Skip to content

BASP Frontiers 2027

24 – 29 January, Villars-sur-Ollon, Switzerland

General
Important dates
Programme
Calls & Registration

Date

2027-Jan-24

Time

3:20 pm - 5:00 pm

Lightning Keynotes on the Past, Present, and Future of Computational Imaging and Panel Discussion

Moderator: Ulugbek Kamilov

Session Schedule

15:20 - 15:45
Invited talk - The Past, Present, and Future of Computational Imaging
Jean-Christophe Pesquet

15:45 - 16:05
Invited talk - The Past, Present, and Future of Astronomical Imaging
Rick Perley

16:05 - 16:30
Invited talk - The Past, Present, and Future of Medical Imaging
Miki Lustig

16:30 - 17:00
Panel discussion: Challenges and Opportunities at the Interface

PRV Event
NXT Event

© 2026 BASP Frontiers 2027Theme by Puro

Organiser & chair: Daniel Rueckert

Session title: Beyond Passive Reconstruction: Trustworthy Foundation Models and Agentic AI for Autonomous Medical Imaging

Abstract:

While computational imaging has traditionally focused on sophisticated reconstruction algorithms to solve ill-posed inverse problems, a paradigm shift is underway toward generalist foundation models and autonomous, agentic AI. This session explores the transition from passive data processing to active, intelligent imaging systems in medicine.

We will first discuss the development of foundation models capable of transferable representations across highly heterogeneous data. As these models become central to imaging pipelines, ensuring they are trustworthy is paramount. This session will highlight the critical role of uncertainty quantification, exploring how foundation models can move beyond point estimates to provide well-calibrated confidence bounds on their representations. Understanding “what the model doesn’t know” is a fundamental requirement for the safe deployment of AI in clinical diagnostics and precision astronomical measurements.

Building upon these trustworthy foundations, the second half of the session will explore Agentic AI. We will examine how AI can move beyond the post-acquisition reconstruction pipeline to drive the acquisition process itself. By leveraging the uncertainty estimates generated by foundation models, agentic systems can autonomously decide which measurements will maximize information gain. Examples include agentic MRI scanners that dynamically adapt pulse sequences in real-time based on patient-specific physiological feedback, or autonomous telescopes optimizing observation strategies. Ultimately, this session provides a perspective on the future of imaging: closed-loop, agent-driven systems where trustworthy foundation models dictate not just how an image is formed, but what measurements are optimal to acquire in the first place.

Organiser & chair: Jonathan Tamir

Session title: From Reconstruction to Practice: Integrating Physics and Learning in Medical Imaging

Abstract:

This session examines the evolving landscape of computational medical imaging, focusing on the progression from algorithmic advances in image reconstruction to their translation to clinical practice. Modern approaches combine model-based formulations with data-driven learning, enabling flexible, high-performance solutions to imaging inverse problems under realistic constraints.

The session will highlight recent developments across this spectrum, including scalable learning-based reconstruction methods, adaptive and data-driven priors, and approaches that account for deviations from idealized acquisition models in real-world settings.

In parallel, the session will consider the challenges in translating these methods to clinical practice, including generalization across populations and systems, integration into existing workflows, and alignment with clinical and operational requirements. By bringing together perspectives from signal processing, machine learning, and medical imaging practice, the session aims to provide a cohesive view of how advances in computational medical imaging are shaping the transition from reconstruction to practice.

Organiser & chair: He Sun

Session title: Computational Microscopy: Physics, Learning, and Inference across Biological Scales

Abstract:

Computational microscopy is rapidly transforming biological imaging from a hardware-limited measurement process into a joint inference problem over optics, samples, algorithms, and biological priors. Modern microscopes increasingly rely on structured illumination, adaptive optics, ptychography, wavefront shaping, light-field acquisition, matrix imaging, and learned reconstruction to recover high-dimensional biological information beyond the native limits of resolution, depth, field of view, speed, and photon budget.

This session will bring together leaders in optical microscopy, wave physics, computational imaging, and bioimage AI to discuss how physics-based forward models, optimization, statistical inference, and modern machine learning are reshaping microscopy. The scope spans quantitative phase imaging, Fourier ptychographic microscopy, adaptive optical microscopy, imaging through scattering media, deep-tissue fluorescence imaging, smart microscopy, and AI-enabled image restoration and analysis. A central theme is how to make computational microscopy reliable for biological discovery: how to co-design optics and algorithms, integrate learned priors without hallucination, quantify uncertainty, and scale methods from controlled phantoms to complex living systems.

By positioning microscopy as a multidisciplinary inverse problem, the session will connect the BASP communities in computational imaging theory and biomedical imaging, while also highlighting methodological links to astronomical imaging, including adaptive optics, phase retrieval, wavefront sensing, and imaging through turbulence or scattering. The session aims to stimulate discussion on the next generation of microscopes that are not merely image acquisition devices, but closed-loop, physics-aware, AI-assisted scientific instruments.

Organiser & chair: Ivan Dokmanic

Session title: Fast Neural Physics Surrogates for Forward and Inverse Problems in Imaging

Abstract:

Physics is clearly central to imaging. Modern deep learning approaches to reconstruction indeed incorporate imaging physics in various ways. In the adjacent AI4Science community, there has recently been rapid progress on learning surrogate models for physics, including neural operators and related architectures that approximate mappings between function spaces associated with forward and inverse models used in imaging. These methods have shown exciting results for wave propagation, transport, and other physically grounded settings, and they intersect with core problems in computational imaging.

The premise of this session is that there is significant opportunity in bringing together researchers developing fast learned surrogates for complex physics with researchers advancing the next generation of computational imaging. The session will focus on that interface. We will survey recent developments in neural operators and related learned physics modules for forward and inverse modeling in imaging, with particular emphasis on fast, scalable surrogates that can be embedded within reconstruction pipelines. We will look at methods that generalize across resolution, sampling patterns, and sensor geometries, and that support challenging imaging regimes. Emphasis will be on signal-processing structure, physical fidelity, and computational efficiency. While neural operators will provide the central lens, the session will more broadly consider learned forward surrogates, and modern modality-aware neural architectures for imaging.

Organiser & chair: Mario Figueiredo

Session title: Algorithms for Imaging Inverse Problems: From Splitting to Learning

Abstract:

This session is devoted to algorithms for imaging inverse problems built on learned components. The basic observation is that the classical structure, an analytical prior handed to an optimizer, has dissolved: the prior is now implicit in a learned denoiser, score, or operator. From this point the field followed essentially two roads. One keeps the operator-theoretic view of classical splitting to retain convergence to a well-defined solution. The other road builds on denoising diffusion and related score-matching and flow-matching tools, accessing the prior only through sampling, shifting from a point solution to posterior samples and trading formal guarantees for distributional expressiveness. The presentations in the session will travel both roads, spanning method design and the connection back to optimization-based approaches.

Organiser & chair: Thomas Pock & Martin Holler

Session title: Generative Models for Imaging

Abstract:

The workshop session “Generative Models for Imaging” will explore recent advances in generative approaches for modern imaging problems, with a particular focus on inverse problems in computer vision and medical image reconstruction. The session will highlight how Bayesian modeling, energy-based methods, and deep neural architectures can be combined to design robust and interpretable models. Emphasis will be placed on generative priors, diffusion-inspired models, and sampling algorithms that enable high-quality reconstruction, synthesis, and uncertainty quantification. Participants will gain insights into modern energy-based formulations, Langevin dynamics, and accelerated stochastic sampling techniques that bridge optimization and probabilistic inference. The session will further discuss scalable algorithms for inverse problems such as MRI reconstruction, image denoising, and computational imaging. By integrating mathematical modeling with state-of-the-art machine learning, the workshop aims to provide a unified perspective on generative imaging methodologies. Through a combination of theoretical concepts and application-driven examples, the workshop will showcase how principled generative learning can advance imaging science across different domains.

Organiser & chair: Berthy Feng & Katie Bouman

Session title: Physics-informed Machine Learning for Astronomy and Astrophysics

Abstract:

Many inference problems in astronomy and astrophysics involve recovering hidden structure from limited sensor measurements. ML methods can fill in the gaps, but purely data-driven approaches fall short when training data is limited or measurements are extremely sparse. Incorporating physics knowledge can be a principled way to impose structure on especially challenging inference problems.

This session focuses on the role of physics-informed ML across a range of imaging and inference problems in astronomy and astrophysics. Topics will include dark matter mapping (Laurence Perreault Levasseur), data-driven priors for astronomical problems (Francois Lanusse), etc. Together, these contributions illustrate how physics-informed ML brings previously invisible astrophysical phenomena into view for scientific research.

Organiser & chair: Aviad Levis

Session title: Imaging Beyond 2D: Time, Spectra, 3D Structure and Vector Fields from Sparse Astronomical Data

Abstract:

Modern astronomical imaging is evolving beyond static two-dimensional reconstructions toward the recovery of high-dimensional structure from sparse and indirect measurements. Rather than producing images alone, emerging approaches aim to reconstruct temporal variability, hyperspectral structure, three-dimensional emission, and vector-valued physical fields directly from observational data.

This session focuses on computational imaging methods that enable such high-dimensional reconstruction across complimentary axes including time-resolved imaging of dynamical systems, hyperspectral data cubes, three-dimensional imaging, and the recovery of magnetic structure. In all of these settings, the underlying challenge is the same: to infer complex, latent structure from incomplete and heterogeneous data using forward models, inversion techniques, and increasingly, learned priors.

By bringing together advances from black hole imaging, spectral-line interferometry, solar polarimetric inference, and neural-field-based tomography, this session aims to identify common principles in modern high-dimensional inverse problems. These include handling sparse sampling, enforcing physical consistency, and disentangling line-of-sight and multi-scale effects. The goal is to provide a unified perspective on how computational imaging is enabling the reconstruction of dynamic, multi-physical structure in the universe.

Organiser & chair: Arwa Dabbech

Session title: Advances in Radio-Interferometric Image Formation, and Scalable Data Processing

Abstract:

This session provides an overview of current and emerging approaches to radio-interferometric imaging in Astronomy throughout the data life cycle, from acquisition and correlation to storage and image formation. The session will open with a tutorial introducing the challenges of image formation in radio interferomety and state-of-the-art image formation methods. The invited talks will then cover: (i) Bayesian and machine-learning frameworks for joint image reconstruction, source separation, and uncertainty quantification, (ii) scalable real-time imaging including FPGA- and GPU-based direct imaging correlators for next-generation aperture arrays such as SKA, (iii) data dimensionality reduction approaches for LOFAR and SKA through visibility compression. The two Invited posters will focus on transient imaging and SETI with MeerKAT and a data dimensionality reduction technique for scalability throughout the data life cycle via a random-phased array approach.

Medical Imaging in a Low-resource Setting

Organiser & chair: Andrew Webb (Chair) & Florian Knoll

Abstract:

This session will explore different challenges associated with medical imaging in low resource settings. MRI, X-ray/CT and ultrasound image requirements will be compared with one another, as well as issues of data transmission and interpretation. The different potential roles of AI in image acquisition, processing and diagnosis will be a common theme throughout.

Medical Imaging in a High-resource Setting

Organiser & chair: Daniel K. Sodickson

Abstract:

If we had all the hardware power, compute resources and data analysis techniques we could dream of, what could we achieve in medical imaging? What fundamental scientific or clinical questions could we answer? This session will explore advanced imaging capabilities which are currently being enabled or may soon be enabled by emerging hardware platforms, software tools, and datasets. Speakers will share a focus on new information that can be gleaned from MRI, PET, CT, ultrasound, and other imaging or sensing modalities. They will also comment on which developments may eventually translate to, or otherwise influence, the low-resource settings to be explored in more detail in a companion session.

“R4” Reconstruction, Resolution, Regularization and Representation - Overcoming Medical Imaging Challenges

Organiser & chair: Julia Schnabel

Abstract:

This session explores the critical “R4” challenges—Reconstruction, Resolution, Regularization, and Representation—that are central to advancing medical imaging technologies. With a strong focus on how novel computational methods, including machine learning and physics-informed techniques, can address longstanding limitations in image quality, acquisition time, and motion artifacts, this session highlights innovative approaches across a range of medical imaging modalities, including MRI, PET, and photoacoustic tomography (PAT), demonstrating synergistic opportunities for overcoming common imaging challenges.

Foundation and Multimodal Models in Computational Imaging

Organiser & chair: Jong Chul Ye

Abstract: Recently, foundation models, which are large scale models pretrained on massive amounts of diverse data in a self-supervised manner, have been quickly replacing the existing CNN based end-to-end and/or supervised-learning approaches due to their superior performance. As a result of large-scale self-supervised pre-training, foundation models have excellent generalization capabilities with intriguing emergent properties. This session targets recent advances in foundation models for computational imaging, such as text+image models, as well as diffusion models for inverse problems, to understand their advantages and opportunities.

On the Interface of Optimization and Deep Learning for Computational Imaging

Organiser & chair: Audrey Repetti

Abstract:

This session focuses on computational imaging methods at the interface of optimization and deep learning. Recent advances aiming at pairing these two areas has led to significant advancements in image reconstruction and enhancement. Optimization techniques traditionally provided mathematically rigorous solutions for imaging problems, but they often struggled with high-dimensional data and non-linearities. Deep learning, particularly neural networks, has transformed this landscape by learning complex mappings from data, enabling faster and higher expressivity in image solutions. When combined, these approaches harness optimization algorithms to refine and guide the training of deep networks, resulting in robust models that are highly efficient for performing computational imaging tasks.

Scalable Interferometric Imaging in the Big Data Era

Organiser & chair: Stefan Wijnholds (Chair) & Kazunori Akiyama

Abstract:

With the development and deployment of LOFAR2.0, SKA and ngVLA, radio interferometry is entering the Big Data Era. To keep up with the data deluge that will be generated by these instruments with an acceptable costing envelope for computing hardware and energy consumption, at least an order of magnitude improvement in computational performance needs to be realized. Also, data processing pipelines will need to scale well and use processing components optimized for energy efficiency. These challenges require us to rethink the way we process radio interferometric data, opening opportunities for exploration of new avenues. This session aims to present an overview of the challenges and provide a forum for discussion of new ways to process radio interferometric data.

Computational Imaging for Precision Astrophysics

Organiser & chair: Aviad Levis / Marie Ygouf

Abstract:

The topic of this session is Computational Imaging for Precision Astrophysics, covering a variety of exciting new approaches for ultra high resolution astronomy. The session focuses on black holes and exoplanets both of which with their unique challenges. Horizon scale black hole science is incredibly challenging due the extreme resolution required (~20 micro arcseconds) to resolve the largest black holes on the sky: Sagittarius A* and M87*. Nonetheless, innovations and advancements in interferometry over the last decade have opened the door to new horizon scale science: from the first images with the Event Horizon Telescope, to precise astrometry with GRAVITY. The talks and posters in this session cover a variety of topics from novel computational imaging that integrates interferometry with information theory and modern machine learning to new mission concepts for space interferometry. For exoplanets, beyond resolution, a key challenge is that of contrast: the signal from the planet could be up to a billion times fainter than that of its host star. Nonetheless, direct imaging holds the key to characterizing composition and habitability. In this session, we will hear about advancements in the field and how precise modeling of instrumental optics could be the key to pushing the detection limit towards the fundamental noise.

Uncertainty Quantification in Computational Imaging

Organiser & chair: Mujdat Cetin

Abstract:

This session explores recent advances in the theory and methods of uncertainty quantification for computational imaging. As computational imaging involves solving ill-posed inverse problems using complicated estimators, quantifying uncertainties associated with formed imagery, including both aleatoric (stochastic) and epistemic (systematic or modeling) uncertainties, has been an important topic of interest. Recent emergence of deep learning-based image formation methods, including generative models, has both increased the need for proper uncertainty quantification with an eye towards trustworthiness, and also provided tools that can possibly enable progress in that direction. This session highlights a variety of statistical and learning-driven recent work in this area, including estimation and visualization of epistemic and aleatoric uncertainties, posterior variance-based error quantification, distribution-free uncertainty quantification using conformal prediction, as well as task-driven uncertainty quantification for computational imaging, among others.

High-dimensional Bayesian Astronomical Imaging and Inverse Problems with Machine Learning

Organiser & chair: Laurence Perreault Levasseur

Abstract:

Recent progress in machine learning and generative modeling has opened new avenues to tackle previously insoluble high-dimensional inverse problems in astronomy and astrophysics, particularly in Bayesian image reconstruction. While these methodologies show great promise in a range of applications from field-level cosmology to differentiable optics systems, multiple open problems stand in the way of groundbreaking discoveries. This session will explore recent applications and proposals addressing the development of computationally tractable methodologies to reconstruct posterior samples in imaging problems. We will also discuss assessing their accuracy in real-world settings and addressing the problem of robustness to distributional shifts, which remains a key focus of current research.

Lightning Keynotes on Modern Perspectives on Imaging and Panel Discussion

14:50 – 15:15:

Speaker: Greg Zaharchuk

Topic: Perspectives on the past, present, and future of medical imaging

Abstract: TBC

15:15 – 15:40:

Speaker: Jean-Luc Starck

Topic: Perspectives on the past, present, and future of astronomical imaging

Abstract: TBC

15:40 – 16:05:

Speaker: Carola Schoenlieb

Topic: Perspectives on the past, present, and future of computational imaging

Abstract: TBC

16:05 – 16:30:

Panel Discussion

Topic: Challenges and opportunities at the interface

Large-scale Optimisation for Computational Imaging

Organisers: Gitta Kutyniok (Chair) & Ulugbek Kamilov (Committee member)

Abstract:

Large-scale optimization problems arise in a variety of imaging tasks. Examples include dictionary learning, low-rank matrix recovery, blind deconvolution, and phase retrieval. Conventional approaches for solving many of these optimization problems involve designing algorithms that can effectively leverage a wide-variety of structural constraints.

This session will provide an excellent opportunity for the wider signal processing and imaging community to come together and share recent developments, open challenges, and future directions in large-scale optimization methods suitable for analyzing imaging data.

Medical Imaging in Low-Resource Settings

Organisers: Nicholas Durr (Chair) & Audrey Bowden (Committee member)

Abstract:

Despite significant progress in modern healthcare, state-of-the-art healthcare in first-world environments differs significantly from that in low-resource settings. Constraints on cost, size, usability and environmental stability pose interesting challenges to engineer suitable solutions in these spaces. This session will bring together impactful research that showcases innovative strategies for imaging that are suitable for implementing in low-resources settings such as rural areas, primary care centers and low- and middle-income countries.

Physics Informed Machine Learning in Astronomy

Organisers: Francois Lanusse (Chair) & Jean-Luc Starck (Committee member)

Abstract:

Machine Learning had a significant success in Astronomy in recent years, and it becomes obvious that useful applications of ML require a tight connection to physical modelling. In this session, we will explore several aspects of imbuing physics as part of a ML model, from building hybrid models that merge both deep learning and physical models, to using known physical symmetries and equivariances to design dedicated neural architectures.

Potential Pitfalls of Deep Learning in Medical Image Reconstruction

Organisers: Efrat Shimron (Chair) & Florian Knoll (Committee member)

Abstract:

This session will focus on scenarios in which deep learning algorithms developed for medical imaging might produce unreliable results, e.g. due to distribution shifts, bias, hallucinations, or other factors. The session is planned following the increasing interest in studying sensitivities and instabilities of such algorithms. The talks will discuss strategies for exposing algorithmic sensitivities and addressing them, preventing inverse crimes, and increasing algorithmic interpretability. The aim of the session is to raise awareness to the growing problem of unreliable AI performance in the context of medical imaging, suggest guidelines and solutions, and invoke community discussions.

Astronomical Imaging with Optical Telescopes: From thirty-meter Observatories to future space missions

Organisers: He Sun & Katie Bouman (Committee member)

Abstract:

Large optical telescopes have a rich history of pushing the boundaries of imaging technology and astronomical science. With a series of revolutionary optical telescopes launched recently or planned to be launched soon, in the next decade we expect to discover astronomical phenomena that will redefine our view of the universe. These telescopes will open windows to studying galaxies in the early universe, Earth-like planets around distant stars, and unknown discoveries beyond our imagination. This session will review recent development of computational imaging techniques in large space- or ground-based optical observatories. Topics include adaptive optics for future space telescopes (e.g. RST) and 30m level ground telescopes (e.g. E-ELT), data processing and early results from JWST, and machine learning methods for gravitational lensing and exoplanet study.

Modern Regularisation

Organisers: Thomas Pock (Chair) & Philip Schniter (Committee member)

Abstract:

The availability of expressive regularizers is a very important component in solving ill-posed inverse problems in imaging. In recent years, hand-designed regularizers have been gradually replaced by data-driven ones. Provided sufficient training data is available, it is nowadays possible to learn tailored regularizers for a certain problem class. This usually leads to a huge increase in reconstruction quality, but the learned regularizers are usually much harder to analyse and it is much harder to give guarantees on convergence behaviour, generalisation ability, or reconstruction error. In this session, we will present and discuss the latest methods, techniques and applications in this cutting-edge field of research.

Medical Image Reconstruction and Analysis

Organisers: Kerstin Hammernik (Chair) & Julia Schnabel (Committee member)

Abstract:

The aim of this session is to guide the audience through the medical imaging pipeline – from acquisition and reconstruction to analysis – with a little help from AI. The talks will cover different image modalities, and focus on inverse problems in medical image reconstruction, motion estimation and motion correction. As image quality and accurate diagnostic assessment are key for the applicability of AI-based solutions in the medical context, the talks cover these aspects and provide insights into the uncertainty of AI-based algorithms in image reconstruction and analysis.

Frontier of Interferometric Imaging from SKA to ngEHT

Organisers: Kazunori Akiyama (Chair) & Yves Wiaux (Committee member)

Abstract:

Computational imaging is a key process in radio and optical/near-infrared interferometry to reveal the fine views of the universe from observational data taken in Fourier space. Over the last decade, significant progress has been made in the development of new computational imaging techniques to address and overcome various challenges brought by the advent of the new instruments including the Event Horizon Telescope (EHT), Low Frequency Array (LOFAR), MeerKAT and Very Large Telescope Interferometer (VLTI), as well as upgrades of existing facilities such as Very Large Array (VLA). Many algorithmic and data processing challenges arise in our quest to endow these instruments, set to observe the sky at new regimes of sensitivity and resolution, with their expected acute vision. In this new era, imaging encompasses not only forming 2D spatial maps of observed fields of view, but also the reconstruction of the spectrum, polarisation, and dynamics of the sources of interest, not to mention the mapping of underlying physical quantities. This focused session will gather interferometric imaging experts to review the exciting frontiers in the field.

Inference and Calibration

Organisers: Ivan Dokmanic (Chair) & Philip Schniter (Committee member)

Abstract:

Generative models are experiencing a second youth in imaging and scientific inference. New ideas include injective models for sampling high-dimensional posteriors, theoretical advances on statistical, approximation-theoretic, and topological questions, generating continuous functions which dovetail with the downstream PDE solvers, and creative uses of generative models to probe performance limits of inference systems. The session “Generative Inference and Calibration” brings together prominent researchers spearheading these exciting new directions.