New algorithm represents a promising tool for the early detection and classification of cancer, with the potential to streamline the diagnostic process

 

For the approximately 20 million new cases of cancer reported each year, early and accurate diagnosis is crucial.  The demand for precise diagnostic tools has never been higher, and traditionally, the visual examination of tissue slides by pathologists has been the gold standard in cancer detection. However, with advancements in digital pathology, these tissue slides can now be digitized into multi-gigapixel whole slide images (WSIs). These high-resolution images hold immense potential for machine learning applications, but their sheer size presents a significant challenge.

 

A team of researchers including Khalifa University’s Dr. Sajid Javed and Prof. Naoufel Werghi, proposed an innovative, fully unsupervised machine learning approach to classify WSIs, enabling faster and more accurate cancer detection, among other clinical uses. The team also included researchers from Information Technology University, Pakistan, and the University of Warwick, United Kingdom. Their results were published in Medical Image Analysis, a top 1% journal.

 

Classifying WSIs involves analyzing the image to determine whether it contains cancerous tissue. Deep learning models for classifying WSIs already exist but they often require manual annotations for expert pathologists, which is both time-consuming and costly. Recent advancements have introduced weakly supervised learning methods to alleviate this burden, but these still depend on large, labelled datasets. The research team’s fully unsupervised approach bypasses the need for any labelled data.

 

Their algorithm divides WSIs into smaller, more manageable patches. These patches are then transformed and subsequently inverse-transformed back to their original spaces. The transformation error — the difference between the original and inverse-transformed patches — is used to generate “pseudo labels”. This method hinges on a crucial observation: Normal tissue patches tend to be more homogenous than cancerous patches, which exhibit greater variability in texture and patterns.

 

The algorithm then further refines the labels to reduce noise and enhance accuracy. This mutual learning process continues iteratively, with each cycle improving the model’s performance.

 

The researchers tested their algorithm on four publicly available datasets, with their model outperforming existing state-of-the-art approaches in fully unsupervised settings, underscoring its potential for effective cancer diagnosis without the need for expensive and labor-intensive annotations from human experts. These implications are profound, with this algorithm a promising tool for the early detection and classification of cancer, making it more efficient, accurate and accessible.

 

The research team says further research could explore the integration of this method with semi-supervised or weakly supervised approaches to further enhance its accuracy and applicability in clinical settings. 

 

Jade Sterling
Science Writer
11 July 2024

Synergy between computational power and molecular science heralds a new era in the rational design of ionic liquids 

 

In a leap forward for green chemistry, a team of researchers at Khalifa University has harnessed the power of deep learning to predict the properties of over 300,000 novel ionic liquid variants.

 

Ionic liquids are a class of compounds known for their unique, tunable properties and minimal environmental impact. They have applications in energy storage, nano-engineering, drug delivery, and environmental remediation, among many others, but the sheer number of possible combinations — created by pairing different cations and anions — presents a daunting challenge. Traditionally, identifying the right ionic liquid for a specific application has required laborious and time-consuming experimental work.

 

To overcome this, the Khalifa University team from the Research & Innovation Center for Graphene and 2D Materials (RIC-2D), and the Center for Membranes and Advanced Water Technology (CMAT), turned to computational methods, combining robust molecular modeling with advanced ensemble deep learning techniques. Tarek Lemaoui, Tarek Eid, Ahmad Darwish, Prof. Hassan Arafat, Prof. Fawzi Banat, and Prof. Enas Nashef developed an artificial neural network model capable of reliably predicting how different ionic liquids will behave based on their molecular structures.

 

Their results were published in Materials Science and Engineering R: Reports, a top 1% journal.

 

The research team’s model screened 303,880 ionic liquids, created by systematically combining 1070 cations with 284 anions. This screening process allows researchers to identify ionic liquids with specific property profiles, drastically reducing the need for extensive experimental validation. The team also developed an open-source “Inverse Designer Tool”, which acts as an advanced database filter, enabling users to explore ionic liquids based on defined criteria, streamlining the identification of promising candidates for various applications.

 

The integration of data-driven models with molecular insights represents a significant advancement in the field of materials science. The team’s approach enhances the efficiency of ionic liquid design and promotes the development of environmentally friendly solvents. By significantly reducing the experimental workload, their system accelerates the adoption of ionic liquids in various industrial fields, from energy storage to pharmaceuticals.

 

The principles demonstrated by the team could also be applied to other complex chemical systems, fostering innovations in material design and environmental sustainability, underscoring the importance of interdisciplinary approaches in tackling research challenges. 

 

Jade Sterling
Science Writer
4 July 2024

Professor Merouane Debbah Recognized for Groundbreaking Research Work on Reconfigurable Intelligent Surfaces

 

Professor Merouane Debbah, Director, 6G Research Center, and Professor, Computer Communication, Khalifa University has been awarded the 2024 IEEE Communications Society Best Tutorial Paper Award by the IEEE Communications Society, honoring his groundbreaking research and advancements in wireless communications.

 

The award-winning paper titled ‘Smart Radio Environments Empowered by Reconfigurable Intelligent Surfaces: How It Works, State of Research, and The Road Ahead’ was published in the IEEE Journal on Selected Areas in Communications in November 2020. It provides a comprehensive overview of Reconfigurable Intelligent Surfaces (RIS) – an emerging wireless transmission technology with the potential to transform future 6G networks.

 

Co-authors of the research paper and notable 2024 IEEE Communications Society Best Tutorial Paper Award recipients include Dr. Marco Di Renzo, Research Director, Centre National de la Recherche Scientifique (CNRS), Dr. Alessio Zappone, Professor, the University of Cassino and Southern Lazio, Dr. Mohamed-Slim Alouini, Al- Professor, King Abdullah University of Science and Technology (KAUST), Dr. Chau Yuen, Associate Professor, Singapore University of Technology and Design (SUTD), Dr. Julien de Rosny, CNRS Senior Scientist with the Institut Langevin, Paris, and Sergei Tretyakov, Professor, Department of Radio Science and Engineering, Aalto University, Finland.

 

The 2024 IEEE Communications Society Best Tutorial Paper Award includes a plaque and an honorarium of up to US$500 for each author. Nominations were shortlisted and finalized by the Magazines and Journals Paper Awards Selection committees with the approval of the Magazines/Journals Editors-in-Chief, Director of Magazines, and Director of Journals.

 

In the award-winning paper, the research team has shown the great potential of RIS in enhancing wireless networks, making it easier to understand and develop these advanced systems. It highlights how RIS use many simple, low-cost antennas or special materials to control wireless signals. Unlike other technologies like phased arrays and relays, which are more complex, RIS is more affordable and can improve signal coverage, capacity, and energy efficiency. 

 

Jade Sterling
Science Writer
4 July 2024

Metal oxyhydroxides represent a promising path towards a sustainable energy future but research needs to optimize synthesis and explore new nanostructures to fully realize their potential 

 

The transition from fossil fuels to renewable energy is a pressing challenge, with climate change, environmental degradation, and resource depletion driving the search for sustainable alternatives. Electrochemical energy conversion and storage (EECS) technologies hold significant promise and central to these are advanced materials that enhance their efficiency and performance. One such group of materials, metal oxyhydroxides (MOOHs), is poised to play a crucial role in the future of energy.

 

In this review article, a team of researchers including Khalifa University’s Dr. Karuppasamy Karuppasamy, under the guidance of Prof. Akram AlFantazi, reported on the advancements in MOOHs, focusing on their synthesis, structural engineering and applications in EECS. Dr. Karuppasamy collaborated with researchers from Gyeongsang National University, South Korea; Vellore Institute of Technology, India; Federal University of Mato Grosso do Sul, Brazil; The Oxford College of Science, India; Chulalongkorn University, Thailand; and Dongguk University-Seoul, Republic of Korea.

 

Their results were published in Coordination Chemistry Reviews, a top 1% journal.

 

Metal oxyhydroxides are a type of transitional metal compound that includes elements like manganese, nickel, iron, and cobalt. These materials have unique electronic structures and variable valence states, which make them particularly effective as electrocatalysts and electrode materials. Their 2D layered structures, comprising edge-sharing octahedral subunits, allow for high conductivity and improved surface texture. Various ions are also inserted into the structure to enhance the material’s overall electrochemical performance.

 

MOOHs are especially promising in supercapacitor applications. Supercapacitors are energy storage devices with high power density, long cycle life, and fast charge-discharge capabilities.

 

“MOOHs such as cobalt oxyhydroxide and nickel oxyhydroxide have shown remarkable performance as supercapacitor electrodes,” Dr. Karuppasamy says. “For instance, cobalt oxyhydroxide exhibits a high specific capacitance and stability due to its mixed valence states, which facilitate excellent reversible redox reactions.”

 

In battery technology, transition metal oxyhydroxides can serve as electrode materials in alkali metal ion batteries, providing high energy density and cycling stability. Their ability to undergo reversible redox reactions makes them ideal for next-generation battery applications, with the potential to surpass traditional materials in performance.

 

MOOHs are also effective catalysts for water electrolysis, a process which produces hydrogen fuel by splitting water into hydrogen and oxygen and is vital for the development of clean hydrogen energy. The MOOH structural properties allow for efficient electron transfer and ion diffusion, making them highly effective for both the hydrogen evolution reaction and the oxygen evolution reaction.

 

Synthesizing MOOHs involves various innovative methods, each with their own advantages. Hydrothermal and solvothermal processes allow for precise control over the nanostructures and chemical compositions, producing high-purity materials with desirable properties, while the sol-gel process produces multicomponent systems at low temperatures. Microwave-assisted synthesis is rapid and energy efficient but requires specialized equipment.

 

Despite their potential, MOOHs face several challenges. Scaling up production while maintaining quality and performance is a significant hurdle. Additionally, ensuring the long-term stability of MOOH-based materials in various electrochemical environments remains a concern.

 

MOOHs could revolutionize the field of EECS but continued research and development will be essential to fully realize their potential.

 

Jade Sterling
Science Writer
3 July 2024

New catalyst offers higher yield and reduced energy consumption to transform ammonia synthesis in the chemical industry

 

Representing a significant step towards more sustainable industrial processes, a team of researchers from Khalifa University and the Indian Institute of Technology Ropar has developed a novel catalyst designed to revolutionize the production of ammonia.

 

Ammonia is a crucial component for fertilizers and a promising carbon-free fuel. The Center of Catalysis and Separations (CeCaS) researchers published their innovative approach in ACS Energy Letters, a top 1% journal. Safa Gaber, Dr. Kayaramkodath Chandran Ranjeesh and Dr. Dinesh Shetty leveraged a covalent organic framework (COF) to efficiently couple two electrochemical reactions, achieving new highs in efficiency and yield.

 

Ammonia is a cornerstone of modern agriculture, essential for the fertilizers that support global food production. It also holds potential as a clean fuel. However, the current method of producing ammonia, the Haber-Bosch process, is highly energy-intensive and accounts for over 2% of annual global CO2 emissions. This has prompted research into greener alternatives, with electrochemical methods emerging as a promising solution.

 

While the nitrogen reduction reaction has been explored, it presents significant challenges of its own. The KU and IIT research team shifted their focus to the nitrate reduction reaction, which offers higher feasibility due to the greater solubility of nitrate ions and their lower bond dissociation energy. Plus, nitrate is a prevalent pollutant in agricultural runoff, meaning their method offers the additional benefit of pollution remediation.

 

The team developed a bifunctional catalyst that integrates ruthenium nanoclusters within a covalent organic framework. This design allows precise control over the diffusion of nitrate and protons, resulting in a highly selective and efficient conversion of nitrate to ammonia.

 

The researchers also coupled the glucose oxidation reaction at the anode of the catalyst, replacing the traditional oxygen evolution reaction. The glucose oxidation reaction requires less energy, significantly reducing the overall energy consumption of the ammonia synthesis process. It also produces valuable by-products such as gluconic and glucaric acids that can be used in other industries.

The novel catalyst achieved a 2.5 times higher ammonia yield rate compared to traditional catalysts, marking a significant step towards sustainable ammonia production. By addressing both pollution and energy efficiency, the bifunctional COF catalyst offers a practical solution for greener industrial processes.

 

Jade Sterling
Science Writer
2 July 2024

Smart cities can harness machine learning for greener grids, revolutionizing urban energy management with renewables 

 

Integrating renewable energy sources and electric vehicles (EVs) into modern cities is a necessary evolution for the urban landscape. However, the variability introduced by these green technologies poses significant challenges for traditional energy management systems.

 

A team of researchers including Khalifa University’s Prof. Ahmed Al-Durra has introduced an innovative approach that could redefine energy management in smart cities. Prof. Al-Durra collaborated with researchers from Politecnico di Milano University, Italy; Islamic Azad University, Iran; Arman Niroo Hormozgan Company, Iran; and Aalborg University, Denmark, to develop an intelligent energy management strategy for networked microgrids (NMGs) in smart cities considering renewable energy source uncertainties and power fluctuations. Their approach leverages a sophisticated combination of technologies including neural networks and deep reinforcement learning algorithms.

 

The team published their research in Sustainable Cities and Societies, a top 1% journal.

 

Smart cities are increasingly turning to NMGs as a solution to enhance energy reliability and efficiency. These microgrids can operate independently or in conjunction with the main power grid and are essential for integrating renewable energy sources and electric vehicles effectively. However, managing them in real-time, considering the unpredictable nature of solar power, for example, and EV battery usage, requires a robust, adaptable solution.

 

The research team’s solution leverages the power of machine learning to manage the active power and frequency of NMGs dynamically. Key to this strategy is its dual structure: offline training for the algorithm and decentralized operation for real-world application. This setup allows for continuous adjustment based on the operational data each microgrid collects, ensuring optimal decisions for frequency and power control.

 

The system can adapt in real-time. Offline training fine-tunes the algorithm’s responsiveness, and the decentralized operation allows for individual microgrids to make autonomous decisions based on local data.

 

The team’s system demonstrated a computation accuracy exceeding 98 percent, significantly outperforming traditional methods, with a 7.82 percent reduction in computation burden and a 61.1 percent decrease in computation time. These enhancements mean that NMGs can operate more smoothly and efficiently, with less downtime and faster responses to changes in energy demand or supply. This is particularly important in urban settings where energy demands can be unpredictable.

 

For urban planners and energy managers, this represents a step towards more sustainable urban energy practices, where green technology integration is efficient and reliable. The potential for scalability and further development opens new pathways for even smarter, more responsive urban energy grids, powered by the capabilities of machine learning.

 

Jade Sterling
Science Writer
26 June 2024