Alzheimer's disease is an irreversible neuro degenerative disease, and associated declines in cognitive function have a significant effect on daily life. Methods such as mini-mental state examination and Montreal cognitive assessment, the existing diagnostic methods for Alzheimer's disease, take a long time and have poor objective assessment score accuracy. This paper presents a novel method that assesses the brain's cognitive state by utilizing eye-tracking technology. The experiment results confirmed that the number of microsaccade occurrences on the x and y axes gradually increased as aging progressed. In addition, it was confirmed that the number of microsaccade occurrences increased as cognitive function scores declined, even among seniors in the same age groups. Based on these results, we prove the correlation between aging and the brain's cognitive functions and verifies that declines in the brain's cognitive functions affect microsaccades.
In recent times, there have been numerous proposals for body extensions, such as the "Third Arm" and other augmented limbs. While these extended limbs effectively address physical limitations in multitasking, they impose a significant cognitive burden by necessitating attention to multiple positions. In this research, we have designed a virtual reality training program that introduces a manageable attentional load, progressively intensifying it to enhance human multitasking capabilities by extending effective attentional resources. The outcomes of our validation test revealed a noteworthy 10%-20% enhancement in test task performance before and after the training, indicating an increase in the number of manageable tasks and quicker task response times. Moreover, the training fostered more efficient attention switching, leading to a reduction in the unnecessary expenditure of attentional resources.
Clinical studies have shown that infant crying is a crucial signal containing physical and mental information, such as hunger and pain, which can provide valuable insights into infants' pathology and demand. However, existing studies either focused on infant crying detection or reasoning (needs/diseases), where the limited data and label types hinder the model's generalization to unknown infants and reasons. To this end, we propose a multi-task Infant Crying Detection and Reasoning (ICDR) model for both the tasks of cry detection and reasoning, which utilizes a shared extractor to extract the deep representations and incorporates two classifiers for different tasks. In this way, ICDR can augment data by mixing datasets from associated tasks and introducing inductive bias as the regularization term to mitigate overfitting. Extensive experiments were performed on four infant crying datasets, showing that ICDR outperforms its corresponding single-task model in both tasks, exhibiting a 0.31% and 6.12% improvement in F1-score for infant crying detection and reasoning when using FNN as the shared extractor, obtaining 0.11% and 1.38% improvement when using CNN as the shared extractor. These results demonstrate that multi-task learning can efficiently leverage data from related tasks to enhance the model's generalization for infant crying detection and reasoning.
The importance of early Alzheimer's Disease screening is becoming more apparent, given the fact that there is no way to revert the patient's status after the onset. However, the diagnostic procedure of Alzheimer's Disease involves a comprehensive analysis of cognitive tests, blood sampling, and imaging, which limits the screening of a large population in a short period. Preliminary works show that rich neurological and cardiovascular information is encoded in the patient's eye. Due to the relatively fast and easy procedure acquisition, early-stage screening of Alzheimer's Disease patients with eye images holds great promise. In this study, we employed a deep neural network as a framework to investigate the relationship between risk factors of Alzheimer's Disease and retinal structures. Our result shows that the model not only can predict several risk factors above the baseline but also can discover the relationship between the retinal structures and risk factors to provide insights into the retinal imaging biomarkers of Alzheimer's disease.
The hippocampus is a disease-prone area of the brain that can be used as an important biomarker for neurodegenerative diseases like Alzheimer's. In recent years, deep neural networks have been applied to segment the hippocampus. However, accurately segmenting the hippocampus using magnetic resonance imaging (MRI) remains a challenging task. To explore a more effective segmentation strategy, this study proposes a new model by integrating the Vision Transformer (ViT) architecture with the UNet++ architecture, which is validated by using manual tracing of the hippocampus performed by clinical experts. The proposed ViT-based model achieved a dice score of 0.885, surpassing similar models by 2.82% in the Dice coefficient score.
The cognitive decline caused by Alzheimer's disease (AD) is closely related to the structural changes in the hippocampus captured by structural magnetic resonance imaging (sMRI). However, current deep model research on the morphological analysis of hippocampus is mainly based on 2D MRI slices, lacking a comprehensive description of the 3D surface morphology and complex textures of the hippocampus. For this reason, we propose a two-stream multi features deep learning model that establishes a descriptive system for 3D spatial structure and morphological atrophy features on the triangular mesh of left and right hippocampus. First, we encode the triangular mesh data into the spatial structural features of the hippocampal surface. Second, considering the tubular structure of the hippocampus and the inhomogeneous morphological changes caused by AD, we introduce the thickness features and Heat Kernel Signature (HKS) features for the morphological atrophy features encoding. Third, we integrate the encoded features of adjacent faces from a macroscopic perspective into the discriminative morphological features induced by AD. Finally, driven by classification tasks, the deep learning model parameters and the discriminative features are continuously optimized, thereby improving the accuracy of AD diagnosis. Our method is evaluated based on the T 1 weighted sMRI baseline data of 269 Aβ+ AD and 437 Aβ-normal cognitively(NC) subjects collected from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The classification accuracy of this method for AD and NC subjects is 93.4%, the sensitivity and specificity are 92.4% and 93.8%, respectively, and the area under the ROC curve (AUC) is 98.3%.
Lumbar punctures are important and delicate procedures necessitating precise training tools for skill acquisition. This study introduces an advanced simulator with enhanced features including dynamic motor systems and an intuitive GUI. The results from motor accuracy tests indicate significant improvements in precision and reliability, crucial for replicating the nuanced conditions of lumbar puncture procedures. An average absolute angular deviation of 0.5 degrees was observed over 19 trials indicating a consistent performance in mimicking real life anatomical vertebral settings. These advancements set a new standard in medical simulation technology for lumbar punctures, aiming to better prepare healthcare professionals for real-world clinical scenarios. This sets the board for future integrations using augmented and virtual reality paired with motorized systems to facilitate unique and complex training scenarios.
Adding supernumerary robotic limbs (SRLs) to humans and controlling them directly through the brain are main goals for movement augmentation. However, whether neural patterns that are distinct from the traditional inherent limbs motor imagery (MI) paradigm can be extracted, which is essential for the high-dimensional control of external equipment. In this study, a novel type of MI paradigm based on SRLs was proposed, consisting of "the sixth-finger", "the third-arm" and "the third-leg", and validated the distinctness of EEG response patterns between the novel and the traditional (hand, arm and leg) MI paradigm. The results showed that imagining extra limbs induced more obvious event-related desynchronization (ERD) phenomenon in sensorimotor areas compared to imagining inherent limbs. Classification results indicate well separable performance among different mental tasks (all above 86%, with a maximum of 90.5%). This work proposed a novel type of MI paradigm, and offered new way for widening the control bandwidth of the BCI system.
Recent advancements in neuroimaging involve the development of a novel electrocorticography (ECoG) grid. This grid is crafted using a polymer-thick film on an organic substrate (PTFOS), a design that offers several significant improvements over conventional ECoG grids, including Reduced Imaging Artifacts: The PTFOS grid produces negligible artifacts on MR images image quality of brain tissue overlaid with PTFOS grids. Lower Temperature Increase During MRI: In a 30-minute MR imaging session, temperature increases with the PTFOS grid were extremely low (0.4°C). These findings suggest that electrocorticography obtained using PTFOS grids could potentially improve the safety and efficacy of neurosurgical procedures, offering clearer imaging results and reducing risk factors such as excessive heating during MRIs.
Ankle-foot orthosis (AFO) prescriptions are common for patients presenting the deficit of foot drop. The intervention aims at improving ankle-foot function and enhancing general gait quality. In this work the asymmetries in gait of foot drop patients and the effectiveness of a plastic passive AFO are investigated by means of statistical methods performed on spatiotemporal measures collected using 3D gait analysis. Step length, stance, swing and single support phases exhibited statistically significant asymmetries between the limbs, while gait cycle time and stride length were meaningfully improved by the use of the orthotic device. The variability of results, with respect to the literature, suggests that general effects of standard AFO are hardly detectable, whereas a customised prescription of the orthosis should be considered.
Previous research shows that both anodal and cathodal high-definition transcranial direct current stimulation (HD-tDCS) may improve function of the upper extremity post stroke. However, most research has focused on the effects separately, therefore the purpose of this study was to determine the effects of performing simultaneous anodal-cathodal HD-tDCS. Five stroke participants received the stimulations in four visits with a two-week washout period: 1) anodal HD-tDCS to the ipsilesional primary motor cortex, 2) cathodal HD-tDCS to the contralesional dorsal premotor cortex, 3) bilateral anodal-cathodal HD-tDCS, and 4) sham. Active stimulation (anodal, cathodal, and bilateral) increased Fugl-Meyer upper extremity scores and decreased latency of ipsilesional M1-induced MEP. These results suggest that HD-tDCS could improve motor function of the upper extremity post-stroke, however, bilateral stimulation may not have an increased effect compared to anodal and cathodal HD-tDCS separately. This early phase study improves our understanding of neural circuitry and plasticity post stroke and HD-tDCS methods for improving function of the impaired arm post-stroke.
Humans possess huge individual differences in behaviors and understanding how the individual differences in brain give rise to the individual differences in behaviors is crucial for understanding the mechanism of brain function. Previous studies indicate that brain functional connectome is important for explaining the individual differences in behaviors and many researchers have already conducted connectome based prediction of individual behaviors. However, the ability of current connectome based prediction model is still limited. In this study, we proposed the convolutional graph propagation network (cGPN) which is a graph neural network that performs graph convolution on the brain connectome and propagates information between brain regions. We performed the individual predictions of many cognitive behaviors and demonstrated that cGPN outperforms various baseline models and achieves state of the art performance in connectome based prediction. The current results contribute to the understanding of behavioral individual differences and may also help to identify the mechanism of brain function.
Accurate segmentation of thyroid nodules is essential for early screening and diagnosis, but it can be challenging due to the nodules' varying sizes and positions. To address this issue, we propose a multi-attention guided UNet (MAUNet) for thyroid nodule segmentation. We use a multi-scale cross attention (MSCA) module for initial image feature extraction. By integrating interactions between features at different scales, the impact of thyroid nodule shape and size on the segmentation results has been reduced. Additionally, we incorporate a dual attention (DA) module into the skip-connection step of the UNet network, which promotes information exchange and fusion between the encoder and decoder. To test the model's robustness and effectiveness, we conduct the extensive experiments on multi-center ultrasound images provided by 17 hospitals. The experimental results show that our proposed method outperforms existing deep learning methods, which provides a new research direction for detecting thyroid nodules.
This paper proposes a semi-and-weak supervised pathological image segmentation method that effectively leverages the pre-recorded long-diameter information of a tumor in clinical as weak supervision. By leveraging the tumor diameter, the proposed method can accurately identify candidate tumor regions for pseudo-label selection. The accurate pseudo labels can improve the segmentation performance. The experimental results demonstrate the effectiveness of our method, which achieved the best performance among the comparative methods.
This study aims to show that multi-channel combinatorial optimization enables sinus rhythm RR interval tracking in electrocardiograms that are too noisy for existing algorithms to handle. The same detector achieves very high performance for arrhythmias and irregular QRS shapes in low/medium noise. To demonstrate high noise performance, results are given for the Non-Invasive Multimodal Foetal ECG-Doppler Dataset for Antenatal Cardiology Research ("NInFEA"). For 57 out of 60 NInFEA records, 84% and 95% of the estimated RR intervals were within 5 ms and 10 ms, respectively, of the ground truth RR intervals. The multi-channel methodology also produces state of the art results when applied to the MIT-BIH Arrhythmia Database ("MIT-BIH DB"), which includes many abnormal rhythms and QRS shapes. For the MIT-BIH DB, the sensitivity (SE) and positive predictive value (PPV) were 99.93% and 99.96% respectively. Although a rules based, Bayesian motivated algorithm is described, combinatorial optimization lends itself to neural network implementations. The strong performance in all noise conditions and flexible framework suggest a way forward for QRS detection in the new era of wearable devices.
OBJECTIVE:Low intensity focused stimulation (LIFUS) has been proved to improve motor function in Parkinson's disease (PD) animal modules. The aim of this study is to investigate whether LIFUS target on the primary motor cortex (M1) can improve motor deficit in the PD rats. METHODS:The PD rat model was induced by injection of 6-hydroxydopamine (6-OHDA) in the medial forebrain bundle (MFB). Two weeks after the injection, LIFUS was used on PD rats for two weeks. Behavioral tests were performed including open field test and rotarod test to examine the motor ability of the rats. The activity of microglia and astrocyte were tested to evaluate the inflammation level in the brain. The tyrosine hydroxylase (TH) staining was done to detect the recovery of dopaminergic (DA) neurons in the substantia nigra (SN) and DA fibers in the striatum (STR). RESULTS:LIFUS treatment decreased the resting time in OFT(p<0.05) and increased the latency to falls in the rotarod test(p<0.05) compared with the untreated PD rats. Moreover, LIFUS reduced the inflammation response reflected in microglia and astrocyte activation. Additionally, TH-immunoreactive fibers increased in the STR after LIFUS. CONCLUSION:These findings demonstrated that LIPUS targeted on M1 can inhibit neuroinflammation and improve movement disorders of PD rats. SIGNIFICANCE:This study provides a new therapeutic strategy for further clinical application in PD.
Alzheimer's disease (AD) is known to affect the lengths and frequencies of certain kinds of pauses in speech. Previous studies have used features based on pause lengths for AD classification. We conjecture that in addition to using pause lengths, it is beneficial to incorporate the "context" behind each pause, i.e., what is being said before and after each pause. We propose an AD detection method based on this idea. As part of the proposed method, pause lengths and context are extracted from the raw audio using automatic speech recognition (ASR) and forced alignment. Then, statistical summaries of pause lengths with context information are extracted from the transcripts and used as features for classification. Our results indicate that incorporating the context significantly improves classification performance compared to using pause lengths alone, with classification accuracy of up to 81%. Additionally, the proposed features largely preserve privacy.
