Professional Documents
Culture Documents
(A) Hypoglycemia-Induced EEG Complexity Changes in Type 1 Diabete
(A) Hypoglycemia-Induced EEG Complexity Changes in Type 1 Diabete
a r t i c l e i n f o a b s t r a c t
Article history: In recent years, hypoglycemia-induced changes in the EEG signal of patients with Type 1 diabetes (T1D)
Received 29 April 2016 have been quantified and studied mainly by linear approaches. So far, sample entropy (SampEn) is the
Received in revised form 23 April 2017 only nonlinear measure used in the literature. SampEn has the disadvantage of being computationally
Accepted 4 June 2017
demanding and, hence, difficult to be used in real-time settings. The present study investigates whether
other nonlinear indicators, less computationally demanding than SampEn, can be equally sensitive to
Keywords:
changes in the EEG signal induced by hypoglycemia. For such a scope, we considered a database obtained
Type 1 diabetes
from 19 T1D patients who underwent a hyperinsulinemic-hypoglycemic clamp while continuous EEG
Hypoglycemia
EEG
was recorded. We analyzed the P3-C3 EEG derivation data using three measures of signal complexity
Signal complexity based on an approach originally proposed by Higuchi in the 80s: the original measure of fractal dimen-
Higuchi fractal dimension sion and two new indexes based on the Higuchi’s curve. All the three indicators revealed a statistically
Nonlinear time-series analysis significant decrease in EEG complexity in the hypo- versus euglycemic state, which is in line with the
results previously obtained with SampEn. However, the lower computational cost of the proposed indi-
cators (∼O(N) versus ∼O(N2 )) makes them potentially more suited for real-time applications such as the
use of EEG to trigger hypoglycemia alerts.
© 2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bspc.2017.06.004
1746-8094/© 2017 Elsevier Ltd. All rights reserved.
F. Scarpa et al. / Biomedical Signal Processing and Control 38 (2017) 168–173 169
may show trends, low frequency oscillations and baseline changes nal itself [25]. Thus, we also extracted two additional features [25],
due to these components. In order to remove these artifacts, we which consider a wider domain of the k curve with respect to the
applied a de-trending procedure to each EEG epoch to correct drifts Higuchi fractal dimension. The first feature evaluates the deviation
and linear trends. Briefly, a first order model consisting of an offset (i.e. the sum of squares of the residuals) of the k curve from the
plus a linear trend was fitted to each epoch, estimated and then regression line computed in its linear region. The second feature is
subtracted from the epoch itself. Furthermore, for each epoch, if a tortuosity measures of k . It consists in a measure of the rate at
the value of one or more of its samples was greater than 100 V which the curve is changing with respect to its coordinates changes
or lower than −100 V, the epoch was considered as affected by (x = log(k) and y = log(L(k))), by using their first and second partial
artifacts and thus not considered for the analysis (on average, 0.4% differences:
epochs rejected per subject). kmax x (n) 2 y (n) − 2 x (n) y (n)
k = | | (3)
n=3 2 2 3/2
3. Methods (x (n)) + (y (n))
Fig. 2. Distributions of the values for 4-s epochs in euglycemia (green distribution) and hypoglycemia (blue distribution) of Higuchi’s fractal dimension (left), residuals
(middle) and tortuosity (right) for a single subject, here used as an example. (Color graphics are available online). (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
Fig. 3. Distributions of the values for 4-s epochs in euglycemia (green distribution) and hypoglycemia (blue distribution) of Higuchi’s fractal dimension (left), residuals
(middle) and tortuosity (right) for all subjects. (Color graphics are available online). (For interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)
tion of N mathematical operations for each k = 1:K, with K smaller of the accuracy in the real-time identification of hypoglycemic
than N (in this work, events by devices similar to that proposed in [6] and to an improved
K = klin = 6 for Higuchi’s fractal dimension, K = kmax = 30 for resid- temporal resolution in the investigation of the mechanisms that
uals and tortuosity, and K = kmax = 30 for all of them simultaneously). link EEG and hypoglycemia. As regard to the on-line detection
The short epoch length and the low computational cost of the frac- of hypo-alerts, it is worth noticing that in subjects with hypo-
tal features allow frequent assessment of the physiological state, glycemia unawareness, both hypoglycemia-induced EEG changes
which is essential for real-time detection of hypoglycemia. and altered cognitive functions during hypoglycemia are distin-
On epochs of the same length (4 s), entropy based methods, i.e. guishable and dissociated from the perception of hypoglycemia
SampEn, could be applied without a scale factor, but their computa- symptoms and the hormonal counter-regulatory responses, which
tional cost, becoming O(N2 ) (see Appendix A), would be extremely are blunted in these patients [35]. Nevertheless, devices develop-
higher than the one of fractal analysis. This computational cost can ing further the prototype documented in [6] require proper clinical
be drastically reduced using SampEn at a scale factor , O((N/)2 ) testing and careful assessment of fundamental issues, such as sen-
(see Appendix A). However, this requires a greater number of sam- sitivity, specificity, and robustness against signal noise and artifacts
ples [33], that involves longer epochs, i.e. 60-s epochs (N = 12,000). caused by daily-life activity. From a methodological point of view,
The high number of samples to be simultaneously analyzed makes proper efforts should be directed towards the development of an
it difficult to be implemented in a real-time application and the long algorithm for efficient and robust identification of hypoglycemic
time interval of each epoch can potentially affect the readiness to events, starting from a careful selection of the EEG indicators that,
identify the hypoglycemic state. alone or in combination, result most sensitive to hypoglycemia.
For sake of completeness, we also analyzed data in euglycemia In this scenario, proving the sensitivity to hypoglycemia and the
after the recovery from the hypoglycemic interval (see Fig. 1). 1-h computational efficiency of quantitative EEG indicators, as done in
data acquired during this latter euglycemic interval (after the hypo- this study, is a pivotal step to clinically study how hypoglycemia
glycemic interval) were available for few subjects only, and thus affects brain function and to develop automatic systems able to
an exhaustive investigation could not be conducted. However, the detect hypoglycemia almost in real-time and in daily-life condi-
analysis on the available data revealed the same results obtained tions.
from the 1-h interval in euglycemia before the entering in hypo-
glycemia, showing again that EEG complexity was diminished in Acknowledgement
hypoglycemia. This suggests that the changes in the signal proper-
ties are not caused by the fact that the hypoglycemic period occurs The authors thank Hyposafe A/S, Lynge, Denmark for assisting
much later than the euglycemic period, nor by the first adjustment with providing the data.
period before entering in euglycemia.
Appendix A.
The computational complexity of the SampEn algorithm is O(N2 ) [13] J. Escudero, D. Abásolo, R. Hornero, P. Espino, M. López, Analysis of
because two loops are required. electroencephalograms in Alzheimer’s disease patients with multiscale
entropy, Physiol. Meas. 27 (11) (2006) 1091–1106.
As a development of SampEn, the MSE approach may be consid- [14] S. Farashi, A multiresolution time-dependent entropy method for QRS
ered to investigate the complexity of a time-series, assessing how complex detection, Biomed. Signal Process. Control 24 (2016) 63–71.
its entropy varies at increasing temporal scales. From the original [15] H. Azami, J. Escudero, Improved multiscale permutation entropy for
biomedical signal analysis: interpretation and application to
time-series u = {u(1), u(2), . . ., u(N)}, a new time-series y = {y (1), electroencephalogram recordings, Biomed. Signal Process. Control 23 (2016)
y (2), . . ., y (N/)} is defined as: 28–41.
1 j
[16] C. Fabris, W. De Colle, G. Sparacino, Voice disorders assessed by (cross-)
sample entropy of electroglottogram and microphone signals, Biomed. Signal
y (j) = u (i) , j = 1, . . ., N/ (A.6) Process. Control 8 (6) (2013) 920–926.
i=(j−1)+1
[17] U.R. Acharya, F. Molinari, S.V. Sree, S. Chattopadhyay, K.H. Ng, J.S. Suri,
where the integer positive parameter represents the chosen time Automated diagnosis of epileptic EEG using entropies, Biomed. Signal Process.
Control 7 (4) (2012) 401–408.
scale. Then, SampEn is computed for several values of .
[18] S. Simons, D. Abásolo, J. Escudero, Classification of Alzheimer’s disease from
quadratic sample entropy of electroencephalogram, Healthc. Technol. Lett. 2
MSE (u, , m, r) = SampEn (y , m, r) (A.7)
(3) (2015) 70–73.
[19] L. Faes, G. Nollo, A. Porta, Compensated transfer entropy as a tool for reliably
In practice, the MSE approach reduces the length of the
estimating information transfer in physiological time series, Entropy 15 (1)
original time-series by averaging consecutive samples belong- (2013) 198–219.
ing to non-overlapping blocks. This coarse-graining procedure [20] X. Zhang, R. Roy, E. Jensen, EEG complexity as a measure of depth of
anesthesia for patients, IEEE Trans. Biomed. Eng. 48 (12) (2001) 1424–1433.
allows obtaining averaged down-sampled versions of the original
[21] T. Higuchi, Approach to an irregular time series on the basis of the fractal
time-series, whose SampEn values are related to the complex fluc- theory, Physica D 31 (2) (1988) 277–283.
tuations of the original signal running at temporal scales times [22] M. Ronzhina, O. Janoušek, J. Kolářová, M. Nováková, P. Honzík, I. Provazník,
larger than the sampling time used to acquire the data. At a scale Sleep scoring using artificial neural networks, Sleep Med. Rev. 16 (3) (2012)
251–263.
factor of , the execution time required to calculate the SampEn [23] A. Accardo, M. Affinito, M. Carrozzi, F. Bouquet, Use of fractal dimension for
of the coarse-grained time series is O((N/)2 ). Therefore, the over- analysis of the electroencephalogram time series, Biol. Cybern. 77 (5) (1997)
339–350.
s
all computational complexity MSE approach, at s increasing
of the
2 [24] G.E. Polychronaki, P.Y. Ktonas, S. Gatzonis, A. Siatouni, P.A. Asvestas, H. Tsekou,
temporal scales is O =1
(N/) . D. Sakas, K.S. Nikita, Comparison of fractal dimension estimation algorithms
for epileptic seizure onset detection, J. Neural Eng. 7 (4) (2010) 46007.
References [25] F. Finotello, F. Scarpa, M. Zanon, EEG signal features extraction based on
fractal dimension, in: Conf Proc IEEE Eng Med Biol Soc, Milan, 2015.
[26] A.S. Sejling, T.W. Kjaer, U. Pedersen-Bjergaard, S.S. Diemar, C.S. Frandsen, L.
[1] P. Mergenthaler, U. Lindauer, G.A. Dienel, A. Meisel, Sugar for the brain: the Hilsted, J. Faber, J.J. Holst, L. Tarnow, M.N. Nielsen, L.S. Remvig, B.
role of glucose in physiological and pathological brain function, Trends Thorsteinsson, C.B. Juhl, Hypoglycemia-associated changes in
Neurosci. 36 (10) (2013) 587–597. electroencephalogram in patients with type 1 diabetes and normal
[2] S. Pramming, B. Thorsteinsson, I. Bendtson, C. Binder, The relationship hypoglycemia awareness or unawareness, Diabetes 64 (5) (2015) 1760–1769.
between symptomatic and biomechanical hypoglycaemia in [27] U. Pedersen-Bjergaard, S. Pramming, B. Thorsteinsson, Recall of severe
insulin-dependent diabetic patients, J. Intern. Med. 228 (6) (1990) 641–646. hypoglycaemia and self-estimated state of awareness in type 1 diabetes,
[3] G. Tribl, K. Howorka, G. Heger, P. Anderer, H. Thoma, J. Zeitlhofer, EEG Diabetes Metab. Res. Rev. 19 (2003) 232–240.
topography during insulin-induced hypoglycemia in patients with [28] D. Kaplan, L. Glass, Understanding Nonlinear Dynamics, Springer-Verlag, New
insulin-dependent diabetes mellitus, Eur. Neurol. 36 (5) (1996) 303–309. York, 1995.
[4] L. Hyllienmark, J. Maltez, A. Dandenell, J. Ludvigsson, T. Brismar, EEG [29] V. Cabukovski, N.M. d. Rudolf, N. Mahmood, Measuring the fractal dimension
abnormalities with and without relation to severe hypoglycaemia in of EEG signals: selection and adaptation of method for real-time analysis, WIT
adolescents with type 1 diabetes, Diabetologia 48 (3) (2005) 412–419. Trans. Biomed. Health 1 (1993) 282–292.
[5] M. Rubega, G. Sparacino, A.S. Sejling, C.B. Juhl, C. Cobelli, [30] B. Klinkenberg, A review of methods used to determine the fractal dimension
Hypoglycemia-induced decrease of EEG coherence in patients with type 1 of linear features, Math. Geol. 26 (1) (1994) 23–46.
diabetes, Diabetes Technol. Ther. 18 (3) (2016) 178–184. [31] M. Bachmann, J. Lass, A. Suhhova, H. Hinrikus, Spectral asymmetry and
[6] R. Elsborg, L.S. Remvig, H. Beck-Nielsen, C.B. Juhl, Detecting Hypoglycemia by Higuchi’s fractal dimension measures of depression electroencephalogram,
Using the Brain as a Biosensor, InTech, Rijeka, Croatia, 2011, pp. 273–292. Comput. Math. Method Med. 2013 (2013) 1–8.
[7] C. Fabris, G. Sparacino, A.S. Sejling, A. Goljahani, J. Duun-Henriksen, L.S. [32] F. Scarpa, X. Zheng, Y. Ohashi, A. Ruggeri, Automatic evaluation of corneal
Remvig, C.B. Juhl, C. Cobelli, Hypoglycemia-related EEG changes assessed by nerve tortuosity in images from in vivo confocal microscopy, Invest.
multiscale entropy, Diabetes Technol. Ther. 16 (10) (2014) 688–694. Ophthalmol. Vis. Sci. 52 (9) (2011) 6404–6408.
[8] J.S. Richman, J.R. Moorman, Physiological time-series analysis using [33] S.-D. Wu, C.-W. Wu, S.-G. Lin, K.-Y. Lee, C.-K. Peng, Analysis of complex time
approximate entropy and sample entropy, Am. J. Physiol. Heart Circ. Physiol. series using refined composite multiscale entropy, Phys. Lett. A 378 (20)
278 (6) (2000) 2039–2049. (2014) 1369–1374.
[9] M. Costa, A.L. Goldberger, C.K. Peng, Multiscale entropy analysis of complex [34] C.B. Juhl, K. Hojlund, R. Elsborg, M.K. Poulsen, P.E. Selmar, J.J. Holst, C.
physiological time series, Phys. Rev. Lett. 89 (6) (2002) 1–4. Christiansen, H. Beck-Nielsen, Automated detection of hypoglycemia-induced
[10] R. Ferenets, T. Lipping, A. Anier, V. Jäntti, S. Melto, S. Hovilehto, Comparison of EEG changes recorded by subcutaneous electrodes in subjects with type 1
entropy and complexity measures for the assessment of depth of sedation, diabetes—The brain as a biosensor, Diabetes Res. Clin. Pract. 88 (1) (2010)
IEEE Trans. Biomed. Eng. 53 (6) (2006) 1067–1077. 22–28.
[11] C. Liu, C. Zhang, L. Zhang, L. Zhao, C. Liu, H. Wang, Measuring synchronization [35] P.E. Cryer, Mechanisms of hypoglycemia-associated automatic failure in
in coupled simulation and coupled cardiovascular time series: a comparison diabetes, N. Engl. J. Med. 369 (2013) 362–372.
of different cross entropy measures, Biomed. Signal Process. Control 21
(2015) 49–57.
[12] O. Faust, U.R. Acharya, F. Molinari, S. Chattopadhyay, T. Tamura, Linear and
non-linear analysis of cardiac health in diabetic subjects, Biomed. Signal
Process. Control 7 (3) (2012) 295–302.