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

General General

Advanced Reinforcement Learning and Its Connections with Brain Neuroscience.

In Research (Washington, D.C.)

In recent years, brain science and neuroscience have greatly propelled the innovation of computer science. In particular, knowledge from the neurobiology and neuropsychology of the brain revolutionized the development of reinforcement learning (RL) by providing novel interpretable mechanisms of how the brain achieves intelligent and efficient decision making. Triggered by this, there has been a boom in research about advanced RL algorithms that are built upon the inspirations of brain neuroscience. In this work, to further strengthen the bidirectional link between the 2 communities and especially promote the research on modern RL technology, we provide a comprehensive survey of recent advances in the area of brain-inspired/related RL algorithms. We start with basis theories of RL, and present a concise introduction to brain neuroscience related to RL. Then, we classify these advanced RL methodologies into 3 categories according to different connections of the brain, i.e., micro-neural activity, macro-brain structure, and cognitive function. Each category is further surveyed by presenting several modern RL algorithms along with their mathematical models, correlations with the brain, and open issues. Finally, we introduce several important applications of RL algorithms, followed by the discussions of challenges and opportunities for future research.

Fan Chaoqiong, Yao Li, Zhang Jiacai, Zhen Zonglei, Wu Xia

2023

General General

Modifying the Power and Performance of 2-Dimensional MoS2 Field Effect Transistors.

In Research (Washington, D.C.)

Over the past 60 years, the semiconductor industry has been the core driver for the development of information technology, contributing to the birth of integrated circuits, Internet, artificial intelligence, and Internet of Things. Semiconductor technology has been evolving in structure and material with co-optimization of performance-power-area-cost until the state-of-the-art sub-5-nm node. Two-dimensional (2D) semiconductors are recognized by the industry and academia as a hopeful solution to break through the quantum confinement for the future technology nodes. In the recent 10 years, the key issues on 2D semiconductors regarding material, processing, and integration have been overcome in sequence, making 2D semiconductors already on the verge of application. In this paper, the evolution of transistors is reviewed by outlining the potential of 2D semiconductors as a technological option beyond the scaled metal oxide semiconductor field-effect transistors. We mainly focus on the optimization strategies of mobility (μ), equivalent oxide thickness (EOT), and contact resistance (RC ), which enables high ON current (Ion ) with reduced driving voltage (Vdd ). Finally, we prospect the semiconductor technology roadmap by summarizing the technological development of 2D semiconductors over the past decade.

Zhuo Fulin, Wu Jie, Li Binhong, Li Moyang, Tan Chee Leong, Luo Zhongzhong, Sun Huabin, Xu Yong, Yu Zhihao

2023

General General

Attentional Generative Multimodal Network for Neonatal Postoperative Pain Estimation.

In Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention

Artificial Intelligence (AI)-based methods allow for automatic assessment of pain intensity based on continuous monitoring and processing of subtle changes in sensory signals, including facial expression, body movements, and crying frequency. Currently, there is a large and growing need for expanding current AI-based approaches to the assessment of postoperative pain in the neonatal intensive care unit (NICU). In contrast to acute procedural pain in the clinic, the NICU has neonates emerging from postoperative sedation, usually intubated, and with variable energy reserves for manifesting forceful pain responses. Here, we present a novel multi-modal approach designed, developed, and validated for assessment of neonatal postoperative pain in the challenging NICU setting. Our approach includes a robust network capable of efficient reconstruction of missing modalities (e.g., obscured facial expression due to intubation) using an unsupervised spatio-temporal feature learning with a generative model for learning the joint features. Our approach generates the final pain score along with the intensity using an attentional cross-modal feature fusion. Using experimental dataset from postoperative neonates in the NICU, our pain assessment approach achieves superior performance (AUC 0.906, accuracy 0.820) as compared to the state-of-the-art approaches.

Salekin Md Sirajus, Zamzmi Ghada, Goldgof Dmitry, Mouton Peter R, Anand Kanwaljeet J S, Ashmeade Terri, Prescott Stephanie, Huang Yangxin, Sun Yu

2022-Sep

Generative model, Multimodal learning, NICU, Neonatal pain, Postoperative pain

General General

Ultrafast Miniature Robotic Swimmers with Upstream Motility.

In Cyborg and bionic systems (Washington, D.C.)

With the development of materials science and micro-nano fabrication techniques, miniature soft robots at millimeter or submillimeter size can be manufactured and actuated remotely. The small-scaled robots have the unique capability to access hard-to-reach regions in the human body in a noninvasive manner. To date, it is still challenging for miniature robots to accurately move in the diverse and dynamic environments in the human body (e.g., in blood flow). To effectively locomote in the vascular system, miniature swimmers with upstream swimming capability are required. Herein, we design and fabricate a miniature robotic swimmer capable of performing ultrafast swimming in a fluidic environment. The maximum velocity of the swimmer in water is 30 cm/s, which is 60 body lengths. Moreover, in a tubular environment, the swimmer can still obtain a swimming velocity of 17 cm/s. The swimmer can also perform upstream swimming in a tubular environment with a velocity of 5 cm/s when the flow speed is 10 cm/s. The ultrasound-guided navigation of the swimmer in a phantom mimicking a blood vessel is also realized. This work gives insight into the design of agile undulatory milliswimmers for future biomedical applications.

Wang Yibin, Chen Hui, Law Junhui, Du Xingzhou, Yu Jiangfan

2023

General General

Prediction of Needle Physiological Traits Using UAV Imagery for Breeding Selection of Slash Pine.

In Plant phenomics (Washington, D.C.)

Leaf nitrogen (N) content and nonstructural carbohydrate (NSC) content are 2 important physiological indicators that reflect the growth state of trees. Rapid and accurate measurement of these 2 traits multitemporally enables dynamic monitoring of tree growth and efficient tree breeding selection. Traditional methods to monitor N and NSC are time-consuming, are mostly used on a small scale, and are nonrepeatable. In this paper, the performance of unmanned aerial vehicle multispectral imaging was evaluated over 11 months of 2021 on the estimation of canopy N and NSC contents from 383 slash pine trees. Four machine learning methods were compared to generate the optimal model for N and NSC prediction. In addition, the temporal scale of heritable variation for N and NSC was evaluated. The results show that the gradient boosting machine model yields the best prediction results on N and NSC, with R 2 values of 0.60 and 0.65 on the validation set (20%), respectively. The heritability (h 2) of all traits in 11 months ranged from 0 to 0.49, with the highest h 2 for N and NSC found in July and March (0.26 and 0.49, respectively). Finally, 5 families with high N and NSC breeding values were selected. To the best of our knowledge, this is the first study to predict N and NSC contents in trees using time-series unmanned aerial vehicle multispectral imaging and estimating the genetic variation of N and NSC along a temporal scale, which provides more reliable information about the overall performance of families in a breeding program.

Niu Xiaoyun, Song Zhaoying, Xu Cong, Wu Haoran, Luan Qifu, Jiang Jingmin, Li Yanjie

2023

oncology Oncology

Deep learning-based fast volumetric imaging using kv and mv projection images for lung cancer radiotherapy: A feasibility study.

In Medical physics ; h5-index 59.0

PURPOSE : The long acquisition time of CBCT discourages repeat verification imaging, therefore increasing treatment uncertainty. In this study, we present a fast volumetric imaging method for lung cancer radiation therapy using an orthogonal 2D kV/MV image pair.

METHODS : The proposed model is a combination of 2D and 3D networks. The proposed model consists of five major parts: 1) kV and MV feature extractors are used to extract deep features from the perpendicular kV and MV projections. 2) The feature-matching step is used to re-align the feature maps to their projection angle in a Cartesian coordinate system. By using a residual module, the feature map can focus more on the difference between the estimated and ground truth images. 3) In addition, the feature map is downsized to include more global semantic information for the 3D estimation, which is useful to reduce inhomogeneity. By using convolution-based reweighting, the model is able to further increase the uniformity of image. 4) To reduce the blurry noise of generated 3D volume, the Laplacian latent space loss calculated via the feature map that is extracted via specifically-learned Gaussian kernel is used to supervise the network. 5) Finally, the 3D volume is derived from the trained model. We conducted a proof-of-concept study using 50 patients with lung cancer. An orthogonal kV/MV pair was generated by ray tracing through CT of each phase in a 4D CT scan. Orthogonal kV/MV pairs from 9 respiratory phases were used to train this patient-specific model while the kV/MV pair of the remaining phase was held for model testing.

RESULTS : The results are based on simulation data and phantom results from a real Linac system. The mean absolute error (MAE) values achieved by our method were 57.5 HU and 77.4 HU within body and tumor region-of-interest (ROI), respectively. The mean achieved peak-signal-to-noise ratios (PSNR) were 27.6 dB and 19.2 dB within the body and tumor ROI, respectively. The achieved mean normalized cross correlation (NCC) values were 0.97 and 0.94 within the body and tumor ROI, respectively. A phantom study demonstrated that the proposed method can accurately re-position the phantom after shift. It is also shown that the proposed method using both kV and MV is superior to current method using kV or MV only in image quality.

CONCLUSION : These results demonstrate the feasibility and accuracy of our proposed fast volumetric imaging method from an orthogonal kV/MV pair, which provides a potential solution for daily treatment setup and verification of patients receiving radiation therapy for lung cancer. This article is protected by copyright. All rights reserved.

Lei Yang, Tian Zhen, Wang Tonghe, Roper Justin, Xie Huiqiao, Kesarwala Aparna H, Higgins Kristin, Bradley Jeffrey D, Liu Tian, Yang Xiaofeng

2023-Mar-20

deep inspiration breath-hold lung radiotherapy, deep learning, fast 3D imaging