Classification of the Epileptic Seizure Onset Zone Based on Partial Annotation.

Epilepsy is a chronic disorder caused by excessive electrical discharges. Currently, clinical experts identify the seizure onset zone (SOZ) channel through visual judgment based on long-time intracranial electroencephalogram (iEEG), which is a very time-consuming, difficult and experience-based task. Therefore, there is a need for high-accuracy diagnostic aids to reduce the workload of clinical experts. In this article, we propose a method in which, the iEEG is split into the 20-s segment and for each patient, we ask clinical experts to label a part of the data, which is used to train a model and classify the remaining iEEG data. In recent years, machine learning methods have been successfully applied to solve some medical problems. Filtering, entropy and short-time Fourier transform (STFT) are used for extracting features. We compare them to wavelet transform (WT), empirical mode decomposition (EMD) and other traditional methods with the aim of obtaining the best possible discriminating features. Finally, we look for their medical interpretation, which is important for clinical experts. We achieve high-performance results for SOZ and non-SOZ data classification by using the labeled iEEG data and support vector machine (SVM), fully connected neural network (FCNN) and convolutional neural network (CNN) as classification models. In addition, we introduce the positive unlabeled (PU) learning to further reduce the workload of clinical experts. By using PU learning, we can learn a binary classifier with a small amount of labeled data and a large amount of unlabeled data. This can greatly reduce the amount and difficulty of annotation work by clinical experts. All together, we show that using 105 minutes of labeled data we achieve a classification result of 91.46% on average for multiple patients.

Epileptic seizure detection by using interpretable machine learning models.

Objective.Accurate detection of epileptic seizures using electroencephalogram (EEG) data is essential for epilepsy diagnosis, but the visual diagnostic process for clinical experts is a time-consuming task. To improve efficiency, some seizure detection methods have been proposed. Regardless of traditional or machine learning methods, the results identify only seizures and non-seizures. Our goal is not only to detect seizures but also to explain the basis for detection and provide reference information to clinical experts.Approach.In this study, we follow the visual diagnosis mechanism used by clinical experts that directly processes plotted EEG image data and apply some commonly used models of LeNet, VGG, deep residual network (ResNet), and vision transformer (ViT) to the EEG image classification task. Before using these models, we propose a data augmentation method using random channel ordering (RCO), which adjusts the channel order to generate new images. The Gradient-weighted class activation mapping (Grad-CAM) and attention layer methods are used to interpret the models.Main results.The RCO method can balance the dataset in seizure and non-seizure classes. The models achieved good performance in the seizure detection task. Moreover, the Grad-CAM and attention layer methods explained the detection basis of the model very well and calculate a value that measures the seizure degree.Significance.Processing EEG data in the form of images can flexibility to use a variety of machine learning models. The imbalance problem that exists widely in clinical practice is well solved by the RCO method. Since the method follows the visual diagnosis mechanism of clinical experts, the model interpretation results can be presented to clinical experts intuitively, and the quantitative information provided by the model is also a good diagnostic reference.

Satelight: self-attention-based model for epileptic spike detection from multi-electrode EEG.

Objective.Because of the lack of highly skilled experts, automated technologies that support electroencephalogram (EEG)-based in epilepsy diagnosis are advancing. Deep convolutional neural network-based models have been used successfully for detecting epileptic spikes, one of the biomarkers, from EEG. However, a sizeable number of supervised EEG records are required for training.Approach.This study introduces the Satelight model, which uses the self-attention (SA) mechanism. The model was trained using a clinical EEG dataset labeled by five specialists, including 16 008 epileptic spikes and 15 478 artifacts from 50 children. The SA mechanism is expected to reduce the number of parameters and efficiently extract features from a small amount of EEG data. To validate the effectiveness, we compared various spike detection approaches with the clinical EEG data.Main results.The experimental results showed that the proposed method detected epileptic spikes more effectively than other models (accuracy = 0.876 and false positive rate = 0.133).Significance.The proposed model had only one-tenth the number of parameters as the other effective model, despite having such a high detection performance. Further exploration of the hidden parameters revealed that the model automatically attended to the EEG's characteristic waveform locations of interest.

Spontaneous and reflex movements after diagnosis of clinical brain death: A lesson from acute encephalopathy.

BACKGROUND: Organ transplantation after brain death (BD) of the donor has been promoted in many countries as an established medical treatment. However, some problems with brain-dead organ transplantation have been reported. For example, there is no evidence as to the optimal observation period for a diagnosis and no evidence to support the interpretation of the various body movements observed after the determination of BD. CASE REPORT: A previously healthy 17-month-old girl with severe febrile convulsive status was transferred to our intensive care unit. The convulsions were refractory and the patient required respiratory management due to whole brain edema on head CT. Later she was diagnosed with acute encephalopathy. The patient showed a flat EEG, no responses on auditory brainstem responses (ABR), and loss of brainstem reflexes on repeated daily examinations. No apnea test was performed. Based on the diagnosis of clinical BD, coordinator of Japan Organ Transplant Network explained about organ donation on the 17th day of the disease. Subsequently, the family responded that they could not consent to organ donation, and the patient did not proceed to the legal BD determination. Around five weeks after the onset, spontaneous body movements began to appear, as not only the spinal reflexes but also the brainstem involvement. CONCLUSION: The pathophysiology of acute encephalopathy is largely unknown, and it is difficult to determine the observation period necessary for BD determination. What we have learned from this case is that clinical BD remains ambiguous and cannot be confirmed even with a thorough neurological examination, EEG, and ABR.

Epileptic Spike Detection Using Neural Networks With Linear-Phase Convolutions.

To cope with the lack of highly skilled professionals, machine learning with proper signal processing is key for establishing automated diagnostic-aid technologies with which to conduct epileptic electroencephalogram (EEG) testing. In particular, frequency filtering with the appropriate passbands is essential for enhancing the biomarkers-such as epileptic spike waves-that are noted in the EEG. This paper introduces a novel class of neural networks (NNs) that have a bank of linear-phase finite impulse response filters at the first layer as a preprocessor that can behave as bandpass filters that extract biomarkers without destroying waveforms because of a linear-phase condition. Besides, the parameters of the filters are also data-driven. The proposed NNs were trained with a large amount of clinical EEG data, including 15 833 epileptic spike waveforms recorded from 50 patients, and their labels were annotated by specialists. In the experiments, we compared three scenarios for the first layer: no preprocessing, discrete wavelet transform, and the proposed data-driven filters. The experimental results show that the trained data-driven filter bank with supervised learning behaves like multiple bandpass filters. In particular, the trained filter passed a frequency band of approximately 10-30 Hz. Moreover, the proposed method detected epileptic spikes, with the area under the receiver operating characteristic curve of 0.967 in the mean of 50 intersubject validations.

Training Physical and Geometrical Mid-Points for Multi-person Pose Estimation and Human Detection Under Congestion and Low Resolution.

This paper introduces the design and evaluation of NeoPose which is developed for multi-person pose estimation and human detection. The design of NeoPose is targeting the issue of human detection under congested situation and with low resolution in the image. Under such situations, we compared the performance of different versions of NeoPose as well as other existing algorithms in a human detection task. Throughout the task, the usefulness of two kinds of mid-point (physical and geometrical mid-points) and a deconvolution structure was discussed. Experiment results indicated that NeoPose which applied geometrical mid-points and deconvolution structure performed the best in terms of both precision and recall in the evaluation.

Association between the frequency of bedwetting and late preterm birth in children aged ≥5 years.

AIM: We examined the associations between late preterm (LPT) birth children aged ≥5 years and the frequency of bedwetting. Moreover, those who were born full-term/low birthweight (BW), LPT/low BW, LPT/normal BW and LPT/low BW were compared. METHODS: In total, we evaluated 614 patients who underwent assessments for frequent bedwetting at the three hospitals from January 2014 to December 2016. Data at the initial visit were collected from the electronic medical records. We assessed the patients' bladder diaries and questionnaires containing detailed information on demographics and frequency of bedwetting per month. Neonatal data were collected from the Maternal and Child Health Handbook. RESULTS: Frequency of bedwetting in the LPT/low BW group was higher than in the term/low BW group (28 vs. 22.5, p < 0.05). However, the frequency between the LPT/normal BW group and the LPT/low BW group was not significantly different (28 vs. 28, p = 1.00). Multiple regression analyses were conducted to eliminate potential confounding factors, attention-deficit/hyperactivity disorder and intellectual disability, but results were not changed. CONCLUSION: This study revealed that LPT/low BW was associated with increased frequency of bedwetting in children. The results suggest that gestational age should be considered when examining patients with severe bedwetting.

Fluorescent detection for cyclic and acyclic alcohol guests by naphthalene-appended amino-beta-cyclodextrins.

Spectroscopic and thermodynamic investigations on complexation of naphthalene-appended amino-beta-cyclodextrins 1 and 2 with cyclic alcohols (cyclohexanol and cycloheptanol) and acyclic alcohols (1-pentanol, 2-pentanol, 1-hexanol, and 1-heptanol) have been carried out. Host 1 exhibits a drastic fluorescent enhanced-signal change in the presence of alcohol guests in aqueous solution. [structure: see text]

Facile formation of water-insoluble cyclodextrin/Nafion composite films on solid surfaces.

The water-insoluble poly-beta-cyclodextrin (poly-CD)/Nafion composite film was easily prepared by casting a mixed solution of poly-CD and Nafion onto substrate plates. FT-IR measurements showed that the 50/50 wt% poly-CD/Nafion composite film remained stable on the glass and quartz substrates after immersion in water for more than 3 h, while a pure poly-CD film was almost completely dissolved by immersion within 1 h. The film stability was also evaluated from the amount of p-nitrophenol (p-NP) inclusion in the film, which was determined from the decrease in UV-vis absorbance of the p-NP solution into which the film was immersed. The composition dependence of the inclusion amount showed that the film was stable up to 50 wt% CD, but became less stable with further increase in the CD concentration in the film. From the isotherms for the inclusion of p-NP and 1-naphthoate (1-Naph) into the film, the inclusion (stability) constants were determined to be 3.7x10(3) M(-1) and 1.9x10(2) M(-1), respectively. These results show that the selective inclusion of CD is retained after preparation of the composite film.

Angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers effectively and directly potentiate superoxide scavenging by polymorphonuclear leukocytes from patients with type 2 diabetes mellitus.

BACKGROUND: Angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin II type 1 receptor blockers (ARBs) have potent antioxidant effects in addition to antihypertensive effects. METHODS: We investigated the ability of ACEIs and ARBs to enhance the superoxide scavenging ability of polymorphonuclear leukocytes (PMNLs) from type 2 diabetic patients (n = 32) and healthy subjects (n = 32). The scavenging ability (U/10(3) cells) of superoxide was measured by electron spin resonance. We used ascorbic acid as a positive control antioxidant and tested captopril, temocapril (an inactive form of ACEI), and temocaprilate (an active form of ACEI) as ACEIs, as well as RNH-6270 as an ARB. RESULTS: Captopril, temocaprilate, and RNH-6270 showed dose-dependent enhancement in scavenging ability. The scavenging ability with captopril and temocaprilate was greater than with RNH-6270. The changes in scavenging ability induced by all of the drugs in diabetic patients were similar to the changes in healthy subjects. A high-glucose medium (400-800 mg/dL) greatly attenuated the drug-induced enhancement of scavenging ability. CONCLUSIONS: We demonstrated that both ACEIs and ARBs enhance superoxide scavenging by PMNLs from type 2 diabetic patients and that a high-glucose environment markedly attenuates the ability of these drugs to augment superoxide scavenging.

High serum pentosidine concentrations are associated with increased arterial stiffness and thickness in patients with type 2 diabetes.

Accumulation of advanced glycation end products in vessel walls may increase arterial stiffness and/or thickness, contributing to a high incidence of cardiovascular disease (CVD) in patients with diabetes. We investigated whether serum concentrations of pentosidine, a well-defined advanced glycation end product, are associated with arterial stiffness or thickness in patients with type 2 diabetes. Pentosidine was measured in sera from 98 patients with type 2 diabetes and 61 age-matched control subjects by a competitive enzyme-linked immunosorbent assay. Arterial stiffness was evaluated by heart-brachial and brachial-ankle pulse wave velocities (PWVs) measured using an automatic device. Arterial thickness was determined ultrasonographically as carotid intima-media wall thickness (IMT). Serum concentrations of pentosidine were significantly higher in patients with diabetes than in control subjects (64.4 +/- 21.0 vs 22.8 +/- 7.0 microg/L; P < .0001). In patients with diabetes, serum pentosidine correlated positively with heart-brachial PWV (r = 0.304; P < .01) but not with brachial-ankle PWV. Serum pentosidine also correlated positively with carotid IMT in patients with diabetes (r = 0.300; P < .01). Serum pentosidine concentrations were significantly higher in patients with diabetes with CVD than in those without (72.3 +/- 23.7 vs 62.3 +/- 19.8 microg/L; P = .0453). By multivariate analysis, only age (partial coefficient = 0.308; P < .05) and serum creatinine (partial coefficient = 0.328; P < .01) retained significant influence on serum pentosidine. After adjustment for renal function, carotid IMT still correlated positively with serum pentosidine (partial coefficient = 0.2736; P = .021). In conclusion, serum pentosidine was positively associated with both arterial stiffness and thickness and CVD in patients with type 2 diabetes.

Coagulation and inflammation in overt diabetic nephropathy: association with hyperhomocysteinemia.

BACKGROUND: Diabetic nephropathy, especially when advanced, is associated with high prevalence of atherosclerotic cardiovascular disease in which inflammation and coagulation may play pathogenic roles. We investigated the relationships between diabetic nephropathy and coagulation, fibrinolysis, or inflammation in patients with Type 2 diabetes. METHODS: We evaluated markers of inflammation and coagulation in 105 Type 2 diabetic patients with various grades of nephropathy and 49 healthy control subjects, in association with plasma total homocysteine (tHcy) measurements. RESULTS: Plasma tHcy concentrations were significantly higher in diabetic patients than in controls (8.96 +/- 3.04 vs. 6.92 +/- 1.36 micromol/l, P < 0.0001). Plasma concentrations of interleukin (IL)-6 were significantly higher in diabetic patients than in control subjects (P < 0.0001). In diabetic patients, plasma tHcy correlated positively with urinary albumin, fibrinogen, IL-6 and plasmin-alpha2-antiplasmin complex (PAP), while plasma tHcy correlated negatively with creatinine clearance (Ccr) and protein C activity. After adjustment for Ccr, IL-6 and protein C activity were significantly associated with plasma tHcy. Plasma tHcy concentrations were significantly higher in patients with overt albuminuria than in those with normoalbuminuria or microalbuminuria, as were plasma concentrations of fibrinogen, prothrombin F1+2, and interleukin-6. CONCLUSIONS: Diabetic nephropathy is associated with elevated markers for both coagulation and inflammation. High plasma homocysteine may be a link between diabetic nephropathy and both chronic inflammation and hypercoagulability, increasing cardiovascular risk.

Dissociation between urinary pyrraline and pentosidine concentrations in diabetic patients with advanced nephropathy.

It has been reported that the concentrations of both pyrraline and pentosidine, well-characterized advanced glycation end products, are increased in the urine of diabetic patients. To determine factors that influence the urinary excretion of pyrraline or pentosidine, we compared pyrraline or pentosidine concentrations with glycemic-control indexes, urinary albumin excretion, and urinary beta2-microglobulin in patients with type 2 diabetes. The study was conducted in 39 age-matched healthy control subjects and 50 diabetic patients, including 22 patients with normoalbuminuria, 15 with microalbuminuria, and 13 with macroalbuminuria. Both urinary pyrraline and pentosidine were measured in early-morning urine specimens with the use of high-pressure liquid chromatography. The urinary pentosidine concentration was significantly higher in diabetic patients than in control subjects (P <.01). In contrast, the urinary pyrraline concentration was significantly lower in diabetic patients than in control subjects (P <.001). Urinary pentosidine concentrations were greater in diabetic patients with macroalbuminuria and microalbuminuria than in those with normoalbuminuria. However, urinary pyrraline concentrations were significantly lower in diabetic patients with advanced nephropathy. Both the hemoglobin A(1c) (HbA(1c)) and the preceding year's mean HbA(1c) were lower in patients with macroalbuminuria than in those with normoalbuminuria or microalbuminuria. Urinary pyrraline, but not pentosidine, showed a significantly positive correlation with the preceding year's mean HbA(1c) (P<0.01). Multivariate analysis disclosed that urinary beta-2-microglobulin was independently correlated with the urinary concentrations of pentosidine and pyrraline (P <.05 for both). We conclude that the urinary concentration of pentosidine is greater in diabetic patients with overt nephropathy, whereas the urinary pyrraline concentration is significantly lower in diabetic patients with overt nephropathy. Because urinary pyrraline is more directly influenced by glycemia than by pentosidine, the difference in glycemic control among diabetic patients with various grades of nephropathy may be responsible for a dissociation between urinary pyrraline and pentosidine concentrations in patients with overt diabetic nephropathy.

Results of blood inflammatory markers are associated more strongly with toe-brachial index than with ankle-brachial index in patients with type 2 diabetes.

OBJECTIVE: Three blood markers of inflammation (high-sensitivity C-reactive protein [hsCRP], interleukin [IL]-6, and fibrinogen) were compared with markers of atherosclerotic cardiovascular disease (CVD) (history of stroke or cardiac ischemia and measured toe-brachial index [TBI]) to determine whether inflammatory markers are associated with atherosclerosis in patients with type 2 diabetes. RESEARCH DESIGN AND METHODS: Of 103 patients with type 2 diabetes, 26 had CVD. TBI was plethysmographically determined in both great toes. Serum hsCRP was immunonephelometrically determined. Plasma IL-6 was measured by an enzyme immunoassay. RESULTS: Both ABI and TBI were lower in diabetic patients with CVD than in those without CVD (1.05 +/- 0.19 vs. 1.14 +/- 0.09, P < 0.05, and 0.75 +/- 0.20 vs. 0.95 +/- 0.21, P < 0.001, respectively). By linear regression, right TBI but not right ABI showed a significant negative correlation with serum hsCRP (r = -0.372, P < 0.01) and plasma fibrinogen (r = -0.224, P < 0.05). Serum hsCRP was also negatively correlated with lower TBI, but not lower ABI. We found no significant correlation between plasma IL-6 and ABI or TBI. CONCLUSIONS: TBI was strongly associated with CVD, serum hsCRP, and plasma fibrinogen. Of these inflammatory markers, serum hsCRP may be the most promising marker for vascular inflammation.

Myotonic dystrophy associated with variable circadian rhythms of serum cortisol and isolated thyrotropin deficiency.

We describe a case of myotonic dystrophy presenting with a disturbed circadian rhythm of the serum cortisol and an isolated thyrotropin deficiency. The diagnosis of myotonic dystrophy was based on clinical characteristics, positive electromyographic findings, and increased number of CTG repeats in the dystrophia myotonica protein kinase (DMPK) gene. The patient presented with a variable circadian rhythm of the serum cortisol, increased excretion of urinary free cortisol, and a high adrenocorticotropin hormone responses to corticotropin-releasing hormone. The basal serum thyrotropin concentration was low and did not increase after thyrotropin-releasing hormone stimulation. The protein encoded by the DMPK gene may act as a second messenger in signal transduction, like a protein kinase. The present patient had a diverse pattern of disturbances in the hypothalamus-pituitary-endocrine organ axis, probably mediated by differences in the action or expression of the gene products in each endocrine cell.