Skip to content

Releases

NVAP v0.1.9 - Soma controls and contact-based segmentation

Choose a tag to compare

@gero-studios gero-studios released this 25 Oct 15:21

NVAP 0.1.9 adds configurable soma requirements and preserves contact-connected microglia structures on the native voxel grid.

  • Soma controls: choose whether a detected soma is required, set the minimum soma radius and supported core volume, and save those settings with the dataset. Viewer colors, cell counts, analysis and exports use the same settings.
  • Contact-only segmentation: separate pieces keep separate IDs. Proximity, faint subthreshold paths and corner-only contact cannot transfer cell ownership. Multiple supported somas can split touching clusters.
  • Gray branches: show or hide unassigned fragments without changing segmentation or cell counts. With the soma requirement off, substantial branching objects that avoid both retained Z boundary planes can qualify as separately labeled Z-interior inferred candidates. Small soma-free twigs stay gray; inferred candidates do not receive fabricated soma measurements.
  • Processing: faster loading and preprocessing, native-grid connectivity, bounded parallel cell analysis, and visibility updates that avoid rebuilding geometry.

The soma and branching thresholds are configurable application heuristics, not a validated biological classifier. The substantial-arbor exception remains available by default; enable Require detected soma for strict soma-only inclusion. Use Apply changes when Auto-apply is off.

Validation: 368 unit/integration tests passed in each of the CPU and DirectML environments, with 3 skipped in each. Both Windows executables are verified against the release source and smoke-tested before publication.

Download NVAP-0.1.9-windows-cpu.exe for standard Windows use, or NVAP-0.1.9-windows-directml.exe for DirectML GPU support. SHA256SUMS.txt contains checksums for both executables.

NVAP v0.1.8 — Single-soma cells and cleaner segmentation

Choose a tag to compare

@gero-studios gero-studios released this 25 Oct 01:04

The colored microglia view now shows only components eligible for cell analysis. Both paths use the same single-soma quality check, so unsupported fragments and ambiguous components with multiple supported soma cores are excluded from the display, cell counts, and per-cell exports. The current green threshold and slice trim apply to eligibility; each accepted cell keeps its full visible arbor.

Large microscopy stacks now retain soma-thickness validation on a bounded grid instead of bypassing it. Faint branch connections are resolved before adding missed-cell markers, reducing unnecessary color boundaries. Background padding also prevents edge branches from receiving artificial soma-sized radii.

On the reviewed cortex sample, the shared eligibility check retained 115 of 188 candidate components and excluded 73. These are application quality thresholds, not a validated biological cell classification; the criteria are documented in HOW_IT_WORKS.md.

Validation: 238 unit tests passed in each of the CPU and DirectML environments, including shared display/analysis eligibility, soma-free and multi-soma rejection, large-stack checks, and dense/sparse display caches. Integration checks: 13 passed, 3 skipped. Packaged executables are checked against the release source and smoke-tested before publication.

Download NVAP-0.1.8-windows-cpu.exe for standard Windows use, or NVAP-0.1.8-windows-directml.exe for DirectML GPU support. SHA256SUMS.txt contains checksums for both executables.

NVAP v0.1.7 — Show analyzed microglia only

Choose a tag to compare

@gero-studios gero-studios released this 22 Oct 01:20

The segmented microglia view now shows only accepted cell labels used by analysis. Unassigned fluorescence no longer appears as a green overlay, including when no cells are accepted. Each accepted cell keeps its individual tint in All and isolated views.

The assignment rules are unchanged: an orphan branch network needs a credible cell body or an unambiguous measured connection to an existing cell before it is included in cell metrics.

Validated on the saved native cortex stack: all 13 accepted cells are preserved in both dense and sparse label caches, while 113,454 unassigned above-threshold voxels are hidden. The active threshold and slice trim still apply.

Download the CPU executable for standard use, or DirectML for Windows GPU acceleration. SHA256SUMS.txt contains checksums for both downloads.

NVAP 0.1.6 — Recover missed microglia cell assignments

Choose a tag to compare

@gero-studios gero-studios released this 21 Oct 23:41

NVAP 0.1.6 improves actual microglia cell and branch assignments, beyond the display-preservation fix in 0.1.5.

Fixed and improved

  • Unassigned signal regions are checked for credible cell bodies at the active threshold, so a brighter neighbouring cell does not prevent detection of a dimmer soma.
  • Branches can recover a cell owner through fainter measured signal paths. These paths establish ownership without adding below-mask bridge voxels to measured cells. Fragments with competing owners or no supporting signal path remain unassigned.
  • Smaller accepted cells are preserved when larger cells are present; the previous 15,000-voxel preference now affects ranking only.
  • Region growth preserves diagonal connections between Z planes.
  • The Microglia Viewer reports the percentage of visible signal assigned within the current slice trim. Remaining unassigned signal stays visible in green.

Verification on the saved cortex reference

At unchanged thresholds, assigned above-threshold signal increased from 68.2% to 92.9%, and accepted cell candidates increased from 11 to 13. The correction assigns 398,481 additional above-threshold voxels to actual cell labels. No previously assigned voxels became unassigned.

All 13 candidates produced per-cell soma and process measurements on the native stack. The test suite passed in both CPU and DirectML development environments (239 passed, 3 skipped in each).

The remaining 7.1% is retained visibly as unassigned signal. Assignment coverage is not a claim of biological ground-truth accuracy.

Windows downloads

  • NVAP-0.1.6-windows-cpu.exe — standalone Windows x64 CPU build.
  • NVAP-0.1.6-windows-directml.exe — standalone Windows x64 build with DirectML GPU support and CPU fallback.
  • SHA256SUMS.txt — checksums for both executables.

After updating, rerun Segmentation and Analysis to refresh cell labels and measurements. Choose Cell: All to see the complete signal and assignment coverage.

NVAP 0.1.5 — Preserve microglia signal after segmentation

Choose a tag to compare

@gero-studios gero-studios released this 21 Oct 22:48

NVAP 0.1.5 fixes microglia structures disappearing from the colored post-segmentation All view.

Fixed

  • The colored All view now preserves the full above-threshold green signal, including voxels that segmentation has not assigned to an accepted cell. Detected cells keep their individual tints; unassigned signal stays green.
  • Unassigned signal is display-only and is not counted as an extra cell or added to per-cell analysis.
  • Large-volume rendering preserves thin foreground branches during downsampling instead of skipping them between sampled positions.
  • Colored surfaces use the same opacity as the original intensity view.

Verification

  • On the saved cortex reference, the old view omitted 511,935 of 1,608,598 above-threshold voxels. The corrected All view preserves the complete threshold mask, with zero missing voxels in both dense and memory-saving cache modes.
  • Threshold and slice trimming remain active. Source intensities, cell labels, and analysis measurements are unchanged.

Windows downloads

  • NVAP-0.1.5-windows-cpu.exe — standalone Windows x64 CPU build.
  • NVAP-0.1.5-windows-directml.exe — standalone Windows x64 build with DirectML GPU support and CPU fallback.
  • SHA256SUMS.txt — checksums for both executables.

Download the appropriate executable and run it. To view the corrected segmented result, run Segmentation and choose Cell: All in the Microglia Viewer with View one microglia enabled.

NVAP 0.1.4 — Distinct microglia component tints

Choose a tag to compare

@gero-studios gero-studios released this 21 Oct 14:35

NVAP 0.1.4 makes segmented microglia easier to distinguish in 3D: each detected component has its own tint.

Fixed and improved

  • All components: distinct tints now appear in both volume and surface rendering.
  • Single-cell inspection: selecting a cell preserves the tint it had in the All view.
  • Touching cells: surface rendering preserves each component's boundary and color.
  • Large stacks: component coloring also works with the memory-saving component cache.
  • Green threshold and slice trimming still apply. Enhanced intensities and analysis data are unchanged.

Using the new colors

Run Segmentation, then choose Cell: All in the Microglia Viewer with View one microglia enabled. Choose a cell index to inspect that component individually.

Windows downloads

  • NVAP-0.1.4-windows-cpu.exe — standalone Windows x64 CPU build.
  • NVAP-0.1.4-windows-directml.exe — standalone Windows x64 build with DirectML GPU support and CPU fallback.
  • SHA256SUMS.txt — checksums for both executables.

Download the appropriate executable and run it. Python does not need to be installed.

NVAP 0.1.3 — Cleanup, export, and project recovery fixes

Choose a tag to compare

@gero-studios gero-studios released this 16 Oct 19:52

NVAP 0.1.3 fixes red-channel cleanup, single-scan CSV export, and missing bundled projects, and includes the current vascular and microglia analysis improvements.

Fixed and improved

  • Red cleanup evaluates connections inside the retained Z slices and separates corner-touching debris. Red rendering stays on the cleaned native grid so interpolation cannot recreate detached dots.
  • Analytics includes Export Current Dataset CSV. Exports compute missing or stale results in the background and include channel metrics, vascular morphometry, per-cell microglia measurements, neurovascular summaries, provenance, and vascular masks.
  • Bundled sample projects use persistent locations in packaged builds. Obsolete temporary-folder duplicates are migrated into one entry. Moved projects can be relinked while preserving pins and history.
  • First and last Z trims default to 10 slices each.
  • Updated desktop presentation, application icon, and inspector layout.
  • Update checks follow newer GitHub releases and select the matching Windows build.

Windows downloads

  • NVAP-0.1.3-windows-cpu.exe — standalone Windows x64 CPU build.
  • NVAP-0.1.3-windows-directml.exe — standalone Windows x64 build with DirectML GPU support and CPU fallback.
  • SHA256SUMS.txt — checksums for both executables.

Download the appropriate executable and run it. Python does not need to be installed. Existing microscopy source files are not modified by the update.

NVAP 0.1.2 — Microglia branch preservation + vasculature tools

Choose a tag to compare

@gero-studios gero-studios released this 14 Jul 13:13

NVAP 0.1.2 — Microglia branch preservation + vasculature tools

Existing NVAP installs will show an in-app upgrade prompt for this build.

Fixed

  • Microglia branch preservation. When isolating a single microglia ("View one microglia"), thin processes were being eroded and left out of both the separation and the analysis. Two causes are fixed:
    • Anisotropic stacks now keep full 8-connectivity within each imaging plane, so thin diagonal branches no longer fragment under the watershed and get dropped from the cell.
    • The separation growth mask now always includes every voxel at or above the render threshold you set, so an internal data-driven floor can never apply a stricter threshold than yours and trim visible dim branches.

Added

  • Vasculature workbench: background enhancement and compact red-blob removal, with UI controls.

Downloads

  • NVAP-0.1.2-windows-cpu.exe — CPU-safe build (Python 3.13). Recommended default; works everywhere.
  • NVAP-0.1.2-windows-directml.exe — DirectML GPU-optimized build for AMD/Intel GPUs (Python 3.12), with CPU fallback included.

A Windows SmartScreen / antivirus warning on first launch is expected for unsigned PyInstaller executables; choose "More info → Run anyway".

NVAP 0.1.1 — Windows GPU + CPU fallback update

Choose a tag to compare

@gero-studios gero-studios released this 13 Jul 04:08

NVAP 0.1.1 update.
Automatic green/red thresholding
CZI loading and metadata spacing support

NVAP 0.1.0 — Windows CPU build

Choose a tag to compare

@gero-studios gero-studios released this 08 Jul 13:42

NVAP 0.1.0 for Windows — two build variants:

NVAP.exe — CPU-safe build (819.9 MB). No Torch bundled, smaller download, avoids the AV/SmartScreen false-positive risk that unsigned Torch-bundled PyInstaller binaries tend to trigger.
SHA-256: e1e9fe98a192d0bc14d2f8c4ea2852fbfa29360fa363d84c565ba30e3bd28815

NVAP-0.1.0-windows-directml.exe — DirectML GPU build for AMD/Intel GPUs (938.3 MB). Bundles Torch + torch-directml.
SHA-256: 7fb020caded4b9c69fdfc2662b6e8ac8c5bc23372584a0dda8db868f6f1fe983

Both are unsigned executables — Windows SmartScreen or your antivirus may flag them on first download. See the docs for verification steps: https://nvap-download.restless-cloud-e25f.workers.dev/docs.html#install