The Choice of Suitable Online Analytical Techniques and Data Processing for Monitoring of Bioprocesses - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 264

2.2.4 Data Pre-Treatment and Enhancement
Finding and exploiting an appropriate correlation method is not the only challenge
to be tackled when implementing dielectric spectroscopy for monitoring a process.
Signal noise and lack of robustness are the principal hurdles that need to be
overcome. The former issue has been extensively addressed by implementing
appropriate filtering techniques [ 44 , 45 , 49 , 50 , 81 , 87 ]. Noise resulting from
agitation can be filtered out using a low-pass filter with time constant of 1 s [ 49 ,
50 ] Most manufacturers of dielectric spectrometers have integrated appropriate
filters into the provided software. However, there is still a need to correct for
changes in conditions over time. Predictive modelling [ 46 ] as well as data rec-
onciliation [ 63 ] can improve the reliability of the measurements.
2.2.5 Range of Measurement and Limits of Application
Dielectric spectroscopy gives real-time information about cell volume and viable
cell number and, depending on the application, even insights into changes in
physiology or morphology of cells. Commercially available devices have different
ranges of validity and linearity of measurements and also find different applica-
tions. Real-time culture monitoring in a 96-well plate is possible using the
xCELLigence system and is applied to mammalian cell cultures at micro-scale
[ 91 ]. Aber Instruments as well as the more recent Fogal range of devices find their
application in monitoring cultures of all types of organisms, but mostly at labo-
ratory scale. Aber devices are very commonly used in brewing industry, where
there are even well-established control tools. Devices from Hewlett Packard [ 43 ,
90 , 92 - 94 ] and Edgerton et al. [ 81 ], as well as some micro-devices [ 80 , 95 , 96 ]
only play a marginal role in the field of bioprocess monitoring by dielectric
spectroscopy. The most commonly encountered devices have validity ranges from
2 9 10 5 cell/mL to a maximum of 10 9 cell/mL, corresponding to approximately
1-200 g/L of dry cell weight for yeast cells. When working with Gram-positive
organisms, it should be kept in mind that the a-dispersion, not relevant when
working with animal cells or Gram-negative microbes, plays an important role [ 45 ,
81 , 97 ]. Cultures involving highly conductive medium cannot be monitored by
traditional dielectric spectrometers since these have an upper conductivity limit of
100 mS/cm [ 61 ]. The sensitivity of the method, especially at low cell concentra-
tion, is limited [ 98 ].
2.2.6 Potential of Industrial Application: Perspectives and Limitations
Monitoring and process understanding is one part of PAT; process control is quite
another. Dielectric spectroscopy is a potentially important tool to gain better
process understanding and to monitor one of the most vital process parameters in
biotechnology. Applications in the area of bioprocess control are less common.