SpikeTaro outputs single unit spikes from the record consisting of multi-unit spike. In general, sorting algorithms use two parameters, that is, spike height and spike duration. However, in case of a record including huge spike units, the interactions between the spikes from different sources lead to the sorting performance depression on such process. SpikeTaro overcomes this problem by developing novel sorting algorithms as follows:

• Extract single unit spike by evaluating confidence level (Peak detection algorithm and others)
• Sorting spikes based on correlation of spike waveforms and spike heights (Time resolution independent correlation algorithm)

  The following are the papers applying this method. SpikeTaro is now widely used, and this algorithm and software are used in international conferences and draft papers.

• Kidokoro-Kobayashi M, Iwakura M, Fujiwara-Tsujii N, Fujiwara S, Sakura M, Sakamoto H, Higashi S, Hefetz A, Ozaki M: Chemical Discrimination and Aggressiveness via Cuticular Hydrocarbons in a Supercolonhy-Forming Ant, Formica yessensis, PLoS ONE, 7(10):e46840(2012)
• Perez Goodwyn P, Katsumata-Wada A, Okada K: Morphology and neurophysiology of tarsal vibration receptors in the water strider Aquarius paludum (Heteroptera: Gerridae), J. Insect Physiol. 55(2009), pp. 855-861

  SpikeTaro comes with FIR filter (Finite Impulse Response filter). This filter is a kind of digital filter, showing liner phase property. In the "FIR Filter Setting Dialog" of SpikeTaro, user can use bandpass or highpass filters and can also set cutoff frequency(s) freely.

  Below is an example of bandpass (500 - 1500 Hz) filter setting and the amplitude characteristics in SpikeTaro.

  With these filtering parameters, the filtering process shows characteristics as follows:

• Signal amplitude depresses 1/10000 within spectrum of 0 - 500 Hz.
• Signal amplitude is the same of a input data within 500 - 1500 Hz.
• Signal amplitude depresses 1/1000 to 1/10000 within 1500 Hz over.
• Amplitude gain sharply changes.

  Also below is an example of filtering process in SpikeTaro. In this case, the filtering result shows that the noise of power supply frequency are removed.

  SpikeTaro's sorting algorithm is composed of two processes below.

• Spike extraction.
• Sorting by correlation coefficients (contribution ratio) between spike waveforms.

  Using peak detection algorithm, SpikeTaro detects first point of the spike of rising phase, peak point and termination point of the spike, automatically. SpikeTaro can extract single unit spike from an integrated signal of several spikes. In case of low reliability, that is, the extracted spike is too short, SpikeTaro ignores the extracted spike.

  In sorting process, SpikeTaro calculates correlation coefficients (contribution ratios) between all combinations of extracted spikes. Subsequently, SpikeTaro sorts spikes according with contribution ratio. User can set "contribution ratio threshold" as a sorting parameter. SpikeTaro shows better sorting performance, because SpikeTaro uses whole shape information of a spike waveform.

  In the calculation of correlation, SpikeTaro applies natural spline interpolation to single spike waveform in order to avoid data clipping error.

  Below is an example of sorting results by SpikeTaro.

  SpikeTaro also outputs a raster plot. The horizontal axis of raster plot indicates time, and the vertical axis indicates cluster ID which corresponds to a sorted cluster. Raster plot can be used for time sequential analysis of the each spike unit.

  SpikeTaro also has clustering function that is independent of sorting algorithm. The clustering algorithm is Ward method indicating high classification sensitivity. Spike heights and spike durations are used as parameters of the clustering process.

  Parameters of spike height and spike duration are measured against only complete spike waveforms. At the clustering process, SpikeTaro automatically extracts this complete spike waveforms, and applies Ward method. SpikeTaro outputs clear classification result, because the input spike waveforms to the clustering process do not include apparent error which is produced by many superimposed spike waveforms around the target spike.

SpikeTaro works under the following environment.

OS Type	Microsoft Windows 7 (64, 32 bit), Vista (64, 32 bit), XP SP2 or later.
Memory	2 GB or more. (NOTE)
CPU	Intel(R) Core(TM)2 Duo 1 GHz or more.
Others	Internet connection is required for installation.

(NOTE) The amount of data SpikeTaro can process is dependent on the memory amount of the environment. If a lot of memories can be used, large data can be processed.

  1. SpikeTaro 1.0 Product Introduction
  2. SpikeTaro 1.0 Install Manual
  3. SpikeTaro 1.0 User Manual

  SpikeTaro ver. 1.0      Price USD 3,699 or EUR 2,699

  Please feel free to send e-mail to the following address to buy SpikeTaro, or to get detailed information for considering to buy.
   

  Payment is through a bank transfer. Payment after delivery is assumed for colleges, universities, and public research organizations. Prepayment before delivery is assumed for other organizations and individuals.

  We provide evaluation software to researchers considering to buy. Satisfy yourself before purchase for performance, accuracy and convenience of SpikeTaro.

  SpikeTaro has been adopted to the following customers:
  Nippon Telegraph and Telephone Corporation (NTT), Kobe University, Hokkaido University, RIKEN, Keio University, Fukuoka University, Tokushima University, Tokai University, Tohoku University

  Chinou Jouhou Shisutemu Inc.
  Room 609 KRP9, 91 Chudoji-Awata-cho,
  Shimogyo-ku, Kyoto-shi, Kyoto, 600-8815, Japan
  TEL: +81-75-321-7300
  FAX: +81-75-321-7305

  Chief Developer: Dr. Koutaroh OKADA
  An experienced engineer with long standing research in the field of neurophysiology. Detailed achievements and attainments can be viewed in the following URL.
  https://www.chino-js.com/en/service-engineer.html#okada.koutaroh