Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference最新文献

Immunotherapies have been proven to have significant therapeutic efficacy in the treatment of cancer. The last decade has seen adoptive cell therapies, such as chimeric antigen receptor T-cell (CART-cell) therapy, gain FDA approval against specific cancers. Additionally, there are numerous clinical trials ongoing investigating additional designs and targets. Nevertheless, despite the excitement and promising potential of CART-cell therapy, response rates to therapy vary greatly between studies, patients, and cancers. There remains an unmet need to develop computational frameworks that more accurately predict CART-cell function and clinical efficacy. Here we present a coarse-grained model simulated with logical rules that demonstrates the evolution of signaling signatures following the interaction between CART-cells and tumor cells and allows for in silico based prediction of CART-cell functionality prior to experimentation.Clinical Relevance- Analysis of CART-cell signaling signatures can inform future CAR receptor design and combination therapy approaches aimed at improving therapy response.

Cochlear implants (CI) have restored hearing to many deaf patients. It is the most successful neuroprosthetic in the field. However, in past decades technical improvements have plateaued and the market has solidified among 3 manufacturers. Proprietary software, and know-how are some of the barriers to innovation and disruption in CIs. In this paper we propose an open data communication protocol for cochlear implants that supports multipolar stimulation, accommodates an expandable number of channels, and minimizes the transmission of redundant information. We also present a method for implementing multipolar stimulation in single supply stimulators with a bridge-type switch matrix through pulse-polarity modulation. This combines the advantages of lower voltage (lower power) operation with more targeted stimulation.Clinical Relevance- In addition to enabling the development of new tools for research and clinical deployment, the presented data communication protocol will promote clinical research in more advanced auditory coding strategies.

As technology advances and sensing devices improve, it is becoming more and more pertinent to ensure accurate positioning of these devices, especially within the human body. This task remains particularly difficult during manual, minimally invasive surgeries such as cystoscopies where only a monocular, endoscopic camera image is available and driven by hand. Tracking relies on optical localization methods, however, existing classical options do not function well in such a dynamic, non-rigid environment. This work builds on recent works using neural networks to learn a supervised depth estimation from synthetically generated images and, in a second training step, use adversarial training to then apply the network on real images. The improvements made to a synthetic cystoscopic environment are done in such a way to reduce the domain gap between the synthetic images and the real ones. Training with the proposed enhanced environment shows distinct improvements over previously published work when applied to real test images.

Brain-computer Interfaces (BCIs) interpret electroencephalography (EEG) signals and translate them into control commands for operating external devices. The motor imagery (MI) paradigm is popular in this context. Recent research has demonstrated that deep learning models, such as convolutional neural network (CNN) and long short-term memory (LSTM), are successful in a wide range of classification applications. This is because CNN has the property of spatial invariance, and LSTM can capture temporal associations among features. A combination of CNN and LSTM could enhance the classification performance of EEG signals due to the complementation of their strengths. Such a combination has been applied to MI classification based on EEG. However, most studies focused on either the upper limbs or treated both lower limbs as a single class, with only limited research performed on separate lower limbs. We, therefore, explored hybrid models (different combinations of CNN and LSTM) and evaluated them in the case of individual lower limbs. In addition, we classified multiple actions: MI, real movements and movement observations using four typical hybrid models and aimed to identify which model was the most suitable. The comparison results demonstrated that no model was significantly better than the others in terms of classification accuracy, but all of them were better than the chance level. Our study informs the possibility of the use of multiple actions in BCI systems and provides useful information for further research into the classification of separate lower limb actions.

Purpose - Clinicians and care workers currently cannot determine the final extent of necrosis once it begins. If the area is relatively small, it may heal like a normal wound, whereas re-operation may be the only option if it spreads. This paper represents work toward a predictive algorithm using full-thickness random rat skin flap photographs to determine whether the tissue will develop irretrievable necrosis.Methods - Using post-surgery images taken over a series of days of ischemic flaps, features were extracted, selected, and input into a classification algorithm to see if it could provide information on the future condition of the flaps. We split our data into two groups: flaps that underwent normal healing and slow healing. When consulting with a specialist, it was observed that the resulting dermal damage was not severe when a flap had ≤ 40% necrosis over its length on the final day. Three classifiers were implemented: K-Nearest Neighbor (KNN), Random Forest, and Support Vector Machine (SVM).Results - The trained KNN was able to correctly determine whether a flap developed a necrotic area larger or less than 40% of its length with an accuracy of 91% using only images one day post-surgery. Under leave-one-out cross-validation, the Random Forest and SVM achieved accuracies of 82% and 79.5%, respectively, using images spanning ten days post-surgery.Conclusion - We have shown that a classifier can accurately determine whether ischemic skin flaps will develop severe necrotic tissue.Clinical Relevance- An algorithm that assesses early timepoint images and predicts necrosis spread is not only a vital tool for patients and clinicians but is also an extremely important tool to accelerate research into necrosis reduction strategies that ultimately may find application for more life-threatening necrosis-related conditions.

Cardiac trabeculae are small samples of heart muscle tissue that can be dissected and studied in vitro to better understand the underlying physiology of cardiac muscle. However, instruments for such experimentation often (1) involve delicate mounting of the muscle, (2) constrain investigations to one muscle at a time and, thus, (3) cannot retain the muscle in the same experimental configuration for post-experimental assessment including imaging analysis. Here, we present a novel device that allows trabeculae to be secured by a visible-light photo-initiated hydrogel, manipulated via a force sensor, and stimulated while being imaged. We use our robust, accurate image registration techniques to measure cantilever and gel deformation during trabecula contraction and thereby provide a measure of trabecula force production during twitches. A variety of experiments can then be conducted, with the potential for the trabecula to be fixed in place using hydrogel for further post-experiment analysis, as well as longitudinal evaluation. The device has multiple wells making it amenable to high-throughput testing.Clinical Relevance- These methods may allow longitudinal and high-throughput studies of cardiac tissue samples in health and disease.

The generation of super resolution ultrasound images from the low-resolution (LR) brightness mode (B-mode) images acquired by the portable point of care ultrasound systems has been of sufficient interest in the recent past. With the advancements in deep learning, there have been numerous attempts in this direction. However, all the approaches have been concentrated on employing the direct image as the input to the neural network. In this work, a stationary wavelet (SWT) decomposition is employed to extract the features from the input LR image which is passed through a modified residual network and the learned features are combined using the inverse SWT to reconstruct the high resolution (HR) image at a 4× scale factor. The proposed approach when compared to the state-of-the art approaches, results in an improved high resolution reconstruction.Clinical relevance- The proposed approach will enable the generation of high-resolution images from portable ultrasound systems, allowing for easier interpretation and faster diagnostics in primary care settings.

Cardiorespiratory interaction is related to the heart rate variability (HRV) synchronized with respiration. These metrics help to comprehend the autonomic nervous system (ANS) functionality in cardiovascular mechanisms. In this work, we aim to study the HRV in healthy subjects aged 18-24 years during the breathing techniques based on deep breaths followed by apnoeas, developed by Wim Hof (WHM). The attributes of all participates have been treated as a group and therefore, separated by gender. A total of 11 intervals have been distinguished: starting of basal respiration (SRI = 1), controlled deep breaths (CDB = 3), long expiratory apnoea (LEA = 3), short inspiratory apnoea (SIA = 3) and ending with basal respiration again (FRI = 1). To strengthen the HRV knowledge extraction from these scenarios, time and frequency analysis is conducted. In general, breathing and apnoea intervals presented significant statistically differences (p < 0.05), heart rate (HR) mean between SRI and FRI (p < 0.001), RR variability of LEA intervals (p < 0.01), root mean square of RR intervals during CDB (p < 0.05), maximum high frequency (HF) peak amplitude between SRI and FRI (p = 0.016), and low frequency (LF) area for LEA intervals (p < 0.001). When performing the frequency analysis, it has been observed that the sympathetic nervous system (SNS) has a higher contribution in the apnoea intervals. In conclusion, the WHM method implementation seems to involve a decrease in the HR. Specific breathing techniques could help to control the body in different conditions.Clinical Relevance- The WHM seems to imply a decrease on HR. Furthermore, after the implementation of the WHM, women presented higher HRV.

Synthetic data generation has become increasingly popular with the increasing use of generative networks. Recently, Generative Adversarial Network (GAN) architectures have produced exceptional results in synthetic image generation. However, time series generation still needs to be studied. This paper proposes a Conditional GAN-based system to generate unique samples of non-REM sleep electroencephalographic (EEG) signals. The CGAN model had a 1-D Convolution Neural Network based architecture. The model was trained using real EEG from healthy controls. The trained model can generate an artificial 30-second epoch of non-REM sleep whose power spectrum is identical to that of a real sleep EEG.Clinical relevance- Sleep EEG simulation can be used to train and enhance the skillset of fellows and technicians in the sleep medicine field. Variations in EEG signals can be highly complex to model mathematically; however, here, we harness the power of deep learning, using generative models such as CGANs to train, model complex data distributions, and generate diverse and artificial but realistic EEG signals during non-REM sleep.

Arterial pulse wave separation analysis (WSA) requires simultaneously measured pressure and flow rate waveform from the same arterial site. Modelling approaches to flow rate waveforms offers a methodological and instrumentational advantage. However, current techniques are limited to the aortic site. For non-aortic sites such as carotid artery, modelling methods that were developed for aortic sites are not likely to capture the intrinsic differences in the carotid flow rate. In this work, a double-Rayleigh flow rate model for the carotid artery is developed to separate the forward and backward pressure waves using WSA (DRMWSA). The model parameters are optimally found based on characteristic features - obtained from the pressure waveform. The DRMWSA was validated using a database of 4374 virtual (healthy) subjects, and its performance was compared with actual flow rate based WSA (REFWSA) at the carotid artery. An RMSE < 2 mmHg were obtained for forward and backward pressure waveforms. The reflection quantification indices (ΔPF, ΔPB), (RM, RI) obtained from DRMWSA demonstrated strong and statistically significant correlation (r > 0.96, p < 0.001) and (r > 0.80, p < 0.001) respectively, with insignificant bias (p > 0.05), upon comparing with counterparts in REFWSA. A moderate correlation (r = 0.64, p < 0.001) was obtained for reflection wave transit time between both methods. The proposed method minimises the measurements required for WSA and has the potential to widen the vascular screening procedures incorporating carotid pulse wave dynamics.Clinical Relevance-This methodology quantifies arterial pressure wave reflections in terms of pressure augmentation and reflection transit time. The methodological advantage of using only a single waveform helps easy translation to technological solutions for clinical research.