Feasibility of decoding covert speech in ECoG with a Transformer trained on overt speech.

Several attempts for speech brain-computer interfacing (BCI) have been made to decode phonemes, sub-words, words, or sentences using invasive measurements, such as the electrocorticogram (ECoG), during auditory speech perception, overt speech, or imagined (covert) speech. Decoding sentences from covert speech is a challenging task. Sixteen epilepsy patients with intracranially implanted electrodes participated in this study, and ECoGs were recorded during overt speech and covert speech of eight Japanese sentences, each consisting of three tokens. In particular, Transformer neural network model was applied to decode text sentences from covert speech, which was trained using ECoGs obtained during overt speech. We first examined the proposed Transformer model using the same task for training and testing, and then evaluated the model's performance when trained with overt task for decoding covert speech. The Transformer model trained on covert speech achieved an average token error rate (TER) of 46.6% for decoding covert speech, whereas the model trained on overt speech achieved a TER of 46.3% ( p > 0.05 ; d = 0.07 ) . Therefore, the challenge of collecting training data for covert speech can be addressed using overt speech. The performance of covert speech can improve by employing several overt speeches.

The project for objective measures using computational psychiatry technology (PROMPT): Rationale, design, and methodology.

INTRODUCTION: Depressive and neurocognitive disorders are debilitating conditions that account for the leading causes of years lived with disability worldwide. However, there are no biomarkers that are objective or easy-to-obtain in daily clinical practice, which leads to difficulties in assessing treatment response and developing new drugs. New technology allows quantification of features that clinicians perceive as reflective of disorder severity, such as facial expressions, phonic/speech information, body motion, daily activity, and sleep. METHODS: Major depressive disorder, bipolar disorder, and major and minor neurocognitive disorders as well as healthy controls are recruited for the study. A psychiatrist/psychologist conducts conversational 10-min interviews with participants ≤10 times within up to five years of follow-up. Interviews are recorded using RGB and infrared cameras, and an array microphone. As an option, participants are asked to wear wrist-band type devices during the observational period. Various software is used to process the raw video, voice, infrared, and wearable device data. A machine learning approach is used to predict the presence of symptoms, severity, and the improvement/deterioration of symptoms. DISCUSSION: The overall goal of this proposed study, the Project for Objective Measures Using Computational Psychiatry Technology (PROMPT), is to develop objective, noninvasive, and easy-to-use biomarkers for assessing the severity of depressive and neurocognitive disorders in the hopes of guiding decision-making in clinical settings as well as reducing the risk of clinical trial failure. Challenges may include the large variability of samples, which makes it difficult to extract the features that commonly reflect disorder severity. TRIAL REGISTRATION: UMIN000021396, University Hospital Medical Information Network (UMIN).

Importance of over-reading ambulatory ECG-based microvolt T-wave alternans to eliminate three main sources of measurement error.

BACKGROUND: Ambulatory electrocardiogram (ECG)-based microvolt T-wave alternans values measured by the modified moving average method (MMA-TWA) can be disrupted by T-wave changes that mimic true repolarization alternans. METHODS: We investigated potential sources of measurement error by studying 19 healthy subjects (12 men; median age, 25) free of known heart disease with 36-month follow-up to establish freedom from significant arrhythmia or syncope. All participants underwent 24-hr continuous 12-lead ECG monitoring. Causes of automated MMA-TWA ≥42 µV episodes were classified based on visual inspection. RESULTS: A total of 2,189 episodes of automated MMA-TWA episodes ≥42 µV were observed in all subjects (peak MMA-TWA: median, 94 µV; interquartile range, 81-112 µV). All episodes included one or more beats with T-wave deformation which lacked "repeating ABAB pattern" and therefore were identified as TWA measurement error. Causes of such error were categorized as: (a) artifact [72.6% (1,589/2,189), observed in 19 (100%) subjects], more frequently in limb than precordial leads; (b) T-wave changes due to changes in heart/body position [25.5% (559/2,189), observed in 14 (73.7%) subjects], frequently observed in leads V1-2; and (c) postextrasystolic T-wave changes [1.9% (41/2,189), observed in 2 (10.5%) subjects]. CONCLUSIONS: Relying only on automated MMA-TWA values obtained during ambulatory ECG monitoring can lead to incorrect measurement of TWA. Our findings offer the potential to reduce false-positive TWA results and to achieve more accurate detection of true repolarization alternans.

Structural Comparison between the Right and Left Atrial Appendages Using Multidetector Computed Tomography.

The three-dimensional (3D) structures of the right atrial appendage (RAA) and left atrial appendage (LAA) were compared to clarify why thrombus formation less frequently occurs in RAA than in LAA. Morphological differences between RAA and LAA of 34 formalin-preserved cadaver hearts were investigated. Molds of RAA and LAA specimens were made and the neck areas, volumes of the atrial appendages (AA), and amount of pectinate muscles (PMs) were analyzed using multidetector computed tomography. In RAA, most PMs were connected to one another and formed a "dendritic" appearance and the inner surface area was smaller than in LAA. RAA had smaller volumes and larger neck areas than LAA. The ratios of the neck area/volume were larger and the amounts of PMs were smaller in RAA than in LAA. The volumes, neck areas, and amount of PMs of RAA were significantly correlated with those of LAA. According to the 3D structure, RAA appears to be suited for a more favorable blood flow, which may explain why the thrombus formation is less common in RAA than in LAA. Examining not only LAA but also RAA by transesophageal echocardiography may be useful in high-risk patients of thrombus formation in LAA because the volume, neck area, and amount of PMs of LAA reflect the shape of RAA.

Fast Coding of Feature Vectors Using Neighbor-to-Neighbor Search.

Searching for matches to high-dimensional vectors using hard/soft vector quantization is the most computationally expensive part of various computer vision algorithms including the bag of visual word (BoW). This paper proposes a fast computation method, Neighbor-to-Neighbor (NTN) search [1] , which skips some calculations based on the similarity of input vectors. For example, in image classification using dense SIFT descriptors, the NTN search seeks similar descriptors from a point on a grid to an adjacent point. Applications of the NTN search to vector quantization, a Gaussian mixture model, sparse coding, and a kernel codebook for extracting image or video representation are presented in this paper. We evaluated the proposed method on image and video benchmarks: the PASCAL VOC 2007 Classification Challenge and the TRECVID 2010 Semantic Indexing Task. NTN-VQ reduced the coding cost by 77.4 percent, and NTN-GMM reduced it by 89.3 percent, without any significant degradation in classification performance.

Sonographic Detection of Subcutaneous Foreign Bodies in 3 Cases.

Subcutaneous masses caused by foreign bodies are frequently encountered in daily practice. Although the majority of foreign bodies such as metals can be detected by radiography, substances such as vegetative materials or wood are difficult to detect. To our knowledge, only a few studies have described the sonographic characteristics of foreign bodies. Herein, we report 3 cases where we studied the sonographic characteristics of the foreign bodies in the dermis and subcutaneous tissue. Our results revealed the following 3 foreign bodies: (1) glass, (2) vegetative material, and (3) a pencil core. Thus, sonographic examination is useful for the detection of foreign bodies.

Sonographic Findings of Subcutaneous Sarcoidosis in 3 Cases.

Subcutaneous sarcoidosis occurs infrequently among cases of cutaneous sarcoidosis. To our knowledge, few studies have described the sonographic characteristics of subcutaneous sarcoidosis. Here we report the sonographic characteristics of 3 cases of this condition. Our results revealed 4 features: (1) an irregular hypoechoic appearance, (2) heterogeneous echogenicity, (3) perilesional hyperechoic changes, and (4) abnormal Doppler signals. Sonography is a rapid, simple, and noninvasive procedure that is useful for initial evaluation of granulomatous lesions such as subcutaneous sarcoidosis.

[Diagnostic ability of ultrasonography for the rotator cuff tear: comparison with ultrasonography and MRI findings].

Ultrasonography (US) is a non-invasive method which can assess not only solid tissue organs but also soft tissues such as tendons and nerves. However, it has not been fully understood that US is a useful tool for the depiction of periarticular structure. We compared the diagnostic accuracies between US and magnetic resonance imaging (MRI) in the patients with rotator cuff tear (RCT). Seventy patients, who underwent arthroscopic surgery, preoperative US and MRI examinations at Gifu University Hospital from January 2010 to April 2013 (49 male, 21 female, mean age 59.7 +/- 15.9) were included in this study. The diagnostic accuracy, sensitivity and specificity of US and MRI were 94.3% and 94.3%, 95.8% and 97.9%, 90.9% and 84.6%, respectively, when the intraoperative finding was regarded as a gold standard. These results suggest that US is useful for the diagnosis of RCT as equal as MRI.