A Modular Approach to Stochastic Optimisation for Inverse Problems Using the Core Imaging Library
Evangelos Papoutsellis, Margaret A. G. Duff, Jakob S. Jørgensen, Sam Porter, Claire Delplancke, Gemma Fardell, Edoardo Pasca, Kris Thielemans · 2026
Read A Modular Approach to Stochastic Optimisation for Inverse Problems Using the Core Imaging Library by Evangelos Papoutsellis, Margaret A. G. Duff, Jakob S. Jørgensen, Sam Porter, Claire Delplancke, Gemma Fardell, Edoardo Pasca, Kris Thielemans (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-21230.txt.
Introduction
Papoutsellis and co-authors expand the Core Imaging Library with a modular framework for stochastic optimisation in inverse problems. The library solves imaging problems such as CT with a plug-in design. The new framework lets researchers combine building blocks to form many stochastic algorithms, instead of coding each one as a black box. They show it on real X-ray CT and PET reconstruction data. The suite aims to make testing and benchmarking stochastic methods easier.
Catalog record
| CREATOR | Evangelos Papoutsellis, Margaret A. G. Duff, Jakob S. Jørgensen, Sam Porter, Claire Delplancke, Gemma Fardell, Edoardo Pasca, Kris Thielemans |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 84378 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.21230
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Modular Approach to Stochastic Optimisation for Inverse Problems Using the Core Imaging Library Evangelos Papoutsellis, Margaret A. G. Duff, Jakob S. Jørgensen, Sam Porter, Claire Delplancke, Gemma Fardell, Edoardo Pasca, Kris Thielemans Abstract . The Core Imaging Library (CIL) is an open-source versatile Python framework for solving inverse problems with special emphasis on imaging applications such as computed tomography (CT), using a plug-in architecture for data and operators, interfacing to toolboxes such as ASTRA, TIGRE and SIRF. A key component of CIL is its optimisation module enabling users to flexibly combine mathematical operators and functionals to form smooth and non-smooth optimisation problems and solve these with a range of first-order algorithms. The present work introduces an expansion of CIL with a new modular framework for stochastic optimisation, allowing researchers to easily use a variety of existing stochastic optimisation algorithms as well form new ones by combining modular building blocks. Users can flexibly configure algorithmic components, adapt to diverse problem structures, and experiment with various sampling and step size strategies. Rather than individual black-box implementations of each fixed algorithm with significant redundancies, our design is modular providing building blocks that can be flexibly combined to realise a wealth of algorithm instances. The framework is particularly well-suited for large-scale applications, where stochastic methods offer notable computational advantages over deterministic approaches. To demonstrate its versatility and practical utility, we present experiments on real-world datasets from imaging inverse problems, such as X-Ray CT and Positron Emission Tomography (PET) reconstruction. In summary, the presented software expansion aims to support the research community with a robust, extensible optimisation suite for developing, testing, and benchmarking stochastic methods for inverse problems. \SetWatermarkText 1. Introduction Imaging Inverse problems are a fundamental challenge in computational imaging, where the objective is to estimate an unknown quantity from indirect, incomplete and often noisy measurements. These arise in various fields including medical imaging, materials science, remote sensing and geoscience. For finite dimensional spaces \mathbb{X},\mathbb{Y} and a forward operator \bm{A}:\mathbb{X}\rightarrow\mathbb{Y} , many inverse problems are characterised by the following operator equation (1) \bm{A}\bm{u}\approx\bm{b}, where \bm{u}\in\mathbb{X} is the unknown quantity and \bm{b}\in\mathbb{Y} is the measurement data. Often these problems are ill-posed, i . e ., they lack existence, uniqueness or stability of a solution. To address this, a common approach is to integrate prior information of the solution in the form of regularisation in a variational (optimisation-based) method. Variational methods with regularisation aim to recover an approximated solution \bm{u}\in\mathbb{X} of ( 1 ) by solving the following optimisation problem (2) \min_{\bm{u}\in\mathbb{X}},D(\bm{A}\bm{u},\bm{b})+\alpha R(\bm{u}). Here, D:\mathbb{Y}\times\mathbb{Y}\rightarrow\mathbb{R}_{\infty}:=\mathbb{R}\cup\{\infty\} denotes the fidelity term, which measures the discrepancy between the forward model applied to the unknown quantity and the measurement data. Depending on the noise statistics of the measurement data, \bm{b} , the fidelity term often takes the form of a least squares distance, an absolute difference, or a Kullback-Leibler divergence for Gaussian, impulse ( 61 ) , and Poisson noise ( 52 ) , respectively. The regularisation term, R(\bm{u}) , promotes certain properties to the solution, \bm{u} , such as smoothness, sparsity or edge preservation and is usually weighted by a regularisation parameter, \alpha>0 , acting as a balance between the two terms in ( 2 ). Popular choices for the regularisation term are Tikhonov ( 82 ) , Total Variation (TV) ( 73 ) and the high order extension, namely the Total Generalised Variation (TGV), ( 17 ) , Total Nuclear Variation, ( 41 ) and more general tensor-based structure regularisation ( 53 ) . Numerous algorithms have been proposed in the literature ( 21 ; 12 ; 16 ; 74 ; 55 ) designed to solve ( 2 ), each tailored to specific imaging modalities, data fidelity requirements, and prior assumptions. Some of the present authors presented the Core Imaging Library (CIL) in ( 45 ) , a versatile open-source software framework designed for imaging inverse problems with an emphasis on advanced tomographic imaging tasks. Besides important utilities related to pre-processing, post-processing and visualisation of imaging data encountered in computed tomography (CT) and related modalities, a flexible optimisation framework was designed to find solutions for inverse problems. The flexible cil . optimisation module, based on three main building blocks, namely Functions , Operators , and Algorithms , allows the users to construct solutions with minimal effort for various optimisation problems tailored to specific imaging applications. Several algorithms have been implemented in CIL to solve minimisation problems in the form of ( 2 ). These are Gradient Descent (GD), Conjugate Gradient Least Squares (CGLS), as well as proximal splitting-based algorithms including Proximal Gradient Descent (PGD) (aliased to ISTA), Fast Iterative Shrinkage Thresholding Algorithm (FISTA) ( 10 ) , Primal-Dual Hybrid Gradient (PDHG) ( 20 ) and Linearised Alternate Direction Method of Multipliers (LADMM) ( 64 ) . These algorithms, implemented in CIL, have been utilised in various applications with a strong emphasis on tomographic reconstruction problems on large and challenging data sets, including low-count, non-standard geometries, limited angle, multispectral or multimodality data, including conventional X-ray CT in ( 45 ) , dynamic X-ray CT, ( 65 ) , limited angle CT, ( 46 ) , Hyperspectral and Neutron Tomography in ( 88 ; 4 ) and fusion plasma imaging, ( 78 ) . Moreover, the CIL optimisation offers a unique solution to the medical imaging community since it is fully compatible with the Synergistic Image Reconstruction Software (SIRF) ( 63 ) , open-source software for applications in medical imaging including Positron Emission Tomography (PET), Single Photon Emission Computed Tomography (SPECT) and Magnetic Resonance Imaging (MRI), ( 18 ; 67 ) . Finally, the CIL library has also been applied beyond tomography, notably in synthetic aperture radar (SAR) imaging, ( 89 ) . Despite the substantial advancements and successes of iterative regularisation algorithms for inverse problems in recent years, uptake of their use is limited because they can be slow to converge or have high memory demands. Traditional first-order solvers and their primal-dual variants are typically designed to operate in a deterministic mode, processing the entire dataset for every update, i . e ., one data pass for each iteration. This approach can be computationally intensive, especially as tomographic scanners continue to advance, producing increasingly larger and higher-resolution datasets. This challenge has motivated the creation of algorithms that provide faster convergence rates per data pass and reduce per-iteration computational costs, with many utilising stochastic optimisation techniques, see Ehrhardt et al. ( 35 ) for a comprehensive review. Unlike traditional deterministic methods that process the entire dataset in every iteration, stochastic approaches randomly select and process a subset of the acquired data, for example, a small number of tomographic projection views. Similar to Stochastic Gradient Descent (SGD) methods widely used in machine learning, this strategy has the potential to reduce per-iteration computation, while still converging to the solution. These advantages position stochastic optimisation as a compelling solution for tackling the challenges posed by high resolution, large-scale datasets in modern inverse problems. 1.1. Contribution In this paper, we extend the deterministic CIL optimisation framework by introducing a unique and user-friendly design that enables efficient implementation and testing of stochastic optimisation techniques. The rapid advancement in optimisation research results in the frequent publication of new algorithms, making it challenging to evaluate their