Table 4 Observed classification matrices resulting from our SS-SPDF method/algorithm [K-TOPS approach —that is, applying the sequence of first, K-means (for sorting the single-unit spikes), and then, template optimization in the phase space (for sorting the overlapping waveforms)] on five simulated datasets (from D_1 to D_5).

From: Spike sorting based on shape, phase, and distribution features, and K-TOPS clustering with validity and error indices

Dataset

Classification Matrix

*Unclassified

*Misclassified

*Well-classified

Error Index

D_1

\(\begin{array}{c}\,\,T1\,\,T2\,\,T3\\ \begin{array}{c}{C}_{1}\\ {C}_{2}\\ {C}_{3}\end{array}(\begin{array}{ccc}2730 & 5 & 1\\ 6 & 2685 & 1\\ 0 & 0 & 2730\end{array})\end{array}\)

23 (13)

13 (4)

8145 (64)

26.665

D_2

\(\begin{array}{c}\,\,T1\,\,T2\,\,T3\\ \begin{array}{c}{C}_{1}\\ {C}_{2}\\ {C}_{3}\end{array}(\begin{array}{ccc}2728 & 4 & 1\\ 5 & 2695 & 1\\ 0 & 0 & 2735\end{array})\end{array}\)

12 (9)

11 (1)

8158 (71)

17.436

D_3

\(\begin{array}{c}\,\,T1\,\,T2\,\,T3\\ \begin{array}{c}{C}_{1}\\ {C}_{2}\\ {C}_{3}\end{array}(\begin{array}{ccc}2731 & 6 & 1\\ 5 & 2691 & 1\\ 0 & 0 & 2733\end{array})\end{array}\)

13 (9)

13 (2)

8155 (70)

20.518

D_4

\(\begin{array}{c}\,\,T1\,\,T2\,\,T3\\ \begin{array}{c}{C}_{1}\\ {C}_{2}\\ {C}_{3}\end{array}(\begin{array}{ccc}2726 & 5 & 1\\ 4 & 2687 & 1\\ 0 & 0 & 2732\end{array})\end{array}\)

25 (18)

11 (1)

8145 (62)

25.357

D_5

\(\begin{array}{c}\,\,T1\,\,T2\,\,T3\\ \begin{array}{c}{C}_{1}\\ {C}_{2}\\ {C}_{3}\end{array}(\begin{array}{ccc}2725 & 10 & 1\\ 5 & 2692 & 1\\ 0 & 0 & 2735\end{array})\end{array}\)

12 (11)

17 (2)

8152 (68)

23.195

Mean

\(\begin{array}{c}\,\,T1\,\,T2\,\,T3\\ \begin{array}{c}{C}_{1}\\ {C}_{2}\\ {C}_{3}\end{array}(\begin{array}{ccc}2728 & 6 & 1\\ 5 & 2690 & 1\\ 0 & 0 & 2733\end{array})\end{array}\)

17 (12)

Percentage (%)

0.21% (14.81%)

13 (2)

Percentage (%)

0.16% (2.47%)

8151 (67)

Percentage (%)

99.63% (82.72%)

22.634

SEM

3.724

  1. The first number in unclassified, misclassified or well-classified columns is the number of spike events. The number in parentheses represents the number of overlapping waveforms for each category. In the last row, the corresponding percentage values (%) are indicated. In the last column, the resulting Error Index [see Eq. (9)] for each simulated dataset and the average value of them are reported. *Here are indicated, well-classified (\({w}_{S}=\sum {d}_{i}\)), misclassified (\({m}_{S}=\sum {r}_{k}\)) and unclassified [\({u}_{S}={N}_{S}-(\sum {d}_{i}+\sum {r}_{k}\))] spike events. Also, NS is the total number of spike events (8181, among which 8100 are single-unit spikes and 81 are overlapping waveforms), while di and rk are the diagonal and nondiagonal elements of the observed classification matrix, respectively. Abbreviations: SEM, standard error of the mean.
Back to article page