to develop algorithms to automatically detect and sort the spiking activity of individual neurons from extracellular recordings. While many algorithms for spike sorting exist, the problem of accurate and fast online sorting still remains a challenging issue.

Results: Here we present a novel software tool, called FSPS (Fuzzy SPike Sorting), which is designed to optimize: (i) fast and accurate detection, (ii) offline sorting and (iii) online classification of neuronal spikes with very limited or human intervention. The method is based on a combination of Singular Value Decomposition for fast and highly accurate pre-processing of spike shapes, unsupervised Fuzzy C-mean, high-resolution alignment of extracted spike waveforms, optimal selection of the number of features to retain, automatic identification the number of clusters, and quantitative quality assessment of resulting clusters independent on their size. After being trained on a short testing data stream, the method can reliably perform supervised online classification and monitoring of single neuron activity. The generalized procedure has been implemented in our FSPS spike sorting software (available free for non-commercial academic applications at the address: http://www.spikesorting.com) using LabVIEW (National Instruments, USA). We evaluated the performance of our algorithm both on benchmark simulated datasets with different levels of background noise and on real extracellular recordings from premotor cortex of Macaque monkeys. The results of these tests showed an excellent accuracy in discriminating low-amplitude and overlapping spikes under strong background noise. The performance of our method is competitive with respect to other robust spike sorting algorithms.

Conclusions: This new software provides neuroscience laboratories with a new tool for fast and robust online classification of single neuron activity. This feature could become crucial in situations when online spike detection from multiple electrodes is paramount, such as in human clinical recordings or in brain-computer interfaces.

Show MeSH

Related in: MedlinePlus

Result of classification of the simulated dataset from Example 2, noise level 0.15.A - The projection of the first three PCs of the dataset not containing overlapping spikes; B - the same, but with overlapped spikes. Three dense clusters are shown in both cases. C and D - original spike shapes of detected clusters in the datasets without and with overlapping spikes, respectively. Different spike events are shown in different colours, according to the outcome of the clustering algorithm. The number of correctly identified spikes is indicated. Values in brackets indicate the total number of spikes in each class. The spike shape shown in the lower part of the figure has been incorrectly classified and therefore marked in black.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3473300&req=5

Figure 8: Result of classification of the simulated dataset from Example 2, noise level 0.15.A - The projection of the first three PCs of the dataset not containing overlapping spikes; B - the same, but with overlapped spikes. Three dense clusters are shown in both cases. C and D - original spike shapes of detected clusters in the datasets without and with overlapping spikes, respectively. Different spike events are shown in different colours, according to the outcome of the clustering algorithm. The number of correctly identified spikes is indicated. Values in brackets indicate the total number of spikes in each class. The spike shape shown in the lower part of the figure has been incorrectly classified and therefore marked in black.

Mentions: Table 1 shows the number of classification errors of the FSPS algorithm and the other tested algorithms when detecting and sorting noisy non-overlapping spikes. FSPS gave the lowest number of false matching spikes in most simulated datasets and did not exceed 2% up to noise level 0.2 in all examples with exception of Example 4, where 7.2% of mismatches were detected. However, even in this case the outcome of FSPS technique was still in 4,5-8,8 times better compared to other methods. The advantage of FSPS becomes apparent when spike shapes are more similar (Table 1, Examples 3 and 4, considered more difficult for clustering), while our results were competitive with those obtained using K-means or SPS clustering on wavelets in Examples 1 and 2, where spike shapes of three simulated neurons were markedly different. A nice feature of the performance of our FSPS algorithm was that it degraded gracefully with increasing noise, in part due to the better outlier identification of fuzzy clustering, and the performance was reasonably good also in the case of overlapping spikes (Table 2). The reason for this improved performance is probably due to better pre-processing strategy that we employed rather than the different clustering procedure. In particular, we verified the alignment procedure and the implementation of PSVD on the clustering performance. Figure 8 illustrates this point by depicting results of classification after clustering of simulated Example 2 with noise level 0.15, a dataset that was particularly difficult to cluster with the traditional 3 PCs clustering method[10]. With our procedure, the distribution of first three PCs at the fragments A and B for the datasets without (Figure 8A) and with (Figure 8B) overlapping spikes demonstrates three clean, compact and well distinguished clusters. The presence of overlapping spikes in the dataset B (763 out of 3411, that is 22,4%) creates less distant and more shaped clusters having complex outliers.


Automatic online spike sorting with singular value decomposition and fuzzy C-mean clustering.

Oliynyk A, Bonifazzi C, Montani F, Fadiga L - BMC Neurosci (2012)

Result of classification of the simulated dataset from Example 2, noise level 0.15.A - The projection of the first three PCs of the dataset not containing overlapping spikes; B - the same, but with overlapped spikes. Three dense clusters are shown in both cases. C and D - original spike shapes of detected clusters in the datasets without and with overlapping spikes, respectively. Different spike events are shown in different colours, according to the outcome of the clustering algorithm. The number of correctly identified spikes is indicated. Values in brackets indicate the total number of spikes in each class. The spike shape shown in the lower part of the figure has been incorrectly classified and therefore marked in black.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3473300&req=5

Figure 8: Result of classification of the simulated dataset from Example 2, noise level 0.15.A - The projection of the first three PCs of the dataset not containing overlapping spikes; B - the same, but with overlapped spikes. Three dense clusters are shown in both cases. C and D - original spike shapes of detected clusters in the datasets without and with overlapping spikes, respectively. Different spike events are shown in different colours, according to the outcome of the clustering algorithm. The number of correctly identified spikes is indicated. Values in brackets indicate the total number of spikes in each class. The spike shape shown in the lower part of the figure has been incorrectly classified and therefore marked in black.
Mentions: Table 1 shows the number of classification errors of the FSPS algorithm and the other tested algorithms when detecting and sorting noisy non-overlapping spikes. FSPS gave the lowest number of false matching spikes in most simulated datasets and did not exceed 2% up to noise level 0.2 in all examples with exception of Example 4, where 7.2% of mismatches were detected. However, even in this case the outcome of FSPS technique was still in 4,5-8,8 times better compared to other methods. The advantage of FSPS becomes apparent when spike shapes are more similar (Table 1, Examples 3 and 4, considered more difficult for clustering), while our results were competitive with those obtained using K-means or SPS clustering on wavelets in Examples 1 and 2, where spike shapes of three simulated neurons were markedly different. A nice feature of the performance of our FSPS algorithm was that it degraded gracefully with increasing noise, in part due to the better outlier identification of fuzzy clustering, and the performance was reasonably good also in the case of overlapping spikes (Table 2). The reason for this improved performance is probably due to better pre-processing strategy that we employed rather than the different clustering procedure. In particular, we verified the alignment procedure and the implementation of PSVD on the clustering performance. Figure 8 illustrates this point by depicting results of classification after clustering of simulated Example 2 with noise level 0.15, a dataset that was particularly difficult to cluster with the traditional 3 PCs clustering method[10]. With our procedure, the distribution of first three PCs at the fragments A and B for the datasets without (Figure 8A) and with (Figure 8B) overlapping spikes demonstrates three clean, compact and well distinguished clusters. The presence of overlapping spikes in the dataset B (763 out of 3411, that is 22,4%) creates less distant and more shaped clusters having complex outliers.

Bottom Line: The results of these tests showed an excellent accuracy in discriminating low-amplitude and overlapping spikes under strong background noise.This new software provides neuroscience laboratories with a new tool for fast and robust online classification of single neuron activity.This feature could become crucial in situations when online spike detection from multiple electrodes is paramount, such as in human clinical recordings or in brain-computer interfaces.

View Article: PubMed Central - HTML - PubMed

Affiliation: Section of Human Physiology, Department of Biomedical Sciences and Advanced Therapies, Faculty of Medicine, University of Ferrara, Via Fossato di Mortara 17/19, 44121, Ferrara, Italy. lynnry@unife.it

ABSTRACT

Background: Understanding how neurons contribute to perception, motor functions and cognition requires the reliable detection of spiking activity of individual neurons during a number of different experimental conditions. An important problem in computational neuroscience is thus to develop algorithms to automatically detect and sort the spiking activity of individual neurons from extracellular recordings. While many algorithms for spike sorting exist, the problem of accurate and fast online sorting still remains a challenging issue.

Results: Here we present a novel software tool, called FSPS (Fuzzy SPike Sorting), which is designed to optimize: (i) fast and accurate detection, (ii) offline sorting and (iii) online classification of neuronal spikes with very limited or human intervention. The method is based on a combination of Singular Value Decomposition for fast and highly accurate pre-processing of spike shapes, unsupervised Fuzzy C-mean, high-resolution alignment of extracted spike waveforms, optimal selection of the number of features to retain, automatic identification the number of clusters, and quantitative quality assessment of resulting clusters independent on their size. After being trained on a short testing data stream, the method can reliably perform supervised online classification and monitoring of single neuron activity. The generalized procedure has been implemented in our FSPS spike sorting software (available free for non-commercial academic applications at the address: http://www.spikesorting.com) using LabVIEW (National Instruments, USA). We evaluated the performance of our algorithm both on benchmark simulated datasets with different levels of background noise and on real extracellular recordings from premotor cortex of Macaque monkeys. The results of these tests showed an excellent accuracy in discriminating low-amplitude and overlapping spikes under strong background noise. The performance of our method is competitive with respect to other robust spike sorting algorithms.

Conclusions: This new software provides neuroscience laboratories with a new tool for fast and robust online classification of single neuron activity. This feature could become crucial in situations when online spike detection from multiple electrodes is paramount, such as in human clinical recordings or in brain-computer interfaces.

Show MeSH
Related in: MedlinePlus