A Machine Learning Enabled Wireless Intracranial Brain Deformation Sensing System.

A leading cause of traumatic brain injury (TBI) is intracranial brain deformation due to mechanical impact. This deformation is viscoelastic and differs from a traditional rigid transformation. In this paper, we describe a machine learning enabled wireless sensing system that predicts the trajectory of intracranial brain deformation. The sensing system consists of an implantable soft magnet and an external magnetic sensor array with a sensing volume of 12 × 12 × 4 mm3. Machine learning algorithm predicts the brain deformation by interpreting the magnetic sensor outputs created by the change in position of the implanted soft magnet. Three different machine learning models were trained on calibration data: (1) random forests, (2) k-nearest neighbors, and (3) a multi-layer perceptron-based neural network. These models were validated using both in vitro (a needle inserted into PVC gel) and in vivo (blast exposure to live and dead rat brains) experiments. The in vitro gel deformation predicted by these machine learning models showed excellent agreement with the camera measurements and had absolute error = 138 µm, Fréchet distance = 372 µm with normalized Procrustes disparity = 0.034. The in vivo brain deformation predicted by these models had absolute error = 50 µm, Fréchet distance = 95 µm with normalized Procrustes disparity = 0.055 for dead animal and absolute error = 125 µm, Fréchet distance = 289 µm with normalized Procrustes disparity = 0.2 for live animal respectively. These results suggest that the proposed machine learning enabled sensor system can be an effective tool for measuring in situ brain deformation.

AN ANALYSIS OF TWO COMMON REFERENCE POINTS FOR EEGS.

Clinical electroencephalographic (EEG) data varies significantly depending on a number of operational conditions (e.g., the type and placement of electrodes, the type of electrical grounding used). This investigation explores the statistical differences present in two different referential montages: Linked Ear (LE) and Averaged Reference (AR). Each of these accounts for approximately 45% of the data in the TUH EEG Corpus. In this study, we explore the impact this variability has on machine learning performance. We compare the statistical properties of features generated using these two montages, and explore the impact of performance on our standard Hidden Markov Model (HMM) based classification system. We show that a system trained on LE data significantly outperforms one trained only on AR data (77.2% vs. 61.4%). We also demonstrate that performance of a system trained on both data sets is somewhat compromised (71.4% vs. 77.2%). A statistical analysis of the data suggests that mean, variance and channel normalization should be considered. However, cepstral mean subtraction failed to produce an improvement in performance, suggesting that the impact of these statistical differences is subtler.

SEMI-AUTOMATED ANNOTATION OF SIGNAL EVENTS IN CLINICAL EEG DATA.

To be effective, state of the art machine learning technology needs large amounts of annotated data. There are numerous compelling applications in healthcare that can benefit from high performance automated decision support systems provided by deep learning technology, but they lack the comprehensive data resources required to apply sophisticated machine learning models. Further, for economic reasons, it is very difficult to justify the creation of large annotated corpora for these applications. Hence, automated annotation techniques become increasingly important. In this study, we investigated the effectiveness of using an active learning algorithm to automatically annotate a large EEG corpus. The algorithm is designed to annotate six types of EEG events. Two model training schemes, namely threshold-based and volume-based, are evaluated. In the threshold-based scheme the threshold of confidence scores is optimized in the initial training iteration, whereas for the volume-based scheme only a certain amount of data is preserved after each iteration. Recognition performance is improved 2% absolute and the system is capable of automatically annotating previously unlabeled data. Given that the interpretation of clinical EEG data is an exceedingly difficult task, this study provides some evidence that the proposed method is a viable alternative to expensive manual annotation.

Automated Identification of Abnormal Adult EEGs.

The interpretation of electroencephalograms (EEGs) is a process that is still dependent on the subjective analysis of the examiners. Though interrater agreement on critical events such as seizures is high, it is much lower on subtler events (e.g., when there are benign variants). The process used by an expert to interpret an EEG is quite subjective and hard to replicate by machine. The performance of machine learning technology is far from human performance. We have been developing an interpretation system, AutoEEG, with a goal of exceeding human performance on this task. In this work, we are focusing on one of the early decisions made in this process - whether an EEG is normal or abnormal. We explore two baseline classification algorithms: k-Nearest Neighbor (kNN) and Random Forest Ensemble Learning (RF). A subset of the TUH EEG Corpus was used to evaluate performance. Principal Components Analysis (PCA) was used to reduce the dimensionality of the data. kNN achieved a 41.8% detection error rate while RF achieved an error rate of 31.7%. These error rates are significantly lower than those obtained by random guessing based on priors (49.5%). The majority of the errors were related to misclassification of normal EEGs.

Improved EEG Event Classification Using Differential Energy.

Feature extraction for automatic classification of EEG signals typically relies on time frequency representations of the signal. Techniques such as cepstral-based filter banks or wavelets are popular analysis techniques in many signal processing applications including EEG classification. In this paper, we present a comparison of a variety of approaches to estimating and postprocessing features. To further aid in discrimination of periodic signals from aperiodic signals, we add a differential energy term. We evaluate our approaches on the TUH EEG Corpus, which is the largest publicly available EEG corpus and an exceedingly challenging task due to the clinical nature of the data. We demonstrate that a variant of a standard filter bank-based approach, coupled with first and second derivatives, provides a substantial reduction in the overall error rate. The combination of differential energy and derivatives produces a 24% absolute reduction in the error rate and improves our ability to discriminate between signal events and background noise. This relatively simple approach proves to be comparable to other popular feature extraction approaches such as wavelets, but is much more computationally efficient.

Obsessive and compulsive symptoms in schizophrenia: clinical and neurocognitive correlates.

Although research suggests that the presence of obsessive and compulsive symptoms in schizophrenia is associated with graver levels of psychosocial dysfunction, it is unclear whether it is also related to clinical features of illness. Accordingly, the present study compared the symptom levels and neurocognitive function of participants with schizophrenia who had and did not have significant obsessive or compulsive symptoms. Analyses of variance revealed that participants with significant levels of either obsessive or compulsive symptoms (N = 21) had higher levels of positive and emotional discomfort symptoms on the Positive and Negative Syndrome Scale (PANSS) and performed more poorly on the Wisconsin Card Sorting Test, a measure of executive function, than participants without obsessions or compulsions (N = 25). ANCOVAs controlling for level of obsessions also revealed that participants with significant levels of compulsions (N = 12) in particular had higher levels of negative and positive symptoms on the PANSS than participants without compulsions (N = 34). The impact of obsessive-compulsive phenomena on the course and outcome of schizophrenia is discussed.

Transferring clients from intensive case management: impact on client functioning.

The effects of transferring clients from assertive community treatment to a less intensive (step-down) case management program were examined. Service use decreased significantly after transfer to the step-down program, and no negative effects of transfer on hospital use or client functioning were evident. Critical elements for successful step-down are suggested and discussed.

The hand-held, voice-controlled terminal: medical information interface of the future.

In its push to cross the barrier of fixed-site, fixed-format information displays, Texas Instruments is busy working on prototypes of a small device that healthcare knowledge workers can carry with them onsite or offsite to enter orders, display test results, view images, listen to audio comments and record other patient data. TI won't specify a delivery date, and they say it's still several years away. Vision must precede form.

Intraoperative coagulation changes in children undergoing liver transplantation.

Intraoperative changes in blood coagulation were observed in eight children undergoing liver transplantation using a simplified coagulation profile (prothrombin time [PT], activated partial thromboplastin time [aPTT], and platelet count) and thrombelastography. Preoperatively, PT and aPTT were moderately prolonged (1.5 times control), and platelet count was greater than 100,000/mm3 in all patients but one (91,000/mm3). During the preanhepatic and anhepatic stages, PT, aPTT, reaction time, and coagulation time improved toward normal values, but platelet count and maximum amplitude did not change. Significant changes in coagulation occurred on reperfusion of the grafted liver: PT, aPTT, reaction time, and coagulation time were prolonged, and platelet count, maximum amplitude, and clot formation rate decreased. A heparin effect, which did not require treatment, was seen on reperfusion in four patients. Fibrinolysis occurred during the operation in five patients and was treated with Epsilon-aminocaproic acid (EACA) in one. Blood coagulation improved slowly, and values were close to baseline 90 min after reperfusion. In general, the coagulation changes seen in these children are similar to those in adults but less severe, possibly because of the preponderance of cholestatic disease in children compared with the more common hepatocellular disease in adults.

Gap junction uncoupling and discontinuous propagation in the heart. A comparison of experimental data with computer simulations.

The effects of octanol on longitudinal propagation in guinea pig papillary muscles were measured by intracellular microelectrodes. These data were compared with alterations in conduction induced by stepwise removal of gap junction channels in computer simulations of propagation based on a discontinuous cable model. Octanol reduced the velocity (theta) of propagating action potentials (APs) from 53.2 +/- 3.5 to less than 6.6 +/- 2.1 cm/s before block occurred. The maximal rate of rise (Vmax) changed in a biphasic manner, increasing from 133.1 +/- 5.4 in controls to 201.7 +/- 11.0 V/s when theta was 20.5 +/- 2.8 cm/s, and then declining to less than 58.6 +/- 15.2 V/s just before block. The input resistance and time constant of the AP foot increased, and the ascending limb of phase-plane loops became increasingly nonlinear and notched during octanol treatment. All effects of octanol reversed upon washout. A strand of cardiac tissue was modeled as a discontinuous cable composed of 40 cells, each with 10 isopotential membrane segments described by Beeler-Reuter kinetics, and coupled by a variable number of gap junction channels (156 pS). Decreasing the number of channels from 40,000 to 400 to 60 slowed conduction from 62.6 to 16.4 to 3.1 cm/s. As noted in the experimental data, Vmax increased from 103 to 130 and then fell to less than 96 V/s. The AP foot increased and became nonexponential. Distinct notches developed during phase 1 of the APs at slower propagation velocities in the experiments and simulations. The close similarities between the experimental and theoretical data obtained in this study supports the applicability of a discontinuous cable model for describing longitudinal propagation in the heart.

Radiation therapy for squamous cell carcinoma of the esophagus.

Between 1965 and 1981, 119 patients with squamous cell carcinoma of the esophagus were treated with radiation therapy with curative intent. Radiation was employed in combination with surgery and delivered pre- and/or postoperatively in 20 patients (17%). The remainder received radiotherapy alone. The overall survival rate was statistically higher in patients who had surgery and radiation compared to the group receiving radiation alone. The one-, two-, and five-year survival rates of patients receiving combined treatment vs radiotherapy alone were 65% vs 35%, 25% vs 14%, and 15% vs 6%. Age, total radiation dose, and inclusion of the supraclavicular areas in the radiation portals did not impact on outcome. Other prognostic factors are discussed. Long term survivors were noted to be at substantial risk for the development of a second epidermoid malignancy in the upper aerodigestive tract. Cumulative risk at five years was approximately 25%.

Propagation through electrically coupled cells. Effects of regional changes in membrane properties.

The normal process of excitation of the heart involves propagation of action potentials through cardiac regions of different anatomy and different intrinsic membrane properties. Although our understanding of these properties is still incomplete, it is well accepted that the parameters measured from a single cell penetration in an electrical syncytium (e.g., action potential duration, rate of rise, and velocity) reflect not only the properties of that cell but also the electrotonic interactions with other cells to which the recorded cell is electrically coupled. We have used simulation techniques to predict the spatial distribution of action potential parameters resulting from discretely localized alterations in the intrinsic membrane properties of some of the cells of an electrical syncytium. We have shown that the resulting spatial distribution is markedly different for alterations in plateau and pacemaker currents vs. rising phase currents, and that other factors, such as the site of stimulation and the underlying spatial pattern of cell-cell coupling resistance, also modify the spatial distribution of action potential properties resulting from a discrete regional change in intrinsic membrane properties.

Lytic effects of normal serum on isolated postonchospheral and metacestode stages of Taenia taeniaeformis.

Postonchospheral stages of Taenia taeniaeformis liberated from rat livers by enzymatic digestion at 1 to 10 days postinfection (DPI) and metacestodes dissected from infected livers at 22, 34, and 69 DPI were exposed in vitro to immune rat serum (IRS) and to normal serum from rats (NRS), human beings (NHS), or guinea pigs (NGS). The onset of rapid and destructive tegumental changes in all organisms exposed to any of the sera was demonstrated to be complement-dependent because the reaction was: (a) inhibited by treatment of serum at 56 C for 30 min; (b) inhibited by prior incubation of serum with zymosan or with complement-fixing, soluble products derived from larvae of T. taeniaeformis maintained in vitro (IVP); and (c) abolished by the addition of EDTA. Lytic effects occurred on exposure to agammaglobulinemic sheep serum, and lysis in the presence of IRS and NRS was shown to result in consumption of available hemolytic complement. Surface changes consisted of vesiculation in the microvillar or microthrix layers followed by sloughing of the tegument, eventually leading to collapse of the cystic bladder and cessation of flame cell activity, or, in the case of early postonchospheral forms, complete disintegration of the organism. When IVP was added to NHS, reduction of hemolytic complement activity was associated with the electrophoretic conversion of C3, and Factor B, but there was little or no consumption of C1. The observations support the hypothesis that complement-mediated effector mechanisms must be counteracted to ensure survival of parasites in vivo, and that the capacity for release of soluble nonspecific complement-fixing factors by taeniid larvae may have an important role to play in this process.