62. A method for filtering ultra-rapid cooling profiles using orthogonal wavelets

The temperature profile during the process of cooling is an important parameter in the design and description of any protocol for cryopreservation. Rapid and ultra-rapid cooling procedures, in which liquid nitrogen (LN 2 ) is used as cooling agent, require special equipment and methods in order to record and analyze the thermal history of the sample to be cryopreserved. Natural time series are a combination of stochastic behavior (noise) and determinist chaotic behavior. In cases where noise is only present on specific high frequency bands, non-linnear filtering process can be implemented in order to reduce noise without altering metric invariants. This is the case of temporal series obtained from cooling process which we are currently studying. Using a small diameter thermocouple embedded within the sample (20 μl of distilled water) mounted on a device which ensures a uniform velocity of immersion, temperature is recorded at high rate (1000 samples/s) during the cooling of the sample in LN 2 . (Juan de Paz et al., OBI 2011, O15, ISBN 978-987-27301-0-9; Kleinhans et al., Cryobiology 61 (2010) 61–63) With this configuration, system noise was ∼±2.9 °C at room temperature and ∼±6.9 °C at LN 2 temperature (−196 °C). This noise may, in some cases, cause difficulties with the analysis and interpretation of the signal. With the intention of eliminating high frequency noise, we applied a filter in such a way that in level “ j ” of the process, the residuals will be a softer version of the original signal having less high frequency signals in contrast with level “ j  + 1” and half number of data. These are the main points by which, in the present work, we have used a signal isolation method based on orthogonal wavelets, in order to analyze the remaining signal with minimum modification of the associated dynamics. With this filtering method system noise was virtually eliminated allowing a more precise interpretation of the significance of cooling curves.