Research HighIights
Regulating stem cell functions by controlling ligand presentation can have an impact on tissue engineering and regenerative medicine. Dr Siu Hong Dexter WONG, Research Assistant Professor of Biomedical Engineering, alongside his collaborators, has developed light switchable tethers that are rationally designed and with lengths modulated by switching a light-responsive protein called pdDronpa in between monomer and dimer states.
This new approach enables consistent, stable control of the bioactive ligand presentation, without the spontaneous dissociation and translocation of ligands over a long cell culture period. The signaling of stem cells can be fine-tuned, by adjusting the tether lengths using selected wavelength lights. The approach does not change the biochemical environment and can decouple the effect of macroscopic matrix elasticity and local mechanical signal transduction, providing a bio-orthogonal, reversible, and spatiotemporally controlled strategy for bioactivity manipulations and a new route to control stem cell fate noninvasively.
Dr WONG successfully published his findings in the international journal, Advanced Materials.
To learn more: https://onlinelibrary.wiley.com/doi/abs/10.1002/adma.202105765
SERS (surface-enhanced Raman scattering) signals are promising for the detection of virus N genes. Prof Mo YANG, Associate Head and Professor of Biomedical Engineering, and Dr Siu Hong Dexter WONG, Research Assistant Professor of Biomedical Engineering, employed SERS for the detection of the N gene in SARs-CoV-2.
Together, the professors have developed a magnetic-responsive substrate consisting of heteoronanostructures, which controls the coupling distance for ultrasensitive and highly selective detection of the N gene of SARS-CoV-2.The platform magnetically modulates interparticle coupling for enhanced electric field strength and Raman signals.
The study proves the manipulation of DNA flexibility to control the coupling distance of metallic NPs by magnetic modulation. It also provides a promising strategy for detecting long-chain viral nucleic acids such as SARS-CoV-2 genes.
The department’s professors successfully published their findings in the international journal, ACS Applied Material Interfaces.
To learn more: https://pubs.acs.org/doi/10.1021/acsami.1c21173
The Department of Computing (COMP) recently received an Innovation and Technology Fund (ITF) grant for its “An Innovative Hybrid Intelligence Web System for the Hong Kong Stock Market” project. The project seeks to combine human intelligence and artificial intelligence (also known as Hybrid Intelligence), to analyse the Hong Kong stock market.
The Fintech project was awarded under the Partnership Research Programme and is in collaboration with the Hong Kong Institute of Financial Analysts (HKIFA) and Institute of Financial Technologists of Asia (IFTA).
A project launch seminar titled “Hong Kong Stock Market: Review and Outlook” was launched by COMP on 20 December 2021, supported by its industry partners. The seminar was attended by over a hundred industry practitioners and representatives, as well as students and alumni.
At the opening, Dr Miranda LOU, Executive Vice President at PolyU and Professor Christopher CHAO, Vice President of Research and Innovation at PolyU, greeted audiences and delivered welcome addresses. We were also honoured to invite Mr Joseph CHAN, JP, Under Secretary for Financial Services and the Treasury for the Government of the HKSAR, and Dr David CHUNG, JP, Under Secretary for Innovation and Technology for the Government of the HKSAR, as our distinguished guests.
Dr Kenny TANG, Chairman of HKIFA and Mr Paul PONG, Chairman of IFTA were the seminar’s guest speakers with presentations on the Hong Kong Stock Market. Moderated by Dr Henry CHAN, Associate Professor and Associate Head of COMP, Dr TANG and Mr PONG exchanged views on the Hong Kong Stock Market in 2021, as well as insights and forecasts for attractive sectors over the next year.
At the end of the seminar, project coordinator Dr Henry CHAN gave an overview on the collaborative research and development of “An Innovative Hybrid Intelligence Web System for the Hong Kong Stock Market”. By aiming for both knowledge transfer and entrepreneurship, he shared that the project will seek to contribute to Hong Kong’s overall Fintech development.
(https://www.polyu.edu.hk/comp/news-and-events/news/2022/0104_project-launch-sem/)
The Centre for Advances in Reliability and Safety (CAiRS), a new research centre utilizing advance equipment to ascertain product reliability and system safety, was recently jointly established by the Hong Kong Polytechnic University (PolyU) and the University of Maryland - College Park (UMD) in the USA.
The Centre gathers top local and international researchers, using advanced equipment and innovative artificial intelligence technology to conduct comprehensive product reliability and system safety research. The research will eventually accurately predict failure occurrence with an aim for prevention.
CAiRS has been included in the InnoHK Clusters, a major initiative of the HKSAR, with Ir. Professor Winco YUNG, Professor of the Department of Industrial and Systems Engineering at PolyU, appointed as the Centre Director and Executive Director of CAiRS.
Thus far, CAiRS has carried out five research programmes:1) “Anomaly Detection and Syndromic Surveillances”, 2) “Innovative Diagnostics for Health Management”, 3) “Prognostics for Remaining Useful Life Assessment”, 4) “Safety Assurance: Improve functional safety” and 5) “Data Analytics Platform for Reliability”. Under the programmes, 15 projects are running in parallel, with research spanning robots, medical devices , vehicles, telecommunications, consumer goods, public utilities, transportation, microelectronics, electrical installations, sensors, IoT products and advanced manufacturing.
CAiRS has also signed non-disclosure agreements with 31 well-regarded representative local companies within the industry, with 20 collaborative projects begun to improve the safety and reliability of their products and systems.
Going forward, PolyU is committed to conducting state-of-the-art interdisciplinary research, to aid the industry and society, with over 20 academics and scholars from the Faculty of Engineering of PolyU and UMD holding impressive research track records in product reliability. Combined with strong industry support, CAiRS will benefit and contribution to Hong Kong’s smart city development and manufacturing.
Gynaecological exams are essential procedures in women’s health, necessitating both prudence and a variety of specific techniques.
Practicing with a model has often been challenging, but a new virtual reality (VR) program has been designed to simulate the pelvic exam process, integrated alongside an Arduino hardware dummy.
The use of the VR program allows exam training to take place anywhere, while reducing medical waste such as lubricants, gloves and cervix brushes. Through the program, medical students are quickly familiarised with the assessment process.
The project is developed by the Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University and co-developed with Dr Vincent CHEUNG, The University of Hong Kong.
Fire drill safety is critical in the workplace. Particularly for medical services, rapid response is essential to minimize casualties and losses. But exercises can often be significant expenditures, alongside increased viral dangers during the pandemic.
Virtual reality fire drill simulations have now been developed as viable alternatives. In the simulation, fire drill evacuations are created for workers on a hospital ward, demonstrating the use a fire extinguisher, evacuation in proper sequence and choosing the correct exit door. The simulation also evaluates the user’s fire safety awareness through a questionnaire with scores at the end.
The project is developed by the Department of Industrial and Systems Engineering.
Light detection and illumination analysis are important in photonics research, particularly where specular highlight detection and removal are challenging tasks. Recent methods have achieved promising results, but they are often designed for highlight detection or removal, with performance deteriorating on real-world images.
We have developed a novel network to detect and remove highlights from natural images. We started by introducing a dataset with about 16K real images, each with corresponding ground truths of highlight detection and removal. Using the presented dataset, we then developed a multi-task network for joint highlight detection and removal, based on a new specular model. Experiments show that our approach clearly outperforms state-of-the-art methods for both highlight detection and removal.
The project is developed by Dr LI Ping, Department of Computing.
Visual comparison of our method against state-of-the-art highlight removal methods on real-world images from the Internet.
Statistics of our dataset. We show that the proposed dataset has a diversity of materials on which highlight characteristically appear and highlight with reasonable property distributions in terms of area, number and location.
Pipeline of our data generation. We provide a typical example in each stage.
The pipeline of our joint highlight detection and removal network. Our network applies an encoder-decoder structure with DSCFA (Dilated Spatial Contextual Feature Aggregation) modules to extract highlight features F from the input highlight image I. Based on F, M, S and D are then subsequently predicted to perform the joint highlight detection, estimation and removal.
Visual comparison of our method against state-of-the-art highlight removal methods on real-world images from the Internet.
Visual comparison of our method against state-of-the-art highlight removal methods on real-world images from the Internet.
Humans rely on vision to perceive the world around us. When light rays hit the retina, cells in our eyes turn these into electrical signals, creating 2D mental projections of our physical 3D world.
Artificial intelligence has long had a goal to build systems using 2D images to infer 3D real-world scenes, with wide applications in robotics and augmented reality (AR), such as intelligent robots understanding living room blueprints to easily navigate around furniture.
From this, we have created deep-learning methods that emulate light ray tracing, enabling machines to see 3D physical worlds from 2D images. Both the geometry and appearance of each 3D point is modelled as a simple neural network and using 2D images captured from different viewing angles, these are used as light ray projections of the 3D space.
Our method reprojects the 2D pixel information back to 3D space along the light ray direction, so the 3D point shape and appearance is learned from the 2D images. We recently published the method at ICCV 2021, a computer vision and machine intelligence conference.
Using 2D views to understand the geometry and appearance of real-world 3D scenes has many applications for robotics or mixed reality, including novel view synthesis and semantic scene understanding.
For novel view synthesis: after feeding 2D images collected from mobile phones into the network, we query new viewing angles. As the neural network understands the 3D geometry and appearance, it can generate photorealistic images, despite never having seen those views before.
For semantic scene understanding: our method can emulate humans by recognizing all objects in 3D scenes. After collecting 2D images of a bathroom, our network infers categories for every pixel in all novel views. For example, pink represents the bathtub while cyan is the mirror.
Overall, natural light rays are modelled as a neural network, enabling the machine to perceive real-world 3D scenes from the 2D images. This technique thus has profound applications across robotics, computer games, and other technology.
The project is developed by Dr YANG Bo, Department of Computing.
PolyU recently partnered with the Biel Crystal (HK) Manufactory limited, to establish the Biel-PolyU Glass Research Joint Laboratory.
The laboratory will focus on collaborative research on precision glass manufacturing, with Dr RUAN Haihui, Department of Mechanical Engineering receiving HK$5m of industrial support Biel. Research conducted will focus predominantly on curved cover glass, used in smartphones, wearable electronics, VR/AR glasses, and electronic vehicles. The lab will develop models to predict glass stress and deformation, protocols to aid mold design to compensate for shrinkage and chemical strengthening, as well as novel materials and manufacturing technology for transparency of thermoformed glass and ensure high-volume production.
Over the past decade, manufacturing curved glass covers has become costly and time-consuming due to testing processes and an upgrade to its computation-based process development protocol was urgently needed.
