InsightSoftwareConsortium/ITK: ITK 6.0 Beta 2: CMake Module Targets

Open MIND · 2026-03-25

We are excited to announce the second beta release of ITK 6.0! This release introduces modern CMake module targets — a major build system modernization that makes it easier than ever to link ITK into your projects. Alongside this headline feature, Beta 2 delivers continued C++ modernization, Python enhancements, performance improvements, and expanded agentic engineering practices. As a beta release, we actively encourage the community to test and provide feedback as we approach the first Release Candidate. 🚀 ✨ 🎉 🔦 Highlights CMake Module Targets The most significant change in Beta 2 is the introduction of modern CMake interface library targets for every ITK module. Each module now exports a namespaced target — for example, ITK::ITKCommonModule, ITK::ITKImageGradientModule, or ITK::ITKThresholdingModule — enabling clean, explicit dependency management with proper transitive dependency resolution. Pre-defined factory meta-modules simplify linking for common IO use cases: ITK::ITKImageIO — all image IO modules (JPEG, PNG, NIFTI, DICOM, etc.) ITK::ITKMeshIO — all mesh IO modules ITK::ITKTransformIO — all transform IO modules Before (ITK 5): find_package(ITK REQUIRED) include(${ITK_USE_FILE}) add_executable(HelloWorld HelloWorld.cxx) target_link_libraries(HelloWorld ${ITK_LIBRARIES}) After (ITK 6): find_package(ITK REQUIRED) itk_generate_factory_registration() add_executable(HelloWorld HelloWorld.cxx) target_link_libraries(HelloWorld ITK::ITKCommonModule) For projects that read images, simply link the IO meta-module: find_package(ITK REQUIRED) itk_generate_factory_registration() add_executable(MyFilter MyFilter.cxx) target_link_libraries( MyFilter PRIVATE ITK::ITKImageGradientModule ITK::ITKImageIO ) UseITK.cmake (${ITK_USE_FILE}) is now deprecated in favor of these interface library targets. ITK can also be consumed via CMake FetchContent, illustrated in the new HelloFetchITK example. For full migration details, see the ITK 6 Migration Guide — also a useful reference for both humans and AI agents working with ITK projects. We especially welcome feedback on the CMake module targets during this beta period. Please report any issues you encounter with the new build system on GitHub. C++ Modernization VariableLengthVector gains a std::vector-like interface: front(), back(), data(), and a (length, value) constructor FixedArray gains front(), back(), and empty() member functions ImageConstIterator::ComputeIndex() added as the replacement for GetIndex() (now deprecated), with performance benefits from avoiding redundant offset table lookups New itk::MakeIndexRange utility functions for convenient index-space iteration SobelOperator extended to N-D with corrected 3D coefficients and a UseLegacyCoefficients option NumericTraits::GetLength is now constexpr Replaced vnl/vnl_sample.h with standard <random> across tests Expanded use of constexpr, [[nodiscard]], aggregate initialization, and range-based for loops Python Enhancements Python 3.10+ is now required for wrapping New ITK_PYTHON_RELEASE_GIL build option enables GIL release during ITK filter execution Updated bundled pygccxml to v3.0.2 Performance Improvements Tuned ImageRegionRange copy used in CastImageFilter IndexRange iterators made significantly faster via m_Position data member ImageRegionRange bypasses OffsetTable[0] (which is always 1) Prefer "with index" iterators when the index is used in each iteration, avoiding redundant ComputeIndex() calls Third-Party Library Updates GDCM 3.2.2, double-conversion 3.4.0, Expat, VXL, NrrdIO, CastXML v2026.01.30 Platform & CI Updates Updated CI to Windows 2022 images Added Linux ARM CI testing Migrated to macOS 15 runners (Apple Silicon) Support building with system Eigen3 >= 5 New ITK_FREEZE_REMOTE_MODULES option to prevent remote module updates during build Remote Module Updates CudaCommon v2.0.0 with __cuda_array_interface__ support for interoperability with CuPy and PyTorch RTK major updates: new Python applications, CUDA improvements, spectral CT documentation Remote modules updated for CMake modernization (itk_module_add_library) Agentic Engineering ITK continues to advance its agentic engineering practices, improving the developer experience with agent-friendly workflows, better CI, and review agents. In this release, copilot-swe-agent[bot] contributed the GIL release feature, NumericLocale GoogleTest coverage, bug fixes for imread and locale-dependent NRRD parsing, and more. The AGENTS.md guide and GitHub Copilot setup workflow continue to be refined. Now is as great a time as ever to contribute to ITK and learn from an active and friendly community of experienced medical imaging professionals. 💾 Download Python Packages Install ITK Python packages with: pip install --pre --upgrade itk Note: Use --pre flag to install beta releases. Guide and Textbook InsightSoftwareGuide-Book1-6.0b02.pdf InsightSoftwareGuide-Book2-6.0b02.pdf Library Sources InsightToolkit-6.0b02.tar.gz InsightToolkit-6.0b02.zip Testing Data Unpack optional testing data in the same directory where the Library Source is unpacked. InsightData-6.0b02.tar.gz InsightData-6.0b02.zip Checksums and Signatures InsightToolkit-6.0b02.tar.gz.asc InsightToolkit-6.0b02.zip.asc InsightData-6.0b02.tar.gz.asc InsightData-6.0b02.zip.asc MD5SUMS SHA512SUMS 📖 Migration Guide For detailed guidance on migrating existing code to ITK 6.0, please consult the ITK 6 Migration Guide. Key changes in Beta 2 include: Modern CMake interface library targets replacing UseITK.cmake Python 3.10+ requirement for wrapping ImageConstIterator::ComputeIndex() replacing GetIndex() GIL release option for Python multithreading 🙏 Acknowledgments Thank you to the 22 contributors who made ITK 6.0 Beta 2 possible! We extend a special welcome to our 5 new contributors: Cavan Riley, Yu-An Chen, Axel Garcia, Michael Cho, and nslay. Thank you for joining the ITK community! We are grateful for Axel Garcia, Ben Boeckel, Brad King, Bradley Lowekamp, Cavan Riley, Cyril Mory, Dave Chen, Davis Marc Vigneault, Dženan Zukić, Hans Johnson, Jon Haitz Legarreta Gorroño, Lucas Gandel, Matt McCormick, Michael Cho, Niels Dekker, nslay, Paul Elliott, Sean McBride, Simon Rit, Tom Birdsong, Vladimir S. FONOV, Yu-An Chen, and the entire ITK community for their code contributions, testing, documentation efforts, and invaluable feedback. 🗣️ What's Next ITK 6.0 Release Candidate 1 Following community testing and feedback on Beta 2, the next ITK 6 release will be the first Release Candidate. Please test the beta — especially the new CMake module targets — in your workflows and report any issues. ITK 5.4 Long-Term Support ITK 5.4 will continue to receive maintenance updates as a long-term support (LTS) release, providing: Critical security patches Bug fixes for production environments Full support for Intel macOS Backward compatibility for existing applications Community Feedback We actively welcome your feedback on ITK 6.0 Beta 2! Please report issues on GitHub or the ITK Discourse forum. Feedback on the CMake module targets is especially valuable as we finalize this major change for the Release Candidate. Enjoy ITK! ITK Changes Since v6.0b01 Ben Boeckel (4): Enhancements update-common.sh: synchronize with upstream (1bc98267c9) update-third-party: sync with upstream (66a45d3bfc) Style Changes git-fetch-file: track update-common.sh (99c9c92652) thirdparty: disable exact tree matching (78982a2b08) Brad King (4): Enhancements itkexpat: Update source layout to make includes ITK-specific (8ec19eb1d6) itkexpat: Update includes to reliably use ITK's expat headers (20fd88b57b) itkexpat: Add script to import from upstream (b12ac06e7e) itkexpat: Remove source tree to make room for fresh import (4957068ebe) Bradley Lowekamp (79): Enhancements Improve initialization in Path classes (9143a4d820) Use ImageRegionRange iterator in ImageAlgorithm::Copy (5b58520f19) Use GTest::gtest and GTest::gtest_main targets (51c24a5a0f) Check IPO/LTO support once globally for all wrapper modules (48b834b5b1) Remove legacy and unused Wrapping EXECUTE_OUTPUT_PATH (ffaa0c946e) Use imported OpenCL targets from CMake (bf1566b687) Use OpenCL imported target in GPUCommon (e3b18e7142) Use target-based includes for ITK modules (87cc629cd4) Create CMake interface libraries for each ITK module (d159777773) Add modern CMake linking with ITK interface modules to examples (602a5e03c6) Configure factory meta-modules with registration properties (056d0a60ab) Link module libraries to interface modules libraries (01971b74c8) Add namespace support for all ITK targets (6dd596e53f) Update Installation Example to use ITK Interface libraries (377423ff66) enable multiple call to itk_generate_factory_registration (bf531980db) Add deprecation warning to UseITK usage. (14665e644c) Allow modules to override GENEX INCLUDE variables (f9638701f6) Remove CMake code block for unsupported versions (22d4ff4f32) Update to use PRIVATE and PUBLIC linking modes (8a4dd39dd6) MINC update script properly ignores trailing whitespace (0bad5e05c0) Enable gersemi to run on ITK CMake files in ThirdParty directory (5a8fb4c71b) teach itk_module_macro to support NAMESPACE argument (d8f37be021) Use ITK CMake interface targets for all example linking (3048ad4230) Use modern CMake interface libraries for ZLIB (8af1b5c5f6) Support installation of FILE_SET HEADER for module targets (b4399480cd) Update double-conversion to 3.4.0 (1aaaf1bef9) Update double-conversion CMake to target_includes (1fec0a4dce) Add example project which used ITK Modules and FetchContent (f8c35f0fb2) Conform GDCM to use CMAKE_*OUTPUT_DIRECTORY (d5cf8de3ed) Performance Improvements Use tuned ImageRangeRegion copy in CastImageFilter (2e8c70a449) Documentation Updates Add migration guide for new CMake Interfa

Deep Computational Anatomy via Latent-Aligned Multiview Normalizing Flows

bioRxiv (Cold Spring Harbor Laboratory) · 2026-05-10

In modeling complex probability distributions, normalizing flows provide exact-likelihood, bijective mappings between empirical data and tractable latent spaces. Building on this foundation, latent-aligned multiview normalizing (LAMNr) flows leverage these salient properties to learn shared latent subspaces across heterogeneous, multimodal datasets while simultaneously topologically unfolding the sampled data manifold into a continuous vector space. Formal latent-alignment constraints are used to model shared structural features separate from view-specific variations, coordinating latent projections into a shared geometric subspace. By applying this transformation in the context of biological imaging, the framework establishes a potential basis for a deep learning interpretation of foundational computational anatomy concepts, such as the population template, latent distances, and geodesic pairwise image interpolation. Additionally, the proposed framework enables closed-form conditional modeling for exact cross-view imputation and other latent space manipulations. Evaluations and illustrations on both imaging-derived phenotypes (IDPs) and multimodal MRI demonstrate the proposed framework and potential applications. To further motivate our work, we provide a robust and comprehensive, 2D and 3D open-source implementation in PyTorch, natively integrated with the ANTsX ecosystem (i.e., ANTsTorch) for efficient training and subsequent data transformation, manipulation, and analysis.

MRI2PET: Realistic PET Image Synthesis from MRI for Automated Inference of Brain Atrophy and Alzheimer's

SSRN Electronic Journal · 2025-01-01

Joint representations from multi-view MRI-based learning support cognitive and functional performance domains

medRxiv · 2025-10-05 · 1 citations

Abstract Background Multimodal MRI (sMRI, dMRI, rsfMRI) encodes complementary aspects of brain structure and function; principled joint representations promise more sensitive and interpretable markers of brain health than single-modality features. Methods We evaluate Normative Neurological Health Embedding (NNHEmbed), a flexible multi-view framework that uses constrained cross-modal similarity objectives to learn low-dimensional embeddings. Models were trained and tested on the UK Biobank (n = 21,300) and evaluated for transfer and longitudinal sensitivity in independent cohorts (Normative Neuroimaging Library, Alzheimer’s Disease Neuroimaging Initiative, Parkinson’s Progression Markers Initiative). Results NNHEmbed produced compact, biologically interpretable components that (a) maps to established neurocognitive systems (e.g., episodic memory, processing-speed, sensorimotor/basal-ganglia circuits), (b) generalizes across cohorts, and (c) captures within-subject change over time. Best configurations balance reconstruction fidelity and shared covariance, improving interpretability while preserving predictive utility. Case demonstrations illustrate individualized normative profiling across multiple visits. Conclusions NNHEmbed yields stable, transferable multimodal embeddings suitable for normative mapping and longitudinal monitoring. Software, NNHEmbed configurations and derived bases are available for reproduction and reuse.

antspymm helper data

Figshare · 2025-01-01

multi-channel and time series example data for antsx processing ... with priors

MRI2PET: Realistic PET Image Synthesis from MRI for Automated Inference of Brain Atrophy and Alzheimer’s

medRxiv · 2025-04-25 · 3 citations

Background: Positron Emission Tomography (PET) scans are a crucial tool in the diagnosing and monitoring of a number of complex conditions, including cancer, heart health, and especially cognitive brain function. However, they are also often much more expensive than comparable imaging modalities such as X-Ray and magnetic resonance imaging (MRI), which can limit their availability and the impact of their use in both medical and machine learning settings. We propose to address this problem by using generative models to simulate the PET scan results based on prior MRI. Methods: While recent work has yielded impressive realism in image generation, this PET synthesis task presents a series of technical challenges based on the scarcity of paired data as well as the complexity and nuance of the 3D images. So, we propose MRI2PET to generate AV45-PET scans from T1-weighted MRI images. MRI2PET is a 3D diffusion-based method which makes use of style transferred pre-training and a Laplacian pyramid loss to address these challenges by utilizing larger available unpaired MRI datasets and structural similarities between the MRI and PET images while simultaneously emphasizing the crucial details. Findings: We evaluate MRI2PET through a series of studies on the ADNI dataset where we show that it both generates realistic images and improves clinically-based disease classification. When compared to training on only the original AV45-PET data, MRI2PET augmentation increases AUROC of brain scan classification to 0.780 ± 0.005 from 0.688 ± 0.014 when classifying brain scans into one of three clinically defined groups: cognitively normal, mild cognitive impairment, and Alzheimer's Disease. Interpretation: The capability to generate high quality, clinically relevant PET scans from MRI has the potential to expand the utility of cost-effective and accessible imaging workflows and improve both image-based machine learning capabilities and patient care. Funding: US National Institute on Aging, US National Institutes of Health, US National Science Foundation.

<i>LRRK2</i> -associated parkinsonism with and without <i>in vivo</i> evidence of alpha-synuclein aggregates: longitudinal clinical and biomarker characterization

Brain Communications · 2025-01-01 · 15 citations

Abstract Among LRRK2-associated parkinsonism cases with nigral degeneration, over two-thirds demonstrate evidence of pathologic alpha-synuclein, but many do not. Understanding the clinical phenotype and underlying biology in such individuals is critical for therapeutic development. Our objective was to compare clinical and biomarker features, and rate of progression over 4 years of follow-up, among LRRK2-associated parkinsonism cases with and without in vivo evidence of alpha-synuclein aggregates. Data were from the Parkinson’s Progression Markers Initiative, a multicentre prospective cohort study. The sample included individuals diagnosed with Parkinson disease with pathogenic variants in LRRK2. Presence of CSF alpha-synuclein aggregation was assessed with seed amplification assay. A range of clinician- and patient-reported outcome assessments were administered. Biomarkers included dopamine transporter scan, CSF amyloid-beta1-42, total tau, phospho-tau181, urine bis(monoacylglycerol)phosphate levels and serum neurofilament light chain. Linear mixed-effects (LMMs) models examined differences in trajectory in CSF-negative and CSF-positive groups. A total of 148 LRRK2 parkinsonism cases (86% with G2019S variant), 46 negative and 102 positive for CSF alpha-synuclein seed amplification assay, were included. At baseline, the negative group was older than the positive group [median (inter-quartile range) 69.1 (65.2–72.3) versus 61.5 (55.6–66.9) years, P &amp;lt; 0.001] and a greater proportion were female [28 (61%) versus 43 (42%), P = 0.035]. Despite being older, the negative group had similar duration since diagnosis and similar motor rating scale [16 (11–23) versus 16 (10–22), P = 0.480] though lower levodopa equivalents. Only 13 (29%) of the negative group were hyposmic, compared with 75 (77%) of the positive group. The negative group, compared with the positive group, had higher per cent-expected putamenal dopamine transporter binding for their age and sex [0.36 (0.29–0.45) versus 0.26 (0.22–0.37), P &amp;lt; 0.001]. Serum neurofilament light chain was higher in the negative group compared with the positive group [17.10 (13.60–22.10) versus 10.50 (8.43–14.70) pg/mL; age-adjusted P-value = 0.013]. In terms of longitudinal change, the negative group remained stable in functional rating scale score in contrast to the positive group who had a significant increase (worsening) of 0.729 per year (P = 0.037), but no other differences in trajectory were found. Among individuals diagnosed with Parkinson disease with pathogenic variants in the LRRK2 gene, we found clinical and biomarker differences in cases without versus with in vivo evidence of CSF alpha-synuclein aggregates. LRRK2 parkinsonism cases without evidence of alpha-synuclein aggregates as a group exhibit less severe motor manifestations and decline. The underlying biology in LRRK2 parkinsonism cases without evidence of alpha-synuclein aggregates requires further investigation.

Prediction, Prognosis and Monitoring of Neurodegeneration at Biobank-Scale via Machine Learning and Imaging

SSRN Electronic Journal · 2024-01-01 · 1 citations

Normative Neuroimaging Library: Designing a Comprehensive and Demographically Diverse Dataset of Healthy Controls to Support Traumatic Brain Injury Diagnostic and Therapeutic Development

Journal of Neurotrauma · 2024-09-05 · 9 citations

The past decade has seen impressive advances in neuroimaging, moving from qualitative to quantitative outputs. Available techniques now allow for the inference of microscopic changes occurring in white and gray matter, along with alterations in physiology and function. These existing and emerging techniques hold the potential of providing unprecedented capabilities in achieving a diagnosis and predicting outcomes for traumatic brain injury (TBI) and a variety of other neurological diseases. To see this promise move from the research lab into clinical care, an understanding is needed of what normal data look like for all age ranges, sex, and other demographic and socioeconomic categories. Clinicians can only use the results of imaging scans to support their decision-making if they know how the results for their patient compare with a normative standard. This potential for utilizing magnetic resonance imaging (MRI) in TBI diagnosis motivated the American College of Radiology and Cohen Veterans Bioscience to create a reference database of healthy individuals with neuroimaging, demographic data, and characterization of psychological functioning and neurocognitive data that will serve as a normative resource for clinicians and researchers for development of diagnostics and therapeutics for TBI and other brain disorders. The goal of this article is to introduce the large, well-curated Normative Neuroimaging Library (NNL) to the research community. NNL consists of data collected from ∼1900 healthy participants. The highlights of NNL are (1) data are collected across a diverse population, including civilians, veterans, and active-duty service members with an age range (18-64 years) not well represented in existing datasets; (2) comprehensive structural and functional neuroimaging acquisition with state-of-the-art sequences (including structural, diffusion, and functional MRI; raw scanner data are preserved, allowing higher quality data to be derived in the future; standardized imaging acquisition protocols across sites reflect sequences and parameters often recommended for use with various neurological and psychiatric conditions, including TBI, post-traumatic stress disorder, stroke, neurodegenerative disorders, and neoplastic disease); and (3) the collection of comprehensive demographic details, medical history, and a broad structured clinical assessment, including cognition and psychological scales, relevant to multiple neurological conditions with functional sequelae. Thus, NNL provides a demographically diverse population of healthy individuals who can serve as a comparison group for brain injury study and clinical samples, providing a strong foundation for precision medicine. Use cases include the creation of imaging-derived phenotypes (IDPs), derivation of reference ranges of imaging measures, and use of IDPs as training samples for artificial intelligence-based biomarker development and for normative modeling to help identify injury-induced changes as outliers for precision diagnosis and targeted therapeutic development. On its release, NNL is poised to support the use of advanced imaging in clinician decision support tools, the validation of imaging biomarkers, and the investigation of brain-behavior anomalies, moving the field toward precision medicine.

Magnetic Resonance Imaging Data Phenotypes for the Parkinson’s Progression Markers Initiative

medRxiv · 2024-09-24 · 3 citations

Abstract The Parkinson’s Progression Markers Initiative (PPMI) delivers multiple modality MRI (M3RI) and biomarker data for a comprehensive longitudinal study of Parkinson’s Disease (PD). These provide quantitative indices of deep brain and cortical structure (T1-weighted MRI), microstructural integrity of brain tissue (diffusion-weighted imaging) and resting brain function (resting state functional MRI). Integrating and uniformly analyzing M3RI alongside non-imaging biological and clinical data is challenging due to the distinct nature of each modality. This study systematically organizes these complex data into a structured format, provides a PD-focused evaluation of the methodologies and evidence for technical robustness of the approach. The cohort encompasses 841 idiopathic PD, 309 genetic PD, 1364 presymptomatic PD and 240 control subjects at baseline with followup at a mean of 1.83 years.