Tapio Seppänen

An experimental comparison of autoregressive and Fourier-based descriptors in 2D shape classification

An experimental comparison of shape classification methods based on autoregressive modeling and Fourier descriptors of closed contours is carried out. The performance is evaluated using two independent sets of data: images of letters and airplanes. Silhouette contours are extracted from non-occluded 2D objects rotated, scaled, and translated in 3D space. Several versions of both types of methods are implemented and tested systematically. The comparison clearly shows better performance of Fourier-based methods, especially for images containing noise. >

Vagal modulation of heart rate during exercise: effects of age and physical fitness

This study was designed to assess the effects of age and physical fitness on vagal modulation of heart rate (HR) during exercise by analyzing the instantaneous R-R interval variability from Poincaré plots (SD1) at rest and at different phases of a bicycle exercise test in a population of healthy males. SD1 normalized for the average R-R interval (SD1n), a measure of vagal activity, was compared at rest and during exercise among subjects of ages 24-34 (young, n = 25), 35-46 (middle-aged, n = 30), and 47-64 yr (old, n = 25) matched for peak O2 consumption (V˙o 2 peak) and among subjects withV˙o 2 peak of 28-37 (poor, n = 25), 38-45 (average, n = 36), and 46-60 ml ⋅ kg-1 ⋅ min-1(good, n = 25) matched for age. SD1n was higher at rest in the young subjects than in the middle-aged or old subjects (39 ± 14, 27 ± 16, and 21 ± 8, respectively; P < 0.001), but the age-related differences in SD1n were smaller during exercise [e.g., 11 ± 5, 9 ± 5, and 8 ± 4 at the level of 100 W; P = not significant (NS)]. The age-matched subjects with good, average, and poor V˙o 2 peakshowed no difference in SD1n at rest (32 ± 17, 28 ± 13, and 26 ± 11, respectively; P = NS), but SD1n differed significantly among the groups from a low to a moderate exercise intensity level (e.g., 13 ± 6, 10 ± 5, and 6 ± 3 for good, average, and poor fitness groups, respectively; P < 0.001, 100 W). These data show that poor physical fitness is associated with an impairment of cardiac vagal function during exercise, whereas aging itself results in more evident impairment of vagal function at rest.

Recognizing human motion with multiple acceleration sensors

In this paper experiments with acceleration sensors are described for human activity recognition of a wearable device user. The use of principal component analysis and independent component analysis with a wavelet transform is tested for feature generation. Recognition of human activity is examined with a multilayer perceptron classifier. Best classification results for recognition of different human motion were 83-90%, and they were achieved by utilizing independent component analysis and principal component analysis. The difference between these methods turned out to be negligible.

Frequency domain measures of heart rate variability before the onset of nonsustained and sustained ventricular tachycardia in patients with coronary artery disease.

Background. Low heart rate variability (HRV) is associated with an increased risk of arrhythmic death and ventricular tachycardia (VT). The purpose of this study was to examine whether there is a temporal relation between changes in HRV and the onset of spontaneous episodes of VT in patients at high risk of life‐threatening arrhythmias. Methods and Results. Components of HRV in the frequency domain were analyzed before the onset of 28 episodes of nonsustained VT (more than four impulses; duration <30 seconds) and 12 episodes of sustained VT (>30 seconds or requiring defibrillation) in 18 patients with coronary artery disease. Seven patients had survived cardiac arrest not associated with acute myocardial infarction, and 11 had a history of sustained VT. All frequency domain measures of HRV, i.e., total power (p <0.001), high‐frequency power (p <0.05), low‐frequency power (p <0.01), very‐low‐frequency power (p <0.01), and ultralow‐frequency power (p <0.05), were significantly lower before the onset of sustained VT than before nonsustained VT. Total power of HRV was also lower during the 1‐hour period before the onset of sustained VT than the average 24‐hour HRV (p <0.05). An indirect correlation existed between the length of VT and the total power of HRV analyzed during the 15 minutes before the onset of VT (r = 0.54, p < 0.01). HRV had a trend toward increasing values before the onset of nonsustained VT (p < 0.01) but not before the sustained VT episodes. The ratio between low‐frequency and high‐frequency powers increased substantially before both nonsustained and sustained VT episodes (p = 0.06 and p = 0.05, respectively). The rate of VT or the coupling interval initiating the VT did not differ significantly between the nonsustained and sustained VT. Conclusions. Spontaneous episodes of VT are preceded by changes in HRV in the frequency domain. Divergent dynamics of HRV before the onset of nonsustained and sustained VT episodes may reflect differences in factors that can facilitate the perpetuation of these arrhythmias. (Circulation 1993;87:1220‐1228)

Power-Law Relationship of Heart Rate Variability as a Predictor of Mortality in the Elderly

BACKGROUND The prognostic role of heart rate (HR) variability analyzed from 24-hour ECG recordings in the general population is not well known. We studied whether analysis of 24-hour HR behavior is able to predict mortality in a random population of elderly subjects. METHODS AND RESULTS A random sample of 347 subjects of > or =65 years of age (mean, 73+/-6 years) underwent a comprehensive clinical evaluation, laboratory tests, and 24-hour ECG recordings and were subsequently followed up for 10 years. Various spectral and nonspectral measures of HR variability were analyzed from the baseline 24-hour ECG recordings. Risk factors for all-cause, cardiac, cerebrovascular, cancer, and other causes of death were assessed. By the end of 10-year follow-up, 184 subjects (53%) had died and 163 (47%) were still alive. Seventy-four subjects (21%) had died of cardiac disease, 37 of cancer (11%), 25 of cerebrovascular disease (7%), and 48 (14%) of various other causes. Among all analyzed variables, a steep slope of the power-law regression line of HR variability (< -1.50) was the best univariate predictor of all-cause mortality (odds ratio, 7.9; 95% confidence interval [CI], 3.7 to 17.0; P<.0001). After adjusting for age and sex and including all univariate predictors of mortality in the proportional hazards analysis, ie, measures of HR variability, history of heart disease, functional class, smoking, medication, and blood cholesterol and glucose concentrations, all-cause mortality was predicted only by the slope of HR variability (adjusted relative risk, 1.74; 95% CI, 1.42 to 2.13; P<.0001) and a history of congestive heart failure (adjusted relative risk, 1.70; P=.0002). The slope of HR variability predicted both cardiac (adjusted relative risk, 2.05; P=.0002) and cerebrovascular death (adjusted relative risk, 2.84; P=.0001) but not cancer or other causes of death. CONCLUSIONS Power-law relationship of 24-hour HR variability is a more powerful predictor of death than the traditional risk markers in elderly subjects. Altered long-term behavior of HR implies an increased risk of vascular causes of death rather than being a marker of any disease or frailty leading to death.

Abnormalities in Beat-to-Beat Dynamics of Heart Rate Before the Spontaneous Onset of Life-Threatening Ventricular Tachyarrhythmias in Patients With Prior Myocardial Infarction

BACKGROUND Beat-to-beat analysis of RR intervals can reveal patterns of heart-rate dynamics, which are not easily detected by summary measures of heart-rate variability. This study was designed to test the hypothesis that alterations in RR-interval dynamics occur before the spontaneous onset of ventricular tachyarrhythmias (VT). METHODS AND RESULTS Ambulatory ECG recordings from 15 patients with prior myocardial infarction (MI) who had spontaneous episodes of sustained VT during the recording and VT inducible by programmed electrical stimulation (VT group) were analyzed by plotting each RR interval of a sinus beat as a function of the previous one (Poincaré plot). Poincaré plots were also generated for 30 post-MI patients who had no history of spontaneous VT events and no inducible VT (MI control subjects) and for 30 age-matched subjects without heart disease (normal control subjects). The MI control subjects and VT group were matched with respect to age and severity of underlying heart disease. All the healthy subjects and MI control subjects showed fan-shaped Poincaré plots characterized by an increased next-interval difference for long RR intervals relative to short ones. All the VT patients had abnormal plots: 9 with a complex pattern, 3 ball-shaped, and 3 torpedo-shaped. Quantitative analysis of the Poincare plots showed the SD of the long-term RR-interval variability (SD2) to be smaller in all VT patients (52+/-14 ms; range, 31 to 75 ms) than in MI control subjects (110+/-24 ms; range, 78 to 179 ms, P<.001) or the normal control subjects (123+/-38 ms, P<.001), but the SD of the instantaneous beat-to-beat variability (SD1) did not differ between the groups. The complex plots were caused by periods of alternating sinus intervals, resulting in an increased SD1/SD2 ratio in the VT group. This ratio increased during the 1-hour preceding the onset of 27 spontaneous VT episodes (0.43+/-0.20) compared with the 24-hour average ratio (0.33+/-0.19) (P<.01). CONCLUSIONS Reduced long-term RR-interval variability, associated with episodes of beta-to-beat sinus alternans, is a highly specific sign of a propensity for spontaneous onset of VT, suggesting that abnormal beat-to-beat heart-rate dynamics may reflect a transient electrical instability favoring the onset of VT in patients conditioned by structurally abnormal hearts.

Dynamic analysis of heart rate may predict subsequent ventricular tachycardia after myocardial infarction.

Dynamics analysis of RR interval behavior and traditional measures of heart rate variability were compared between postinfarction patients with and without vulnerability to ventricular tachyarrhythmias in a case-control study. Short-term fractal correlation of heart rate dynamics was better than traditional measures of heart rate variability in differentiating patients with and without life-threatening arrhythmias.

Physiological Background of the Loss of Fractal Heart Rate Dynamics

Background—Altered fractal heart rate (HR) dynamics occur during various disease states, but the physiological background of abnormal fractal HR behavior is not well known. We tested the hypothesis that the fractal organization of human HR dynamics is determined by the balance between sympathetic and vagal outflow. Methods and Results—A short-term fractal scaling exponent (&agr;1) of HR dynamics, analyzed by the detrended fluctuation analysis (DFA) method, and the high-frequency (HF) and low-frequency (LF) spectral components of R-R intervals (0.15 to 0.4 Hz; n=13), along with muscle sympathetic nervous activity (MSNA) from the peroneus nerve (n=11), were assessed at rest and during cold face and cold hand immersion in healthy subjects. During cold face immersion, HF power increased (from 6.9±1.3 to 7.6±1.2 ln ms2, P<0.01), as did MSNA (from 32±17 to 44±14 bursts/100 heartbeats, P<0.001), and LF/HF ratio decreased (P<0.01). Cold hand immersion resulted in a similar increase in MSNA (from 34±17 to 52±19 bursts/100 heartbeats, P<0.001) but a decrease in HF spectral power (from 7.0±1.3 to 6.5±1.1 ln ms2, P<0.05) and an increase in the LF/HF ratio (P<0.05). The fractal scaling index &agr;1 decreased in all subjects (from 0.85±0.27 to 0.67±0.30, P<0.0001) during cold face immersion but increased during cold hand immersion (from 0.77±0.22 to 0.97±0.20, P<0.01). Conclusions—The fractal organization of human HR dynamics is determined by a delicate interplay between sympathetic and vagal outflow, with the breakdown of fractal HR behavior toward more random dynamics occurring during coactivation of sympathetic and vagal outflow.

Adaptive document binarization

A new method is presented for adaptive document image binarization, where the page is considered as a collection of subcomponents such as text, background and picture. The problems caused by noise, illumination and many source type related degradations are addressed. The algorithm uses document characteristics to determine (surface) attributes, often used in document segmentation. Using characteristic analysis, two new algorithms are applied to determine a local threshold for each pixel. An algorithm based on soft decision control is used for thresholding the background and picture regions. An approach utilizing local mean and variance of gray values is applied to textual regions. Tests were performed with images including different types of document components and degradations. The results show that the method adapts and performs well in each case.

Circadian rhythm of heart rate variability in survivors of cardiac arrest

Reduced heart rate (HR) variability is associated with increased risk of cardiac arrest in patients with coronary artery disease. In this study, the power spectral components of HR variability and their circadian pattern in 22 survivors of out-of-hospital cardiac arrest not associated with acute myocardial infarction were compared with those of 22 control patients matched with respect to age, sex, previous myocardial infarction, ejection fraction and number of diseased coronary arteries. Survivors of cardiac arrest had significantly lower 24-hour average standard deviation of RR intervals than control patients (29 +/- 10 vs 51 +/- 15 ms, p less than 0.001), and the 24-hour mean high frequency spectral area was also lower in survivors of cardiac arrest than in control patients (13 +/- 7 ms2 x 10 vs 28 +/- 14 ms2 x 10, p less than 0.01). In a single cosinor analysis, a significant circadian rhythm of HR variability was observed in both groups with the acrophase of standard deviation of RR intervals and high-frequency spectral area occurring between 3 and 6 A.M. which was followed by an abrupt decrease in HR variability after arousal. The amplitude of the circadian rhythm of HR variability did not differ between the groups. Thus, HR variability is reduced in survivors of cardiac arrest but its circadian rhythm is maintained so that a very low HR variability is observed in the morning after awakening, corresponding to the time period at which the incidence of sudden cardiac death is highest.