Salme Kärkkäinen

EMG, Heart Rate, and Accelerometer as Estimators of Energy Expenditure in Locomotion.

PURPOSE Precise measures of energy expenditure (EE) during everyday activities are needed. This study assessed the validity of novel shorts measuring EMG and compared this method with HR and accelerometry (ACC) when estimating EE. METHODS Fifty-four volunteers (39.4 ± 13.9 yr) performed a maximal treadmill test (3-min loads) including walking with different speeds uphill, downhill, and on level ground and one running load. The data were categorized into all, low, and level loads. EE was measured by indirect calorimetry, whereas HR, ACC, and EMG were measured continuously. EMG from quadriceps (Q) and hamstrings (H) was measured using shorts with textile electrodes. Validity of the methods used to estimate EE was compared using Pearson correlations, regression coefficients, linear mixed models providing Akaike information criteria, and root mean squared error (RMSE) from cross-validation at the individual and population levels. RESULTS At all loads, correlations with EE were as follows: EMG(QH), 0.94 ± 0.03; EMG(Q), 0.91 ± 0.03; EMG(H), 0.94 ± 0.03; HR, 0.96 ± 0.04; and ACC, 0.77 ± 0.10. The corresponding correlations at low loads were 0.89 ± 0.08, 0.79 ± 0.10, 0.93 ± 0.07, 0.89 ± 0.23, and 0.80 ± 0.07, and at level loads, they were 0.97 ± 0.03, 0.97 ± 0.05, 0.96 ± 0.04, 0.95 ± 0.08, and 0.99 ± 0.02, respectively. Akaike information criteria ranked the methods in accordance with the individual correlations. CONCLUSIONS It is shown for the first time that EMG shorts can be used for EE estimations across a wide range of physical activity intensities in a heterogeneous group. Across all loads, HR is a superior method of predicting EE, whereas ACC is most accurate for level loads at the population level. At low levels of physical activity in changing terrains, thigh muscle EMG provides more accurate EE estimations than those in ACC and HR if individual calibrations are performed.

Evaluating the performance of artificial neural networks for the classification of freshwater benthic macroinvertebrates

Abstract Macroinvertebrates form an important functional component of aquatic ecosystems. Their ability to indicate various types of anthropogenic stressors is widely recognized which has made them an integral component of freshwater biomonitoring. The use of macroinvertebrates in biomonitoring is dependent on manual taxa identification which is currently a time-consuming and cost-intensive process conducted by highly trained taxonomical experts. Automated taxa identification of macroinvertebrates is a relatively recent research development. Previous studies have displayed great potential for solutions to this demanding data mining application. In this research we have a collection of 1350 images from eight different macroinvertebrate taxa and the aim is to examine the suitability of artificial neural networks (ANNs) for automated taxa identification of macroinvertebrates. More specifically, the focus is drawn on different training algorithms of Multi-Layer Perceptron (MLP), probabilistic neural network (PNN) and Radial Basis Function network (RBFN). We performed thorough experimental tests and we tested altogether 13 training algorithms for MLPs. The best classification accuracy of MLPs, 95.3%, was obtained by two conjugate gradient backpropagation variations and scaled conjugate gradient backpropagation. For PNN 92.8% and for RBFN 95.7% accuracies were achieved. The results show how important a proper choice of ANN is in order to obtain high accuracy in the automated taxa identification of macroinvertebrates and the obtained model can outperform the level of identification which is made by a taxonomist.

Statistical classification and proportion estimation - an application to a macroinvertebrate image database

We apply and compare a random Bayes forest classifier and three traditional classification methods to a dataset of complex benthic macroinvertebrate images of known taxonomical identity. Since in biomonitoring changes in benthic macroinvertebrate taxa proportions correspond to changes in water quality, their correct estimation is pivotal. As classification errors are passed on to the allocated proportions, we explore a correction method known as a confusion matrix correction. Classification methods were compared using the misclassification error and the χ2 distance measures of the true proportions to the allocated and to the corrected proportions. Using low misclassification error and smallest χ2 distance measures as performance criteria the classical Bayes classifier performed best followed closely by the random Bayes forest.

On the orientational analysis of planar fibre systems from digital images

The orientational characteristics of fibres in digital images are studied. The fibres are modelled by a planar Boolean model whose typical grain is a thick (coloured) fibre. The aim is to make stereological inference on the rose of directions of the unobservable central fibres from observations made on a digital image of the thick fibres. For central fibres, the relation between the rose of directions and the point intensity, observed on a sampling line, is known. We derive, under regularity conditions, the relation between the unobservable point intensity and the scaled variogram observed on the line in a binary and a greyscale image. Using such a relation, it is possible to draw inference about the rose of directions from the scaled variogram, which is easy and quick to determine in a digital image.

Understanding the statistical properties of the percent model affinity index can improve biomonitoring related decision making

The percent model affinity (PMA) index is used to measure the similarity of two probability profiles representing, for example, an ideal profile (i.e. reference condition) and a monitored profile (i.e. possibly impacted condition). The goal of this work is to study the effects of sample size, evenness, true value of the index and number of classes on the statistical properties of the estimator of the PMA index. We derive and extend previous formulas of the expectation and variance of the estimator for estimated monitored profile and fixed reference profile. Using the obtained extension, we find that the estimator is asymptotically unbiased, converging faster when the profiles differ. When both profiles are estimated, we calculate the expectation using transformation rules for expectation and in addition derive the formula for the estimator’s variance. Since the computation of the probabilities in the variance formula is slow, we study the behavior of the variance with simulation experiments and assess whether it could be approximated with the variance for the fixed reference profile. Finally, we provide a set of recommendations for the users of the PMA index to avoid the most common caveats of the index.

Benchmark database for fine-grained image classification of benthic macroinvertebrates

Abstract Managing the water quality of freshwaters is a crucial task worldwide. One of the most used methods to biomonitor water quality is to sample benthic macroinvertebrate communities, in particular to examine the presence and proportion of certain species. This paper presents a benchmark database for automatic visual classification methods to evaluate their ability for distinguishing visually similar categories of aquatic macroinvertebrate taxa. We make publicly available a new database, containing 64 types of freshwater macroinvertebrates, ranging in number of images per category from 7 to 577. The database is divided into three datasets, varying in number of categories (64, 29, and 9 categories). Furthermore, in order to accomplish a baseline evaluation performance, we present the classification results of Convolutional Neural Networks (CNNs) that are widely used for deep learning tasks in large databases. Besides CNNs, we experimented with several other well-known classification methods using deep features extracted from the data.

Dangerous relationships: biases in freshwater bioassessment based on observed to expected ratios

The ecological assessment of freshwaters is currently primarily based on biological communities and the reference condition approach (RCA). In the RCA, the communities in streams and lakes disturbed by humans are compared with communities in reference conditions with no or minimal anthropogenic influence. The currently favored rationale is using selected community metrics for which the expected values (E) for each site are typically estimated from environmental variables using a predictive model based on the reference data. The proportional differences between the observed values (O) and E are then derived, and the decision rules for status assessment are based on fixed (typically 10th or 25th) percentiles of the O/E ratios among reference sites. Based on mathematical formulations, illustrations by simulated data and real case studies representing such an assessment approach, we demonstrate that the use of a common quantile of O/E ratios will, under certain conditions, cause severe bias in decision making even if the predictive model would be unbiased. This is because the variance of O/E under these conditions, which seem to be quite common among the published applications, varies systematically with E. We propose a correction method for the bias and compare the novel approach to the conventional one in our case studies, with data from both reference and impacted sites. The results highlight a conceptual issue of employing ratios in the status assessment. In some cases using the absolute deviations instead provides a simple solution for the bias identified and might also be more ecologically relevant and defensible.

Orientational analysis of planar fibre systems observed as a Poisson shot-noise process.

We consider two‐dimensional fibrous materials observed as a digital greyscale image. The problem addressed is to estimate the orientation distribution of unobservable thin fibres from a greyscale image modelled by a planar Poisson shot‐noise process. The classical stereological approach is not straightforward, because the point intensities of thin fibres along sampling lines may not be observable. For such cases, Kärkkäinen et al. (2001) suggested the use of scaled variograms determined from grey values along sampling lines in several directions. Their method is based on the assumption that the proportion between the scaled variograms and point intensities in all directions of sampling lines is constant. This assumption is proved to be valid asymptotically for Boolean models and dead leaves models, under some regularity conditions. In this work, we derive the scaled variogram and its approximations for a planar Poisson shot‐noise process using the modified Bessel function. In the case of reasonable high resolution of the observed image, the scaled variogram has an approximate functional relation to the point intensity, and in the case of high resolution the relation is proportional. As the obtained relations are approximative, they are tested on simulations. The existing orientation analysis method based on the proportional relation is further experimented on images with different resolutions. The new result, the asymptotic proportionality between the scaled variograms and the point intensities for a Poisson shot‐noise process, completes the earlier results for the Boolean models and for the dead leaves models.