Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Surgery Surgery

Automatic registration of dental CT and 3D scanned model using deep split jaw and surface curvature.

In Computer methods and programs in biomedicine

BACKGROUND AND OBJECTIVES : In the medical field, various image registration applications have been studied. In dentistry, the registration of computed tomography (CT) volume data and 3D optically scanned models is essential for various clinical applications, including orthognathic surgery, implant surgical planning, and augmented reality. Our purpose was to present a fully automatic registration method of dental CT data and 3D scanned models.

METHODS : We use a 2D convolutional neural network to regress a curve splitting the maxilla (i.e., upper jaw) and mandible (i.e., lower jaw) and the points specifying the front and back ends of the crown from the CT data. Using this regressed information, we extract the point cloud and vertices corresponding to the tooth crown from the CT and scanned data, respectively. We introduce a novel metric, called curvature variance of neighbor (CVN), to discriminate between highly fluctuating and smoothly varying regions of the tooth crown. The registration based on CVN enables more accurate fine registration while reducing the effects of metal artifacts. Moreover, the proposed method does not require any preprocessing such as extracting the iso-surface for the tooth crown from the CT data, thereby significantly reducing the computation time.

RESULTS : We evaluated the proposed method with the comparison to several promising registration techniques. Our experimental results using three datasets demonstrated that the proposed method exhibited higher registration accuracy (i.e., 2.85, 1.92, and 7.73 times smaller distance errors for individual datasets) and smaller computation time (i.e., 4.12 times faster registration) than one of the state-of-the-art methods. Moreover, the proposed method worked considerably well for partially scanned data, whereas other methods suffered from the unbalancing of information between the CT and scanned data.

CONCLUSIONS : The proposed method was able to perform fully automatic and highly accurate registration of dental CT data and 3D scanned models, even with severe metal artifacts. In addition, it could achieve fast registration because it did not require any preprocessing for iso-surface reconstruction from the CT data.

Kim Minchang, Chung Minyoung, Shin Yeong-Gil, Kim Bohyoung

2023-Mar-07

Deep learning, Dental computed tomography (CT), Metal artifacts reduction, Registration, Scanned surface model

General General

HARDC : A novel ECG-based heartbeat classification method to detect arrhythmia using hierarchical attention based dual structured RNN with dilated CNN.

In Neural networks : the official journal of the International Neural Network Society

Deep learning-based models have achieved significant success in detecting cardiac arrhythmia by analyzing ECG signals to categorize patient heartbeats. To improve the performance of such models, we have developed a novel hybrid hierarchical attention-based bidirectional recurrent neural network with dilated CNN (HARDC) method for arrhythmia classification. This solves problems that arise when traditional dilated convolutional neural network (CNN) models disregard the correlation between contexts and gradient dispersion. The proposed HARDC fully exploits the dilated CNN and bidirectional recurrent neural network unit (BiGRU-BiLSTM) architecture to generate fusion features. As a result of incorporating both local and global feature information and an attention mechanism, the model's performance for prediction is improved. By combining the fusion features with a dilated CNN and a hierarchical attention mechanism, the trained HARDC model showed significantly improved classification results and interpretability of feature extraction on the PhysioNet 2017 challenge dataset. Sequential Z-Score normalization, filtering, denoising, and segmentation are used to prepare the raw data for analysis. CGAN (Conditional Generative Adversarial Network) is then used to generate synthetic signals from the processed data. The experimental results demonstrate that the proposed HARDC model significantly outperforms other existing models, achieving an accuracy of 99.60%, F1 score of 98.21%, a precision of 97.66%, and recall of 99.60% using MIT-BIH generated ECG. In addition, this approach significantly reduces run time when using dilated CNN compared to normal convolution. Overall, this hybrid model demonstrates an innovative and cost-effective strategy for ECG signal compression and high-performance ECG recognition. Our results indicate that an automated and highly computed method to classify multiple types of arrhythmia signals holds considerable promise.

Islam Md Shofiqul, Hasan Khondokar Fida, Sultana Sunjida, Uddin Shahadat, Lio’ Pietro, Quinn Julian M W, Moni Mohammad Ali

2023-Mar-05

Arrhythmia, BiGRU–BiLSTM, Dilated CNN, ECG, Hierarchical attention, Preprocessing

General General

An integrated evaluation approach of wearable lower limb exoskeletons for human performance augmentation.

In Scientific reports ; h5-index 158.0

Wearable robots have been growing exponentially during the past years and it is crucial to quantify the performance effectiveness and to convert them into practical benchmarks. Although there exist some common metrics such as metabolic cost, many other characteristics still needs to be presented and demonstrated. In this study, we developed an integrated evaluation (IE) approach of wearable exoskeletons of lower limb focusing on human performance augmentation. We proposed a novel classification of trial tasks closely related to exoskeleton functions, which were divided into three categories, namely, basic trial at the preliminary phase, semi-reality trial at the intermediate phase, and reality trial at the advanced phase. In the present study, the IE approach has been exercised with a subject who wore an active power-assisted knee (APAK) exoskeleton with three types of trial tasks, including walking on a treadmill at a certain angle, walking up and down on three-step stairs, and ascending in 11-storey stairs. Three wearable conditions were carried out in each trial task, i.e. with unpowered exoskeleton, with powered exoskeleton, and without the exoskeleton. Nine performance indicators (PIs) for evaluating performance effectiveness were adopted basing on three aspects of goal-level, task-based kinematics, and human-robot interactions. Results indicated that compared with other conditions, the powered APAK exoskeleton make generally lesser heart rate (HR), Metabolic equivalent (METs), biceps femoris (BF) and rectus femoris (RF) muscles activation of the subject at the preliminary phase and intermediate phase, however, with minimal performance augmentation at advanced phase, suggesting that the APAK exoskeleton is not suitable for marketing and should be further improved. In the future, continuous iterative optimization for the IE approach may help the robot community to attain a comprehensive benchmarking methodology for robot-assisted locomotion more efficiently.

Zhang Xiao, Chen Xue, Huo Bo, Liu Chenglin, Zhu Xiaorong, Zu Yuanyuan, Wang Xiliang, Chen Xiao, Sun Qing

2023-Mar-14

General General

Generative Adversarial Network for Personalized Art Therapy in Melanoma Disease Management

ArXiv Preprint

Melanoma is the most lethal type of skin cancer. Patients are vulnerable to mental health illnesses which can reduce the effectiveness of the cancer treatment and the patients adherence to drug plans. It is crucial to preserve the mental health of patients while they are receiving treatment. However, current art therapy approaches are not personal and unique to the patient. We aim to provide a well-trained image style transfer model that can quickly generate unique art from personal dermoscopic melanoma images as an additional tool for art therapy in disease management of melanoma. Visual art appreciation as a common form of art therapy in disease management that measurably reduces the degree of psychological distress. We developed a network based on the cycle-consistent generative adversarial network for style transfer that generates personalized and unique artworks from dermoscopic melanoma images. We developed a model that converts melanoma images into unique flower-themed artworks that relate to the shape of the lesion and are therefore personal to the patient. Further, we altered the initial framework and made comparisons and evaluations of the results. With this, we increased the options in the toolbox for art therapy in disease management of melanoma. The development of an easy-to-use user interface ensures the availability of the approach to stakeholders. The transformation of melanoma into flower-themed artworks is achieved by the proposed model and the graphical user interface. This contribution opens a new field of GANs in art therapy and could lead to more personalized disease management.

Lennart Jütte, Ning Wand, Bernhard Roth

2023-03-16

General General

Edge computing on TPU for brain implant signal analysis.

In Neural networks : the official journal of the International Neural Network Society

The ever-increasing number of recording sites of silicon-based probes imposes a great challenge for detecting and evaluating single-unit activities in an accurate and efficient manner. Currently separate solutions are available for high precision offline evaluation and separate solutions for embedded systems where computational resources are more limited. We propose a deep learning-based spike sorting system, that utilizes both unsupervised and supervised paradigms to learn a general feature embedding space and detect neural activity in raw data as well as predict the feature vectors for sorting. The unsupervised component uses contrastive learning to extract features from individual waveforms, while the supervised component is based on the MobileNetV2 architecture. One of the key advantages of our system is that it can be trained on multiple, diverse datasets simultaneously, resulting in greater generalizability than previous deep learning-based models. We demonstrate that the proposed model does not only reaches the accuracy of current state-of-art offline spike sorting methods but has the unique potential to run on edge Tensor Processing Units (TPUs), specialized chips designed for artificial intelligence and edge computing. We compare our model performance with state of art solutions on paired datasets as well as on hybrid recordings as well. The herein demonstrated system paves the way to the integration of deep learning-based spike sorting algorithms into wearable electronic devices, which will be a crucial element of high-end brain-computer interfaces.

Rokai János, Ulbert István, Márton Gergely

2023-Feb-28

Brain–computer interface, Deep learning, Edge device, Electrophysiology, Feature extraction, Spike sorting

General General

Global distribution of marine microplastics and potential for biodegradation.

In Journal of hazardous materials

Microplastics are a growing marine environmental concern globally due to their high abundance and persistent degradation. We created a global map for predicting marine microplastic pollution using a machine-learning model based on 9445 samples and found that microplastics converged in zones of accumulation in subtropical gyres and near polar seas. The predicted global potential for the biodegradation of microplastics in 1112 metagenome-assembled genomes from 485 marine metagenomes indicated high potential in areas of high microplastic pollution, such as the northern Atlantic Ocean and the Mediterranean Sea. However, the limited number of samples hindered our prediction, a priority issue that needs to be addressed in the future. We further identified hosts with microplastic degradation genes (MDGs) and found that Proteobacteria accounted for a high proportion of MDG hosts, mainly Alphaproteobacteria and Gammaproteobacteria, with host-specific patterns. Our study is essential for raising awareness, identifying areas with microplastic pollution, providing a prediction method of machine learning to prioritize surveillance, and identifying the global potential of marine microbiomes to degrade microplastics, providing a reference for selecting bacteria that have the potential to degrade microplastics for further applied research.

Chen Bingfeng, Zhang Zhenyan, Wang Tingzhang, Hu Hang, Qin Guoyan, Lu Tao, Hong Wenjie, Hu Jun, Penuelas Josep, Qian Haifeng

2023-Mar-11

Biodegradation potential, Global microplastic distribution, Machine learning, Metagenome, Microplastic