Using Space-Filling Curves and Fractals to Reveal Spatial and Temporal Patterns in Neuroimaging Data

Jacek Grela, Zbigniew Drogosz, Jakub Janarek, Jeremi K. Ochab, Ignacio Cifre, Ewa Gudowska-Nowak, Maciej A. Nowak, Paweł Oświęcimka, Dante R. Chialvo, for the Alzheimer’s Disease Neuroimaging Initiative · 2025

Read Using Space-Filling Curves and Fractals to Reveal Spatial and Temporal Patterns in Neuroimaging Data by Jacek Grela, Zbigniew Drogosz, Jakub Janarek, Jeremi K. Ochab, Ignacio Cifre, Ewa Gudowska-Nowak, Maciej A. Nowak, Paweł Oświęcimka, Dante R. Chialvo, for the Alzheimer’s Disease Neuroimaging Initiative (2025) 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-2501-12111.txt.

Introduction

Brain scans are pictures with patterns at every scale, and fractal tools can measure how rough or correlated those patterns are. Jacek Grela and co-authors present in 2025 a method with two steps. First, a space-filling curve flattens a multidimensional scan into a one-dimensional thread, keeping neighbors in order. Then fractal fluctuation analysis measures the thread.

They validated the method on synthetic data with known fractal character, tuned the curve choice for speed and robustness, and turned it loose on real scans. In Alzheimer's MRI data, the fractal exponent fell steadily as the disease progressed. In functional scans during a breath-holding task, the exponent's changes separated the task's phases. The authors offer the tool as a general way to turn spatial and temporal correlation into markers of brain state and disease.

Catalog record

CREATORJacek Grela, Zbigniew Drogosz, Jakub Janarek, Jeremi K. Ochab, Ignacio Cifre, Ewa Gudowska-Nowak, Maciej A. Nowak, Paweł Oświęcimka, Dante R. Chialvo, for the Alzheimer’s Disease Neuroimaging Initiative
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE68 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-12111

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Using Space-Filling Curves and Fractals to Reveal Spatial and Temporal Patterns in Neuroimaging Data Jacek Grela, Zbigniew Drogosz, Jakub Janarek, Jeremi K. Ochab, Ignacio Cifre, Ewa Gudowska-Nowak, Maciej A. Nowak, Paweł Oświęcimka, Dante R. Chialvo, for the Alzheimer’s Disease Neuroimaging Initiative Abstract Objective Magnetic resonance imaging (MRI), functional MRI (fMRI) and other neuroimaging techniques are routinely used in medical diagnosis, cognitive neuroscience or recently in brain decoding. They produce three- or four-dimensional scans reflecting the geometry of brain tissue or activity, which is highly correlated temporally and spatially. While there exist numerous theoretically guided methods for analyzing correlations in one-dimensional data, they often cannot be readily generalized to the multidimensional geometrically embedded setting. Approach We present a novel method, Fractal Space-Curve Analysis (FSCA), which combines Space-Filling Curve (SFC) mapping for dimensionality reduction with fractal Detrended Fluctuation Analysis (DFA). We conduct extensive feasibility studies on diverse, artificially generated data with known fractal characteristics: the fractional Brownian motion, Cantor sets, and Gaussian processes. We compare the suitability of dimensionality reduction via Hilbert SFC and a data-driven alternative. FSCA is then successfully applied to real-world MRI and fMRI scans. Main results The method utilizing Hilbert curves is optimized for computational efficiency, proven robust against boundary effects typical in experimental data analysis, and resistant to data sub-sampling. It is able to correctly quantify and discern correlations in both stationary and dynamic two-dimensional images. In MRI Alzheimer’s dataset, patients reveal a progression of the disease associated with a systematic decrease of the Hurst exponent. In fMRI recording of breath-holding task, the change in the exponent allows distinguishing different experimental phases. Significance This study introduces a robust method for fractal characterization of spatial and temporal correlations in many types of multidimensional neuroimaging data. Very few assumptions allow it to be generalized to more dimensions than typical for neuroimaging and utilized in other scientific fields. The method can be particularly useful in analyzing fMRI experiments to compute markers of pathological conditions resulting from neurodegeneration. We also showcase its potential for providing insights into brain dynamics in task-related experiments. 1 Introduction Technological advancements in experimental science have resulted in an abundance of multidimensional data. This trend is particularly observed in neuroscience, where brain imaging techniques such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) produce four-dimensional scans (comprising three spatial dimensions and one temporal dimension). Such developments require novel, dedicated methods of analysis combining the diversity of input data with interpretable outcomes. In brain imaging, most established data analysis methods are tailored to specific imaging techniques—for example, ICA is mainly applied to functional MRI. In contrast, our method offers a nonparametric, general approach for quantifying the fractal properties of scans. According to the structural MRI method classification discussed in [ 1 ] , our method generates a texture-based, statistical, second-order biomarker. This data-agnostic method aims to provide features similar to ROI-based biomarkers, such as those based on gray-matter volume, cortical thickness, or Voxel-Based Morphometry. One compelling strategy of multidimensional data analysis involves transforming the data into more typical forms, such as a one-dimensional array, to enable standard time series methods. A fitting family of transformations is the Space-Filling Curves (SFCs), an indispensable tool of contemporary scientific computing [ 2 ] . SFCs are continuous functions that can convert multidimensional geometrically embedded data grids into one-dimensional sequences while preserving the local properties of the multidimensional data. This locality, or clustering, property of SFCs is highly desirable, ensuring that nearby points in a multidimensional space correspond to nearby points in a one-dimensional domain. This property renders SFCs prevalent instruments for dimensionality reduction and image compression [ 3 ] , efficient storage and retrieval of data from geographical information systems, cryptography, and data visualization [ 4 , 5 , 6 ] . In neuroscience, SFC-based techniques also find extensive use in the processing of magnetic resonance imaging (MRI) data [ 7 , 8 , 9 , 10 , 11 ] , e.g., selecting brain’s activation maps from schizophrenia or Alzheimer’s study. Mapping geometrically embedded multidimensional data onto a one-dimensional sequence presents an opportunity for utilizing fractal analysis of time series. One aspect of the theory of fractals explores the self-similar properties of data quantified by the fractal dimension exponent. In recent years, fractal characteristics have emerged across various scientific and artistic domains, spanning physics [ 12 , 13 ] , biology [ 14 ] , neuroscience [ 15 , 16 , 17 , 18 ] , chemistry [ 19 ] , economics [ 20 , 21 , 22 ] , quantitative linguistics [ 23 , 24 ] , and music [ 25 ] . The ubiquity of fractal structures owes much to the advancement of robust algorithms, among which detrended fluctuation analysis (DFA) is especially prominent [ 26 ] . DFA quantifies the temporal organization of data through the Hurst exponent, a scaling exponent closely related to the fractal dimension. Thus, as scaling exponents reflect the correlation organization of data, DFA output can be interpreted through an interdisciplinary lens of fractality, thereby broadening the scope for understanding findings within the context of scale-free dynamics. In this contribution, we introduce the Fractal Space-Curve Analysis (FSCA) shown in Fig. 1 , a novel approach to multidimensional data analysis integrating space-filling curves with detrended fluctuation analysis. FSCA utilizes SFCs as a method for locality-preserving dimensionality reduction, while DFA is employed to extract interpretable fractal properties from the data. In this work, we provide an extensive feasibility study of the method. We find FSCA robust to the boundary effects, behaving consistently for different data resolutions and recreating correct fractal properties of synthetic data. We show its versatility in application to both the synthetic data and real-world data of diverse shapes spanning two-dimensional images, three-dimensional scans, and four-dimensional scans changing in time. The paper is structured as follows: Section 2 introduces the concept of the proposed Fractal Space-Curve Analysis (FSCA) methodology. We begin by discussing the basic concepts of space-filling curves and fractal analysis using detrended fluctuation analysis. Section 3 presents an FSCA analysis of synthetic data that mimic the structure of neuroscientific datasets. Within this framework, fractal signals are considered to be two-dimensional fractional Brownian motion images, mirroring the analysis of MRI scans, binary Cantor sets, and two-plus-one-dimensional Gaussian processes similar to fMRI data analysis. We scrutinize FSCA for the optimal selection of the space-filling curve, potential boundary effects in experimental data, and data sampling density. In Section 4 , we apply FSCA to analyze neuroimaging data. We examine MRI datasets obtained from patients diagnosed with Alzheimer’s disease, as well as fMRI signals from a single healthy individual undergoing various tests, referred to as the Poldrack dataset. Section 5 presents summary and conclusions. Figure 1 : The flowchart of the proposed Fractal Space-Curve Analysis comprised of d) a slicing preprocessing step, e) space-filling-curve linearization mapping to reduce problem dimensionality and f) the detrended fluctuation analysis to extract Hurst exponents. The method is applicable to a) two-dimensional, b) three-dimensional, or c) three-plus-one-dimensional data. Method’s output describes the fractality of the data in the form of g) single Hurst exponents, h) Hurst exponent profile resolved along a spatial axis or i) Hurst profile resolved in time. 2 Method: Fractal Space-Curve Analysis (FSCA) This chapter

Read the complete text