Staff Profile

I am a Senior Lecturer (Associate Professor) in Signal Processing and Machine Learning at Newcastle University, where I lead an exciting portfolio of interdisciplinary projects spanning digital health, financial risk modelling, smart sensing, non-destructive evaluation, and structural health monitoring.

From building AI digital twins that mimic expert urologists’ decision making, to designing AI powered diagnostic tools for preserving historical artifacts and building digital clone of the UK economy for real-time financial risk forecasting, my research integrates AI, signal processing, and physics-informed models to deliver explainable, real-time AI solutions that address high-impact societal and industrial challenges. Whether we're capturing hidden corrosion in industrial pipelines using smart RFID, restoring centuries-old artifacts with AI-powered diagnostics, or reconstructing 3D anatomy from low-quality medical scans, my core vision is to build intelligent systems that are not only accurate and interpretable, but also fast, scalable and meaningful to those who use them—clinicians, curators, policy-makers, and the wider public.

My work is distinguished by a commitment to lightweight, interpretable AI, with a strong emphasis on cross disciplinary innovation that bridges engineering, medicine, and the humanities to tackle long-standing challenges. Through cross-disciplinary partnerships, I explore unconventional AI applications to tackle complex challenges that remain largely unexplored or underserved by current technologies. This means working side by side with clinicians, historians, engineers, and economists alike, since solving complex problems demands bringing diverse minds together.

Below is a snapshot of my research projects:

Digital Twin for Bladder Cancer Risk Prediction (Lead; in collaboration with Imperial College, the Institute of Cancer Research, and the NHS, 2022–2026)

Simulating an expert urologist through an AI digital twin - an explainable AI model that delivers patient-specific, trustworthy recurrence predictions for non-muscle invasive bladder cancer using routine clinical data. The goal is to combine transparency, simplicity, and scalability to replace outdated cancer risk tools in clinical practice.

SONNET: Scalability Oriented Novel Network of Event Triggered Systems (Co-Investigator and work package lead: EPSRC Programme Grant, 2023-2028)

Cloning a real-time digital twin of the UK financial system—a distributed, event-triggered AI framework that runs in parallel with live economic data to continuously model emerging risks and systemic shifts. By rethinking conventional approaches to computation, learning, and modelling, this digital twin will extract and synthesise key financial signals at the edge, generating low-latency, high-impact insights for national-scale risk forecasting and decision support.

UNVEIL: Unified Non-destructive Evaluation of Historical Artifacts (Newcastle PI: HORIZON EUROPE Grant, 2026-2029)

Developing a next-generation diagnostic framework for cultural heritage (CH) preservation using advanced, non-destructive digital twins of historical artifacts. Integrating cutting-edge surface and subsurface evaluation techniques, such as terahertz imaging, thermography, photoacoustics, and ultrasonics, with power of AI and machine learning, these models will capture comprehensive, multi-physics insights into the condition, composition, and provenance of CH objects. Novel ML-driven algorithms for feature extraction, image fusion, and material classification will enable high-resolution diagnostics while reducing costs and inspection complexity. These digital twins will serve as powerful tools to support conservation decisions, enhance historical understanding, and engage wider audiences. By fostering close collaboration between engineering and humanities disciplines across Europe, the project seeks to bridge long-standing gaps in CH diagnostics and establish a scalable, multidisciplinary model to protect irreplaceable artifacts.

AI-Enabled UHF RFID for Corrosion Monitoring (PhD Project, Primary Supervisor, 2023-2027)

Designing a low-cost, IoT-integrated RFID system for real-time characterisation of corrosion under coatings in industrial pipelines. By combining multilayer RFID sensors with machine learning-based feature extraction and antenna optimisation, this work pioneers scalable, intelligent NDE for infrastructure health.

Electromagnetic SHM in Complex Aerospace Geometries (PhD Project, Primary Supervisor, 2024-2028)

Creating a miniaturised, flexible eddy current probe for in-situ detection of fatigue cracks and stress patterns in aerospace components. Leveraging nonlinear feature extraction and sparse ML models, this project addresses the challenge of monitoring structural integrity in curved, anisotropic geometries.

Generative AI for 3D Medical Image Reconstruction (PhD Project, Primary Supervisor, 2023-2028)

Applying state-of-the-art generative models (e.g. GANs, diffusion models, NeRF) to produce high-fidelity 3D images from sparse or low-quality 2D inputs. The project focuses on building explainable, resource-efficient AI solutions to advance diagnostic accuracy, especially in data-constrained healthcare settings.

Qualifications

2013-2019: PhD in Real-time Estimation of Physiological Tremor for Robotic Handheld Surgical Instruments, Newcastle University, UK

My PhD study was in biomedical signal processing that was directed to enhance performance of the hand-held surgical robots during vitreoretinal microsurgeries. My PhD research was a fine blend of signal processing, multidimensional and quaternion modelling, and machine learning, which was aimed to offset the undesired vibration from the tip of the hand-held surgical robots.

2008-2010: Postgraduate Certificate in Education (PGCE), Huddersfield University, UK

2006-2007: MSc in Optoelectronics and Communication Systems, Northumbria University, UK

2000-2004: BEng in Electronics and Communication Systems, Institute of Engineering (IOE), Pulchowk Campus, Tribhuvan University, Nepal

Roles and Responsibilities within the School

2021 - : Degree Programme Director for MSc Communications and Signal Processing programme

2019 - 2021: Stage 1 Tutor for undergraduates studying electrical & electronic degree programmes

Memberships

Fellow of Higher Education Academy (HEA)

Institute of Electrical and Electronics Engineers (IEEE)

Previous Positions

2008-2010: Lecturer, South Tyneside College, South Shields, UK

2004-2006: Lecturer, Apex College Nepal

Research

I am passionate about developing AI systems that are not only high-performing, but also interpretable, deployable, and aligned with real-world needs—whether in the NHS, finance, or global infrastructure. My research is inherently collaborative, bringing together engineering, medicine, and computer science to drive the next generation of intelligent sensing and decision-making technologies.

My Google Scholar profile can be viewed here.

Research Grants

2026 - 2029, Newcastle PI, UNVEIL: Unified Non-destructive Evaluation of Historical Artifacts, HORIZON EUROPE Grant, £350k

2023 - 2028, Co-I and work package lead (~570k), SONNETS: Scalability Oriented Novel Network of Event Triggered Systems, EPSRC Programme Grant, £6.5m (in collaboration with Imperial College and University of Southampton)

2022 – 2026, PI, Mimicking the Expert: A Trustworthy Digital Twin for Patient-Specific Bladder Cancer Risk Assessment, EPSRC DTP Award, £100k

Current PhD Students

Saram Abbas (Primary Supervisor, 2022 – ) – Mimicking an Expert Urologist: An Explainable AI Digital Twin for Patient-Specific Bladder Cancer Risk Prediction (in collaboration with Imperial College, the Institute of Cancer Research, and the NHS) 

Bladder cancer affects nearly 10,000 people annually in the UK, with around 400,000 cases globally. Most are non-muscle invasive bladder cancers (NMIBCs), treated with transurethral endoscopic resection. For most bladder cancer patients, treatment is just the start of a relentless cycle. NMIBC which accounts for 75% of cases—recurs in 80% of patients, often repeatedly. Each recurrence demands invasive cystoscopies, expensive scans, and lifelong monitoring, placing a severe and ongoing burden on both patients and healthcare systems. In the UK alone, post-treatment surveillance costs the NHS over £200 million each year; globally, the burden runs into billions—while many patients in low-income regions receive no follow-up at all. Yet clinicians are still guided by outdated risk tools like EORTC and CUETO—built decades ago on narrow, homogeneous datasets. These models lack personalisation, transparency, and clinical adaptability. The result? Patients are over- or under-monitored, clinicians are left without clear direction, and lives remain in limbo while systems waste precious resources. 

This research aims to develop an explainable AI “digital twin” that mirrors the decision-making of an expert urologist—a model that evaluates each patient’s routine clinical data individually to flag high-risk factors and deliver reliable, personalised recurrence predictions. Our goal is to move beyond outdated risk calculators and create a globally scalable tool that brings expert-level insight directly to the point of care. Unlike many AI tools that depend on complex and costly inputs (e.g. MRI, genomics, multi-omics), our model prioritises simplicity and precision, using only routinely collected variables such as age, tumour size, post-operative notes, and lifestyle factors—available in 90% of healthcare settings worldwide. The model is trained on real patients’ data across 20+ NHS hospitals, forming one of the UK’s most comprehensive prospective datasets on NMIBC. These data reflect real-world care across diverse patient profiles, including age (25–90+), tumour grade, smoking history, treatment pathways, and socioeconomic background. Our digital twin combines accuracy, interpretability, and seamless NHS integration—delivering a scalable alternative to outdated tools. With few AI models tackling this challenge, it uniquely balances predictive power, transparency, and real-world readiness.

This is a multidisciplinary collaboration between the School of Engineering, Newcastle University, the Faculty of Medicine, Imperial College, the Institute of Cancer Research, UK and the NHS, UK.

Peilin Hui (Primary Supervisor, 2023 – ) – Undercoating Corrosion Characterization Using Ultra-High Frequency Radio Frequency Identification

This study aims to develop a novel approach for charcterising Corrosion Under Coating (CUC) in pipelines using Ultra-High Frequency Radio Frequency Identification (UHF RFID). A key objective is to design an innovative, low-cost, real-time, and remote sensing system based on commercial off-the-shelf RFID tags. The project focuses on developing a multi-layer UHF RFID sensor with high sensitivity to corrosion beneath coatings, integrated with a compact, IoT-enabled monitoring system. Machine learning algorithms will be employed for automated feature extraction, corrosion pattern recognition, and predictive analytics. By harnessing machine learning capabilities, the system can adaptively learn from sensor data, facilitating intelligent, real-time corrosion assessment and long-term structural health monitoring. To enhance robustness against environmental interference, we have introduced a response map-based measurement system, utilising RSSI values across different frequencies and power levels to capture impedance variations. These maps are processed using Principal Component Analysis (PCA) to enable reliable and simplified detection. Additionally, an AI-driven design framework will be proposed to automate antenna optimisation, making UHF RFID sensor development faster, more accessible, and scalable across various non-destructive evaluation applications.

Wenjie Li (Primary supervisor, 2024 – ) – Electromagnetic Non-Destructive Testing for Structural Health Monitoring of Complex Aerospace Geometries

This study aims to develop advanced electromagnetic non-destructive testing (NDT) techniques for real-time structural health monitoring (SHM) of fatigue cracks in propellers with complex aerospace geometries. The objective is to design an innovative flexible eddy current probe integrated with a miniaturised onboard stress monitoring system, based on an inductance-to-digital converter (LDC) chip. This system is designed to detect stress-induced variations in electrical conductivity and magnetic permeability, arising from piezoresistive and inverse magnetostrictive effects, thereby enabling high-sensitivity, high-resolution stress assessment without the need for bulky and expensive vector network analysers. The tunable resonance frequency of the inductive coil, combined with enhanced lift-off tolerance, ensures accurate measurements even on curved surfaces. In addition to leveraging low-cost sensor technology and wireless sensor networks for NDT and SHM, the approach integrates advanced machine learning techniques. Nonlinear feature extraction methods such as Principal Component Analysis, data fusion, and compressed sensing are combined with sparse machine learning classifiers to tackle the signal complexity caused by surface curvature and multi-directional stress. Our prototyped eddy current testing system has already demonstrated its capability to identify conductivity variations and detect both surface and sub-surface cracks, confirming its potential for in-situ SHM of aerospace components.

Jiaqi Huang (Primary Supervisor, 2024 – ) – Leveraging Generative AI for High-Fidelity 3D Medical Image Reconstruction

This multidisciplinary project focuses on leveraging generative AI for high-fidelity 3D medical image reconstruction. It aims to address major limitations of traditional imaging—such as noise, artefacts, radiation exposure, and high costs—by applying cutting-edge generative models including GANs, VAEs, diffusion models, NeRF, and 3D Gaussian Splatting. These approaches are expected to improve image clarity and anatomical accuracy while reducing reliance on repeated scans and extensive manual annotations. Driven by the clinical need for safer, more efficient, and accurate imaging tools, the research explores how AI can bridge the gap between limited 2D inputs and rich 3D representations. To enhance generalisation and computational efficiency, techniques such as attention mechanisms, self-supervised learning, and model pruning are incorporated. The ultimate goal is to develop robust, explainable AI models capable of generating precise 3D reconstructions even in data-scarce, resource-limited settings—advancing diagnostic safety and improving patient outcomes.

Undergraduate

• EEE2010 : Electrical Engineering II
• ENG1002:  Sustainable Design Creativity and Professionalism  

Previous

• EEE8074-MSc
• Coursework
• Stage 1 Labs
• Stage 2 Labs
• EEE2008-Buggy Project Supervision

• Articles

  • Abbas S, Shafik R, Soomro N, Heer R, Adhikari A. AI predicting recurrence in non-muscle-invasive bladder cancer: systematic review with study strengths and weaknesses. Frontiers in Oncology 2025, 14, 1509362.
  • Cavus M, Allahham A, Adhikari K, Giaouris D. A hybrid method based on logic predictive controller for flexible hybrid microgrid with plug-and-play capabilities. Applied Energy 2024, 359, 122752.
  • Cavus M, Ugurluoglu YF, Ayan H, Allahham A, Adhikari K, Giaouris D. Switched Auto-Regressive Neural Control (S-ANC) for Energy Management of Hybrid Microgrids. Applied Sciences 2023, 13(21), 11744.
  • Cavus M, Allahham A, Adhikari K, Zangiabadi M, Giaouris D. Energy Management of Grid-Connected Microgrids using an Optimal Systems Approach. IEEE Access 2023, 11, 9907-9919.
  • Makarfi AU, Rabie KM, Kaiwartya O, Adhikari K, Nauryzbayev G, Li X, Kharel R. Toward Physical-Layer Security for Internet of Vehicles: Interference-Aware Modeling. IEEE Internet of Things Journal 2020, 8(1), 443-457.
  • Adhikari K, Tatinati S, Veluvolu KC, Chambers JA. Physiological tremor filtering without phase distortion for robotic microsurgery. IEEE Transactions on Automation Science and Engineering 2020, n/a, n/a.
  • Rathore RS, Sangwan S, Prakash S, Adhikari K, Kharel R, Cao Y. Hybrid WGWO: whale grey wolf optimization-based novel energy-efficient clustering for EH-WSNs. Eurasip Journal on Wireless Communications and Networking 2020, 2020, 101.
  • Wang Y, Tatinati R, Adhikari K, Huang L, Nazarpour K, Ang WT, Veluvolu KC. Multi-step prediction of physiological tremor with random quaternion neurons for surgical robotic applications. IEEE Access 2018, 6, 42216-42226.
  • Adhikari K, Tatinati S, Ang WT, Velovulu KC, Nazarpour K. A quaternion weighted Fourier linear combiner for modeling physiological tremor. IEEE Transactions on Biomedical Engineering 2016, 63(11), 2336-2346.

• Conference Proceedings (inc. Abstracts)

  • Adhikari K, Tatinati S, Veluvolu KC, Chambers JA, Nazarpour K. Real-time physiological tremor estimation using recursive singular spectrum analysis. In: Proceedings of the 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 2017, Seogwipo, South Korea: IEEE.
  • Adhikari K, Tatinati S, Veluvolu KC, Nazarpour K. Modeling 3D Tremor Signals with a Quaternion Weighted Fourier Linear Combiner. In: 7th Annual International IEEE/EMBS Conference on Neural Engineering (NER 2015). 2015, Montpellier, France: IEEE.
  • Adhikari K, Tatinati S, Veluvolu KC, Nazarpour K. Improvement in modelling of physiological tremor by inclusion of grip force in quaternion weighted Fourier linear combiner. In: 2nd IET International Conference on Intelligent Signal Processing. 2015, London: IET.