Information Technology Reference
In-Depth Information
II errors as a function of the number of seizures,
duration of interictal data, and prediction horizon
length were derived. The model's application
was demonstrated using a new seizure detection
algorithm that appeared to predicted seizure onset.
The framework provides some benefits involving:
precisely recognize either focal or generalized
seizures for many years (Gotman, 1982). These
methods can directly utilize use of the EEG data
or may involve Fourier or wavelet transformations.
Since recent time, the effort to identify a pre-ictal
state has been shown to be a larger confront (Litt,
2002; Petrosian, 2000; Lehnertz, 2003).
Signal processing is needed, as in Figure 2,
to recognize seizure movements. For a given
time window detection algorithm should evalu-
ate best-fit model compare to observe signal, if
suits, trigger an alarm. This triggering should
differentiate patient normal day today activities
such as general movement, walking and lying with
valid seizure movement (Nijsen, 2006). A myo-
clonic seizure can exist shorter than one second.
Frequently, many myoclonic seizures can happen
during sleeping time; hence, sleep rhythm can be
disturbed. Detection of myoclonic seizures could
be used for early warning as sever motor seizures
often proceeds from it (Nijsen, 2006).
Signal processing and filtering are the methods
to obtain a better form for a set of information,
either by reshaping it or filtering out noise. Wave-
let Transform can be in two forms: the Continuous
Wavelet Transform (CWT) and the Discrete
Wavelet Transform (DWT) and the differences
between them (Khorbotly, 2008). It is possible to
recover the signal from its samples, if, the con-
tinuous signal is sampled at a rate twice as its
highest frequency (Olenewa, 2007). To differen-
tiate numerous complex movements of seizure
patient Short-Term Fourier Transform (Thielgen,
2004) or a Wavelet Transform (Najafi, 2003) can
be used.
1. Helping quantifying the performance of a
seizure prediction algorithm against a null
hypothesis whereby the results of the algo-
rithm can be analyzed in a transparent way,
2. Generating measures facilitating the process
of calculating data requirements in clinical
trials, involving seizure prediction algo-
rithms, which are useful for clinical settings.
3. Redefining a preictal period as a stochastic,
probabilistic state where seizures might arise
from.
4. The notion of a permissive preictal state in
seizure generation modeling has the potential
to improve implanted clinical antiepileptic
devices.
The framework may be a useful tool for the
design and validation of prediction algorithms for
epileptic seizures.
Short-Time Fourier and Wavelet
Transform to be used for our CRESH
In this section we will discuss about Short time
Fourier and Wavelet Transform which is our
recommended method for the implementation
of a CRESH for epileptic seizure detection and
prediction.
Ictal state refers the period during a seizure
and pre-ictal refers the period just before seizure
onset. The periods of normal brain activity between
seizures are called inter-ictal state (Litt, 2002; Litt,
2001; Petrosian, 2000). Seizures can successfully
predicted if pre-ictal state could be identified in
the EEG, hence an early warning to the seizure
patient (Litt, 2002; Litt, 2001; Petrosian, 2000).
Numerous techniques have been developed to
Short-Term Fourier Transform (STFT)
Fourier Transform involves integrating a product
of a signal and oscillating function. Short seg-
ment of the signal are transformed, separately
and this is separated into cosine waves. However
disadvantage is possibility of unexpected break
of between segments (Strang, 1994). The Short-
Search WWH ::
Custom Search