Fluctuations in Upper and Lower Body Movement during Walking in Normal Pressure Hydrocephalus and Parkinson's Disease Assessed by Motion Capture with a Smartphone Application, TDPT-GT.

We aimed to capture the fluctuations in the dynamics of body positions and find the characteristics of them in patients with idiopathic normal pressure hydrocephalus (iNPH) and Parkinson's disease (PD). With the motion-capture application (TDPT-GT) generating 30 Hz coordinates at 27 points on the body, walking in a circle 1 m in diameter was recorded for 23 of iNPH, 23 of PD, and 92 controls. For 128 frames of calculated distances from the navel to the other points, after the Fourier transforms, the slopes (the representatives of fractality) were obtained from the graph plotting the power spectral density against the frequency in log-log coordinates. Differences in the average slopes were tested by one-way ANOVA and multiple comparisons between every two groups. A decrease in the absolute slope value indicates a departure from the 1/f noise characteristic observed in healthy variations. Significant differences in the patient groups and controls were found in all body positions, where patients always showed smaller absolute values. Our system could measure the whole body's movement and temporal variations during walking. The impaired fluctuations of body movement in the upper and lower body may contribute to gait and balance disorders in patients.

Artificial Intelligence Distinguishes Pathological Gait: The Analysis of Markerless Motion Capture Gait Data Acquired by an iOS Application (TDPT-GT).

Distinguishing pathological gait is challenging in neurology because of the difficulty of capturing total body movement and its analysis. We aimed to obtain a convenient recording with an iPhone and establish an algorithm based on deep learning. From May 2021 to November 2022 at Yamagata University Hospital, Shiga University, and Takahata Town, patients with idiopathic normal pressure hydrocephalus (n = 48), Parkinson's disease (n = 21), and other neuromuscular diseases (n = 45) comprised the pathological gait group (n = 114), and the control group consisted of 160 healthy volunteers. iPhone application TDPT-GT captured the subjects walking in a circular path of about 1 meter in diameter, a markerless motion capture system, with an iPhone camera, which generated the three-axis 30 frames per second (fps) relative coordinates of 27 body points. A light gradient boosting machine (Light GBM) with stratified k-fold cross-validation (k = 5) was applied for gait collection for about 1 min per person. The median ability model tested 200 frames of each person's data for its distinction capability, which resulted in the area under a curve of 0.719. The pathological gait captured by the iPhone could be distinguished by artificial intelligence.

Quantitative Gait Feature Assessment on Two-Dimensional Body Axis Projection Planes Converted from Three-Dimensional Coordinates Estimated with a Deep Learning Smartphone App.

To assess pathological gaits quantitatively, three-dimensional coordinates estimated with a deep learning model were converted into body axis plane projections. First, 15 healthy volunteers performed four gait patterns; that is, normal, shuffling, short-stepped, and wide-based gaits, with the Three-Dimensional Pose Tracker for Gait Test (TDPT-GT) application. Second, gaits of 47 patients with idiopathic normal pressure hydrocephalus (iNPH) and 92 healthy elderly individuals in the Takahata cohort were assessed with the TDPT-GT. Two-dimensional relative coordinates were calculated from the three-dimensional coordinates by projecting the sagittal, coronal, and axial planes. Indices of the two-dimensional relative coordinates associated with a pathological gait were comprehensively explored. The candidate indices for the shuffling gait were the angle range of the hip joint < 30° and relative vertical amplitude of the heel < 0.1 on the sagittal projection plane. For the short-stepped gait, the angle range of the knee joint < 45° on the sagittal projection plane was a candidate index. The candidate index for the wide-based gait was the leg outward shift > 0.1 on the axial projection plane. In conclusion, the two-dimensional coordinates on the body axis projection planes calculated from the 3D relative coordinates estimated by the TDPT-GT application enabled the quantification of pathological gait features.

Development of Smartphone Application for Markerless Three-Dimensional Motion Capture Based on Deep Learning Model.

To quantitatively assess pathological gait, we developed a novel smartphone application for full-body human motion tracking in real time from markerless video-based images using a smartphone monocular camera and deep learning. As training data for deep learning, the original three-dimensional (3D) dataset comprising more than 1 million captured images from the 3D motion of 90 humanoid characters and the two-dimensional dataset of COCO 2017 were prepared. The 3D heatmap offset data consisting of 28 × 28 × 28 blocks with three red-green-blue colors at the 24 key points of the entire body motion were learned using the convolutional neural network, modified ResNet34. At each key point, the hottest spot deviating from the center of the cell was learned using the tanh function. Our new iOS application could detect the relative tri-axial coordinates of the 24 whole-body key points centered on the navel in real time without any markers for motion capture. By using the relative coordinates, the 3D angles of the neck, lumbar, bilateral hip, knee, and ankle joints were estimated. Any human motion could be quantitatively and easily assessed using a new smartphone application named Three-Dimensional Pose Tracker for Gait Test (TDPT-GT) without any body markers or multipoint cameras.

Performance Improvement of EEG-Based BCI Using Visual Feedback Based on Evaluation Scores Calculated by a Computer.

In the study of an electroencephalography (EEG)-based brain computer interface (BCI) using the P300, there have been many reports on computer algorithms that identify the target intended by a user from multiple candidates. However, because the P300 amplitude depends on the subject's condition and is attenuated by physical and mental factors, such as fatigue and motivation, the performance of the BCI is low. Therefore, we aim to improve performance by introducing a feedback mechanism that provides the user with an evaluation calculated by the computer during EEG measurement. In this study, we conducted an experiment in which the user input one character from four characters on the display. By changing the character size according to the evaluation score calculated by the computer, the computer's current evaluation was fed back to the user. This is expected to change the consciousness of the user to enable them to execute a task by knowing the evaluation; that is, if the evaluation is high, the user needs to maintain the current state, and if the evaluation is low, a behavioral change, such as increasing attention, is required to improve the evaluation.As a result of comparing 10 subjects with and without feedback, accuracy improved for seven subjects that were given feedback.

Reduction of the ERP Measurement Time by a Weighted Averaging Using Deep Learning.

In clinical examination, event-related potentials (ERPs) are estimated by averaging across multiple responses, which suppresses background EEG. However, acquiring the number of responses needed for this process is time consuming. We therefore propose a method for shortening the measurement time using weighted-average processing based on the output of deep learning. Using P300 as a representative component, here we focused on the shape of the ERP and evaluated whether our method emphasizes the P300 peak amplitude more than conventional averaging, while still maintaining the waveform shape and the P300 peak latency. Thus, using either CNN or EEGNet, the correlation coefficient reflecting the waveform shape, the peak P300 amplitude, and the peak latency were evaluated and compared with the same factors obtained from conventional waveform averaging. Additionally, the degree of background EEG suppression provided by our method was evaluated using the root mean square of the pre-stimulation waveform, and the number of fewer responses required for averaging (i.e., the reduction in measurement time) was calculated.The results showed that compared with P300 values obtained through conventional averaging, our method allowed for the same shape and response latency, but with a higher amplitude, while requiring a smaller number of responses. Our method showed that by using EEGNet, measurement time could be reduced by 13.7%. This corresponds to approximately a 40-second reduction for every 5 minutes of measurement time.

Motor Imagery Classification of Finger Motions Using Multiclass CSP.

Electroencephalogram (EEG) data during motor imagery tasks regarding small-scale physical dynamics such as finger motions have low discriminability because capturing the spatial difference of the motions is difficult. We assumed that more discriminative features can be captured if spatial filters maximize the independence of each class data. This study constructed spatial filters named multiclass common spatial pattern (CSP), which maximize an approximation of mutual in-formation of extracted components and class labels, and applied them to a five-class motor-imagery dataset containing finger motion tasks. By applying multiclass CSP, the classification accuracies were improved (Mean SD: 40.6 ± 10.1%) compared with classical CSP (21.8 ± 2.5%) and no spatial filtering case (38.7±10.0%). In addition, we visualized learned spatial filters to assess the trend of discriminative features of finger motions. For these results, it was clear that multiclass CSP captured task-specific spatial maps for each finger motion and outperformed multiclass motor-imagery classification performance about 2% even when the tasks are small-scale physical dynamics.

Longitudinal analysis of brain structure using existence probability.

INTRODUCTION: We propose a method to evaluate quantitatively the longitudinal structural changes in brain atrophy to provide early detection of Alzheimer's disease (AD) and mild cognitive impairment (MCI). METHODS: We used existence probabilities obtained by segmenting magnetic resonance (MR) images at two different time points into four regions: gray matter, white matter, cerebrospinal fluid, and background. This method was applied to T1-weighted MR images of 110 participants with normal cognition (NL), 165 with MCI, and 82 with AD, obtained from the Japanese Alzheimer's Disease Neuroimaging Initiative database. RESULTS: We obtained the coefficients of probability change (CPC) for each dataset. We found high area under the receiver operating characteristic curve (ROC) values (up to 0.908 of the difference of ROCs) for some CPC regions that are considered indicators of atrophy. Additionally, we attempted to establish a machine-learning algorithm to classify participants as NL or AD. The maximum accuracy was 92.1% for NL-AD classification and 81.2% for NL-MCI classification using CPC values between images acquired at first and sixth months, respectively. CONCLUSION: These results showed that the proposed method is effective for the early detection of AD and MCI.

Fast EEG spike detection via eigenvalue analysis and clustering of spatial amplitude distribution.

OBJECTIVE: In the current study, we tested a proposed method for fast spike detection in electroencephalography (EEG). APPROACH: We performed eigenvalue analysis in two-dimensional space spanned by gradients calculated from two neighboring samples to detect high-amplitude negative peaks. We extracted the spike candidates by imposing restrictions on parameters regarding spike shape and eigenvalues reflecting detection characteristics of individual medical doctors. We subsequently performed clustering, classifying detected peaks by considering the amplitude distribution at 19 scalp electrodes. Clusters with a small number of candidates were excluded. We then defined a score for eliminating spike candidates for which the pattern of detected electrodes differed from the overall pattern in a cluster. Spikes were detected by setting the score threshold. MAIN RESULTS: Based on visual inspection by a psychiatrist experienced in EEG, we evaluated the proposed method using two statistical measures of precision and recall with respect to detection performance. We found that precision and recall exhibited a trade-off relationship. The average recall value was 0.708 in eight subjects with the score threshold that maximized the F-measure, with 58.6 ± 36.2 spikes per subject. Under this condition, the average precision was 0.390, corresponding to a false positive rate 2.09 times higher than the true positive rate. Analysis of the required processing time revealed that, using a general-purpose computer, our method could be used to perform spike detection in 12.1% of the recording time. The process of narrowing down spike candidates based on shape occupied most of the processing time. SIGNIFICANCE: Although the average recall value was comparable with that of other studies, the proposed method significantly shortened the processing time.

Fast spike detection in EEG using eigenvalue analysis and clustering of spatial amplitude distribution.

In the current study, we tested a proposed method for fast spike detection using a general-purpose computer. First, we performed eigenvalue analysis using a gradient calculated from two neighboring samples to detect high-amplitude negative peaks. Clustering was performed to classify detected peaks by considering amplitude distribution at scalp electrodes. Negative peaks were scored by considering electrodes in the detection process and the cluster to which each peak belonged. Spikes were detected using two parameters: score threshold, and the number of clusters. We then used precision and recall to eliminate overestimation of the performance of the method. The results revealed a tradeoff between precision and recall. Recall showed a maximum average value of 0.90 in two subjects. In contrast, average precision was 0.21, and the false positive rate was almost four times higher than the true positive rate on the condition that 64 and 54 spikes were included in two subjects. Analysis of required processing time revealed that our method could complete spike detection in approximately one-eighth of the recording time.

Robust estimation of event-related potentials via particle filter.

BACKGROUND AND OBJECTIVE: In clinical examinations and brain-computer interface (BCI) research, a short electroencephalogram (EEG) measurement time is ideal. The use of event-related potentials (ERPs) relies on both estimation accuracy and processing time. We tested a particle filter that uses a large number of particles to construct a probability distribution. METHODS: We constructed a simple model for recording EEG comprising three components: ERPs approximated via a trend model, background waves constructed via an autoregressive model, and noise. We evaluated the performance of the particle filter based on mean squared error (MSE), P300 peak amplitude, and latency. We then compared our filter with the Kalman filter and a conventional simple averaging method. To confirm the efficacy of the filter, we used it to estimate ERP elicited by a P300 BCI speller. RESULTS: A 400-particle filter produced the best MSE. We found that the merit of the filter increased when the original waveform already had a low signal-to-noise ratio (SNR) (i.e., the power ratio between ERP and background EEG). We calculated the amount of averaging necessary after applying a particle filter that produced a result equivalent to that associated with conventional averaging, and determined that the particle filter yielded a maximum 42.8% reduction in measurement time. The particle filter performed better than both the Kalman filter and conventional averaging for a low SNR in terms of both MSE and P300 peak amplitude and latency. For EEG data produced by the P300 speller, we were able to use our filter to obtain ERP waveforms that were stable compared with averages produced by a conventional averaging method, irrespective of the amount of averaging. CONCLUSIONS: We confirmed that particle filters are efficacious in reducing the measurement time required during simulations with a low SNR. Additionally, particle filters can perform robust ERP estimation for EEG data produced via a P300 speller.

Preliminary study for extraction of P300 and SSVEP by stimulus presentation using phase inversion technique in hybrid BCI.

In this study, we propose a novel stimulation presentation method for the hybrid BCI of the P300 and steady state visual evoked potential (SSVEP) to separate the two components efficiently. The method produces the separation by generating the P300 at two time points whose phase difference is π radians in the SSVEP component corresponding to stimulus frequency. Assuming that the consecutive two P300 responses are identical and the SSVEP is sinusoidal, the P300 can be extracted as a summation of the above two responses by suppressing the SSVEP. Also, the SSVEP can be detected by the subtraction of the above two responses. Accordingly, this method is realized by a stimulus pair consisting of the above two stimuli. In an EEG experiment, we used a checkerboard stimulus and character presentation for obtaining the SSVEP and P300, respectively. The stimulus frequencies of the checkerboard were assigned to 5 Hz and 3 Hz to classify the target character from the two given characters. The results showed the appearance of a prominent P300 component from only one pair of stimuli, even though the fundamental and harmonic frequency components of the SSVEP for lower stimulus frequencies are not very stable. This is because of the asymmetry of the positive and negative potentials for the SSVEP. It is a good idea to use a stimulus frequency that overlaps with the P300 frequency band, because this method does not separate the P300 and SSVEP by EEG frequency difference. Moreover, it reduces the measurement time (i.e., it shortens the number of averagings required for P300 estimation) because the SSVEP cancels out if it is sinusoidal. We consider that this will be a useful method to estimate the P300 and SSVEP simultaneously from these aspects.

Character identification by maximizing the difference between target and non-target responses in EEG without sophisticated classifiers.

We propose a simple character identification method demonstrated by using an electroencephalogram (EEG) with a stimulus presentation technique. The method assigns a code maximizing the minimum Hamming distance between character codes. Character identification is achieved by increasing the difference between target and non-target responses without sophisticated classifiers such as neural network or support vector machine. Here, we introduce two kinds of scores reflecting the existence of the P300 component from the point of time and frequency domains. We then applied this method to character identification using a 3 × 3 matrix and compared the results to that of a conventional P300 speller. The accuracy of character identification with our method indicated a performance of 100% character identification from five subjects. In contrast, the correct character was detected in two subjects and a wrong one was detected for one subject. For the remaining two subjects, no character was detected within ten trials. Our method required 4.8 trials on average to detect the correct character.

Quantitative evaluation for the wakefulness state using complexity-based decision threshold value in EEG signals.

Fully awake state of the subjects tends to be an early drowsy state as a result from the prolonged time of electroencephalography (EEG) measurements. Such situations can complicate the interpretation of EEG signals and hence, the wakefulness of the subject should be considered in the inspection. Thus, in the present study, a new index for quantitative evaluation of the wakefulness (whether either early drowsy or fully awake) state of subjects by using a complexity-based decision threshold value was developed. The proposed index was based on approximate entropy (ApEn) to quantify the complexity metric, but with new parameter values by using a new systematic approach. This index was evaluated using occipital-alpha rhythm during eye closure for 45 healthy adult subjects for each one of two groups: fully awake and drowsy groups. Our index could show more superiority than other conventional spectral-based indices used for evaluating the wakefulness state of subjects including relative delta sub band power (R.δ), relative theta sub band power (R.θ), power ratio between theta and alpha (Pθ/α), and between theta and beta (Pθ/ß) over occipital lobe. Our index is superior than R.δ, R.θ, Pθ/α and Pθ/ß with 10%, 5.5%, 8.9% and 24.4% respectively.

EEG character identification using stimulus sequences designed to maximize mimimal hamming distance.

In this study, we have improved upon the P300 speller Brain-Computer Interface paradigm by introducing a new character encoding method. Our concept in detection of the intended character is not based on a classification of target and nontarget responses, but based on an identifaction of the character which maximize the difference between P300 amplitudes in target and nontarget stimuli. Each bit included in the code corresponds to flashing character, '1', and non-flashing, '0'. Here, the codes were constructed in order to maximize the minimum hamming distance between the characters. Electroencephalography was used to identify the characters using a waveform calculated by adding and subtracting the response of the target and non-target stimulus according the codes respectively. This stimulus presentation method was applied to a 3×3 character matrix, and the results were compared with that of a conventional P300 speller of the same size. Our method reduced the time until the correct character was obtained by 24%.

Imaging renal structures by X-ray phase-contrast microtomography.

X-ray crystal interferometer-based X-ray phase-contrast microtomography (phase-contrast microtomography) is able to image microstructures within soft tissue without the use of a contrast agent. Here we determined the feasibility of using this technique in the non-destructive inspection of formalin-fixed kidney tissue from certain hamsters that spontaneously develop mesangial thickening with focal and segmental glomerulosclerosis, and from age-matched Syrian hamsters. We used a triple Laue-case X-ray interferometer with a 40 microm-thick analyzer, a sample cell, and an X-ray charge-coupled-device camera with a 4.34 microm pixel size. Images of glomeruli and tubular structures were similar to those seen using 40-100 x magnification on an optical microscope. In samples from two female glomerulosclerotic hamsters, seven scattered lesions were detected. The wedge-shaped pathological lesions included mild atrophic tubular walls, markedly dilated tubular lumen, high-density glomeruli, and widening of Bowman's space. The microvasculature was distinctly visualized in the specimens without any contrast agents. Hence, phase-contrast microtomography can detect small scattered lesions in diseased kidney tissue and is a powerful auxiliary tool for pre-histological evaluations.

Quantitative evaluation of photic driving response for computer-aided diagnosis.

The aim of our research is the quantification of the photic driving response, a routine electroencephalogram (EEG) examination, for computer-aided diagnosis. It is well known that the EEG responds not only to the fundamental frequency but also to all sub and higher harmonics of a stimulus. In this study, we propose a method for detecting and evaluating responses in screening data for individuals. This method consists of two comparisons based on statistical tests. One is an intraindividual comparison between the EEG at rest and the photic stimulation (PS) response reflecting enhancement and suppression by PS, and the other is a comparison between data from an individual and a distribution of normals reflecting the position of the individual's data in the distribution of normals in the normal database. These tests were evaluated using the Z-value based on the Mann-Whitney U-test. We measured EEGs from 130 normal subjects and 30 patients with any of schizophrenia, dementia and epilepsy. Normal data were divided into two groups, the first consisting of 100 data for database construction and the second of 30 data for test data. Using our method, a prominent statistical peak of the Z-value was recognized even if the harmonics and alpha band overlapped. Moreover, we found a statistical difference between patients and the normal database at diagnostically helpful frequencies such as subharmonics, the fundamental wave, higher harmonics and the alpha frequency band.

Quantitative evaluation of myocardial function by a volume-normalized map generated from relative blood flow.

Our study aimed to quantitatively evaluate blood flow in the left ventricle (LV) of apical hypertrophic cardiomyopathy (APH) by combining wall thickness obtained from cardiac magnetic resonance imaging (MRI) and myocardial perfusion from single-photon emission computed tomography (SPECT). In this study, we considered paired MRI and myocardial perfusion SPECT from ten patients with APH and ten normals. Myocardial walls were detected using a level set method, and blood flow per unit myocardial volume was calculated using 3D surface-based registration between the MRI and SPECT images. We defined relative blood flow based on the maximum in the whole myocardial region. Accuracies of wall detection and registration were around 2.50 mm and 2.95 mm, respectively. We finally created a bull's-eye map to evaluate wall thickness, blood flow (cardiac perfusion) and blood flow per unit myocardial volume. In patients with APH, their wall thicknesses were over 10 mm. Decreased blood flow per unit myocardial volume was detected in the cardiac apex by calculation using wall thickness from MRI and blood flow from SPECT. The relative unit blood flow of the APH group was 1/7 times that of the normals in the apex. This normalization by myocardial volume distinguishes cases of APH whose SPECT images resemble the distributions of normal cases.

Fundamental study on subharmonic imaging by irradiation of amplitude-modulated ultrasound waves.

The second harmonic and subharmonic components, the frequencies of which are twice and one half the fundamental frequency, are included in echoes from contrast agents. An imaging method, which employs a second harmonic (second harmonic imaging), is widely used in medical diagnoses. On the other hand, subharmonic is expected to provide a higher contrast between biological tissues and blood flow because echo signals are generated only from blood containing the contrast agents. However, the subharmonic component echo signal power from contrast agents is relatively low. This has resulted in little progress in the field of subharmonic imaging. In this study, a new imaging method is proposed using amplitude-modulated waves as transmitted waves combined with the pulse inversion method to enhance subharmonic echo signals. Two optimal frequencies are set, including the modulated waves, F(1) and F(2), so that the subharmonic frequency of F(1) and the second harmonic frequency of F(2) may result in the same value. This allows a more powerful signal at the frequency band because the second harmonic and subharmonic components are integrated. Furthermore, a B-mode ultrasound image of an agar phantom that imitated biological tissue and showed the effectiveness of our method was reconstructed. As a result, the echo power of the subharmonic component was enhanced by approximately 11.8 dB more than the conventional method and the signal to noise ratio showed an improvement of 7.6 dB.

Estimation of temporary change of activation areas by moving an analysis time window in fMRI measurement.

In this paper, we propose a method to acquire temporal changes of activations by moving an analysis time window. An advantage of this method is that it can acquire rough changes of activated areas even with the data having low time resolution. We ascertained that activations from our method do not contradict previous reports on the oddball paradigm, thus showing its effectiveness. Eight normal subjects participated in the study, which consisted of a random series of 30 target and 70 nontarget stimuli. We investigated the activated area in three kinds of analysis time sections, from stimulus onset to 5 s after the stimulus (time section A), from 2 to 7 s after (B) and from 4 to 9 s after (C). In time section A, representative activated areas were regions including the left and supplementary motor areas (SMA), and cerebellum. In B, regions including the left motor area and SMA, right parahippocampal gyrus (Broadmann Area (BA) 30), right limbic lobe and cerebellum were activated. In C, bilaterally postcentral gyrus (BA 3,40), right anterior cingulate (ACC, BA 32), left middle frontal gyrus (BA 9) and right parahippocampal gyrus were activated. Most activations were consistent with previous studies.