v1.0.1

v1.0.1 — Reproducibility, CI, and API curation

A maintenance and infrastructure release: gated continuous-integration test
lanes, deterministic identity/fingerprinting primitives, MCP integration docs,
a curated public API surface, and targeted fixes. It builds on the v1.0.0
two-layer toolbox without changing its workflows.

Testing & continuous integration

• Authoritative test-suite builder (buildSuite) and a runCI gate that runs
  the headless, UI, clean-process, and quick lanes and fails on Failed or
  Incomplete tests.
• runQuickTests now returns a pass flag and errors on failure when no output
  is captured, so it can act as a gate.
• Manual GitHub Actions workflow (workflow_dispatch) running the quick and
  identity lanes, including cross-platform matrix jobs.
• Golden-file regression fixtures and notes under tests/golden/.

Identity & reproducibility primitives (+pf2_base/+identity/)

• Deterministic canonical byte encoding, domain-separated (named) hashing, and
  SHA-256 digests.
• A tagged JSON/MAT transport for round-tripping values canonically.
• Source and imported-data fingerprinting, plus a fail-closed expected-digest
  preflight that gates importers.

Processing methods

• Shipped Takizawa oxy seed methods (takizawa_easy / takizawa_hard, lenient
  and strict automatic channel rejection — Takizawa et al. 2008) added to the
  default library.
• Method reference tables and examples updated to the live registry and to the
  OD_TDDR raw method.

Public API & namespace curation

• The Context processing option now uses the public pf2.ProcessingContext;
  the public fromRecipe helper was removed.
• Method-authoring internals moved out of the primary pf2.* surface into
  pf2_base: seed factories and discovery to pf2_base.methods.seeds,
  function-metadata validation to pf2_base.methods.validateFunction, the
  add/edit-function tools to pf2_base.methods.functions, and the tutorial
  demo-pipeline setup to pf2_base.examples.addDemoPipelines.
• Retired the pf2.GUI.configureRawMethods / configureOxyMethods /
  functions wrappers; the supported method editors are
  pf2.methods.raw.configureMethods and pf2.methods.oxy.configureMethods.

MCP integration

• mcp/README.md and mcp/TOOL_CARD.md: how to drive processFNIRS2 from an MCP
  client through the MATLAB MCP Server.

Documentation

• pf2_sSMART header references added (Curtin & Ayaz 2019; Ayaz et al. 2010 for
  the SMAR detection stage) and an expanded Notes section on the DC-offset
  (ShiftCorrect) correction and its caveats; pf2_functions_default.cfg
  argument list, role, defaults, and parameter metadata corrected to match the
  function.
• Docs clarify pf2_base as an advanced-and-internal package.

Fixes

• getTakizawaRejection now uses the first FFT bin above 1 Hz as a scalar index
  rather than the full index vector.
• Signed-zero handling: negative zero round-trips canonically as +0 through
  the JSON transport, with explicit test coverage.

Full Changelog: v1.0.0...v1.0.1

v1.0.0: Two-Layer fNIRS Toolbox

v1.0.0 — Two-Layer fNIRS Toolbox

This release brings processFNIRS2 to a complete two-layer toolbox: single-subject
processing (pf2) and group analysis / statistics (exploreFNIRS), spanning the
full workflow from raw device import through preprocessing, hemoglobin conversion,
block/GLM modeling, group-level statistics, and publication-grade visualization.

Single-subject processing (pf2)

• Programmatic Pipeline API (RawPipeline/OxyPipeline, PipelineFunction)
  for building, editing, saving, and reloading named processing methods.
• ProcessingContext for global-state isolation, enabling reproducible and
  parallel (parfor) processing via copy(), recipes, and per-worker settings.
• Modular raw→OD→Hb staged pipeline with DPF modes (None/Fixed/Calc), improved
  error handling, and reduced duplication.
• Block averaging (defineBlocks → extractBlocks → blockAverage) and
  marker-free sliding-window epoching for dynamic FC / model input.

Group analysis & statistics (exploreFNIRS)

• Experiment class: grouping, aggregation, averaging modes, behavioral
  variables, ROI, and temporal/bar plotting.
• LME engine with contrasts, FDR correction, growth-curve (TimeModel)
  analysis, and stats formatting.
• Connectivity and coupling (Pearson, partial correlation, mutual
  information, wavelet coherence), graph metrics, and HB-ICA /
  hyperscanning inter-brain synchrony (including PPI).
• GLM / AR-IRLS first-level modeling with design matrices, contrasts,
  diagnostics, and optional GPU support.

Signal processing

• First-party DSP: butter, fir1, sgolay, window functions, and a
  first-party wavelet package (FFT-based CWT, translation-invariant transforms).
  The bundled WaveLab850 dependency has been removed.
• Motion correction (TDDR, SMAR), filtering, and global/systemic interference
  removal (CAR, PCA-GSR, short-channel regression).

Auxiliary signals

• Typed, first-class Aux handling (HR, EKG, PPG, ACCEL, GSR/EDA, RESP, TEMP,
  EEG): canonical form, type inference, grid alignment, feature extractors, and
  use as GLM nuisance regressors, LME covariates, and motion gates.

Spatial modeling & visualization

• Diffuse Optical Tomography: PMDF forward sensitivity model, coverage,
  honest banana projection, and depth-weighted Tikhonov image reconstruction with
  cortical render (layered head model, resolution diagnostics).
• High-quality 3D rendering with style presets (showcase/publication),
  matcap materials, ambient occlusion, brain colormaps, and an interactive
  Explore3D explorer.
• Spatial visualizations: topographic maps, activation movies, Brodmann
  parcel projection, optode parcel maps, brain-anchored connectome, circular
  connectogram, dual-brain synchrony, and statistical projections (p-values,
  F-stats, correlations).
• Consistent dark-mode theming with white-background export.

Import / export

• Import: fNIR Devices/Biopac, SNIRF (auto BIDS events.tsv), Hitachi, Artinis
  Oxy3, directory-tree batch import, tidy-table import (fromTable), and
  foundation-model embedding re-import.
• Export: NIR, SNIRF, BIDS-NIRS dataset export, and a self-describing HDF5
  tensor export contract for foundation-model consumers.
• Marker dictionary subsystem unifying COBI/BIDS event labels; markers are
  now a canonical table.
• Montage descriptor, cross-device anatomical canonicalization, and
  device .cfg save.

Devices, QC & probe anatomy

• Added/expanded device support (fNIR3000, Artinis, LD) with robust probe
  detection and MNI/2D layout fallback.
• Headless QC pipeline (assess/apply/report/snapshot), refactored
  Takizawa rejection, and an enhanced multi-file ChannelCheck GUI with headless
  gating.
• Probe anatomy: nearest-Brodmann mapping and forward/sensitivity models.

Infrastructure

• Expanded unit/integration, golden, and benchmark test suites; Octave and
  Statistics Toolbox compatibility shims; standardized error identifiers and a
  warning-suppression API; tutorial scripts and documentation.

Full Changelog: v0.9a...v1.0.0

v0.8.1: Documentation & SNIRF Improvements

What's New

• Comprehensive function documentation: Added detailed headers across import, export, data manipulation, plotting, and methods configuration modules. Headers include input/output specifications, examples, and algorithm references where applicable.

• New concatenateHorizontal function: Added pf2.data.concatenateHorizontal for joining fNIRS segments side-by-side (channel dimension) with probe geometry merging.

• exploreFNIRS version info: Added exploreFNIRS.versInfo for version tracking in the analysis module.

Bug Fixes

• Fixed pf2.data.resample to skip missing fields gracefully instead of erroring
• Fixed SNIRF import to use stable sort for consistent channel ordering
• Fixed SNIRF export to handle NaN wavelengths correctly
• Improved SNIRF import/export support for fNIRS data
• Improved probe layout ordering to prefer source/detector index
• Fixed channel filter to determine channels in advance during processing
• Fixed probe check GUI
• Fixed help file to reflect moved export functions
• Fixed "strip extra raw" functionality

Spelling Corrections

• Renamed Concatonate to Concatenate and ConcatonateHorizontal to ConcatenateHorizontal

Full Changelog: v0.8a...v8.1

v0.9a: API Standardization & Testing Infrastructure

v0.9a — API Standardization & Testing Infrastructure

This release standardizes the public API naming and lays down a comprehensive
testing foundation ahead of the v1.0 feature work.

Breaking Changes

• camelCase API migration. All public pf2.* namespaces now use camelCase:

  • pf2.Import.* → pf2.import.*
  • pf2.Export.* → pf2.export.*
  • pf2.Data.* → pf2.data.*
  • pf2.Process.* → pf2.process.*
  • pf2.Methods.* → pf2.methods.*
  • pf2.Settings.* → pf2.settings.*
  • pf2.Probe.* → pf2.probe.*

  Function names follow suit (e.g. pf2.Import.ImportNIR() → pf2.import.importNIR(),
  pf2.Methods.Raw.SetMethod() → pf2.methods.raw.setMethod()). Update call
  sites accordingly.

New Features

• Unit and integration testing infrastructure built on MATLAB's unittest
  framework, with runners pf2_base.tests.runAllTests() and
  pf2_base.tests.runQuickTests().
• Coverage across import, data-structure invariants, signal processing,
  Beer-Lambert conversion, TDDR motion correction, ROI definition/building,
  NIR/SNIRF export, FDR correction, method configuration, and auxiliary data;
  end-to-end pipeline and SNIRF round-trip integration tests.
• Synthetic fNIRS data generators with configurable HRF, cardiac/respiratory,
  noise, and motion-artifact components for reproducible testing.
• Quick validation scripts for rapid sanity checks.

Documentation

• Added CHANGELOG.md and a testing strategy document (testing_plan.md).
• Added golden-file infrastructure documentation.

Full Changelog: v0.8.1...v0.9a

v0.8a: SNIRF Support & Visualization Enhancements

What's New

• SNIRF format support: Added import and export for the standardized SNIRF file format, including embedded probe information and short separation channel handling

• NIR export: Added pf2.export.asNIR to export fNIRS structs back to .nir file format

• Violin plots: Added violin plot support in exploreFNIRS bar charts with kernel density estimation

• IQR error bars: Added interquartile range (IQR) and IQR-without-outliers options for error bar display

• Non-ordinal optode support: Channels no longer require sequential optode numbering

• Datetime support in SetT0: pf2.data.setT0 now accepts datetime parameters

• Auxiliary data splitting: pf2.data.split now properly splits and flattens auxiliary data channels

• Short separation filtering: 3D interpolation plots filter short separation channels by default

Bug Fixes

• Fixed scatter plots displaying flipped orientation
• Fixed barchart plotting with model indices
• Fixed temporal ROI plotting
• Fixed export for long and wide table formats
• Fixed channel ordering after Beer-Lambert conversion
• Fixed crash when markers are missing
• Fixed crash with mismatched channel numbers
• Fixed t0 calculation from datetime/unix timestamps
• Fixed marker sorting on SNIRF import
• Fixed probe check GUI in auto mode
• Fixed errorbar means calculation
• Fixed autocontrast indexing

Full Changelog: v0.7a...v0.8a

v0.7a: 3D Visualization & Multi-Probe Support

What's New

• MNI coordinate system: Migrated 3D visualization to MNI coordinates for standardized brain mapping

• Brodmann area visualization: Added Brodmann area plotting, labeling, and legend support on 3D brain models

• Multi-probe support: Added ability to plot and concatenate data from multiple probes

• EEG 10-20 integration: Added 10-20 EEG probe plotting and ability to overlay EEG/fNIRS data

• Voxel brain display: Added voxel-based brain visualization mode

• External plot functions: Moved temporal, scatter, and barchart plotting to standalone functions in exploreFNIRS.plot

• Auxiliary data plotting: Added Aux channel support in temporal, scatter, and barchart plots

• Categorical variable support: exploreFNIRS now handles categorical values in grouping and analysis

• Improved marker matching: Enhanced regex pattern matching for marker extraction

• Hitachi import improvements: Faster import and datetime extraction

• Resample enhancements: Support for datetime/duration fields and recursive struct/table resampling

Bug Fixes

• Fixed autocontrast sorting and indexing
• Fixed LME model failures with categorical types
• Fixed vline support for datetime/duration axes
• Fixed interpolate values chart orientation
• Fixed grandavg for auxiliary data with time integration
• Fixed suptitle spacing issues
• Fixed ICA clean and wavelet clean functionality

Full Changelog: v0.5a...v0.7a

v0.5a: Initial Package Release

What's New

• Package architecture: Migrated codebase to MATLAB package format (+pf2, +pf2_base)

• ROI support: Added region of interest functionality with automatic channel grouping

• Automatic channel rejection: Added threshold-based channel quality assessment

• Plotting functions: Added pf2.data.plot.Oxy, pf2.data.plot.Raw, arranged values, and interpolate values visualization

• Method management: Added pf2.methods.raw.setMethod, pf2.methods.oxy.setMethod, and method listing/description functions

• Device configuration: Devices now use source/detector geometry information when available

• fNIR Devices support: Updated import for newer COBI format files

• Automatic demographic loading: ImportNIR now extracts demographic data from file headers

• Channel rejection loading: Auto-loads channel rejection data when present in files

• Help documentation: Added built-in help document and version information

• exploreFNIRS: Initial group analysis GUI with LME modeling, FDR correction, and temporal/bar/scatter plots

Full Changelog: https://github.com/AdrianCurtin/processFNIRS2/commits/v0.5a