Extend registration_pipelines and bench_registration with stock antspyx
ANTs presets (SyNRA, SyN, Repro sr/br, QuickRepro sr/br) plus the legacy
Rigid -> SyNOnly workflow, and update the deformable benchmark
documentation with follow-up ANTs-variant runs and results.
Migrate register_impact from impact-reg-konfai (pip + one-time glibc/LibTorch
shim) to the vboussot/elastix_impact Docker image running a rigid 6-DOF
EulerTransform with MIND + TotalSegmentator static features and a Mattes-MI
secondary metric. Delete scripts/setup_impact_konfai.sh.
Vendor ImpactLoss (vboussot/ImpactLoss) into src/impact_loss.py and add
impact_loss as a per-case semantic scoring metric in the benchmark runner.
Update docs (AGENTS.md, REGISTRATION_BENCHMARK.md) to reflect the
all-rigid pipeline set and the new Docker-based IMPACT runtime.
This commit introduces a new script, `setup_impact_konfai.sh`, which facilitates a one-time setup for the IMPACT registration pipeline (impact-reg-konfai). The script builds a self-contained shim that includes:
- A glibc 2.35 loader and libraries extracted from the niftyreg:2.1.1-ubuntu Docker image.
- A self-contained LibTorch 2.8.0 (cu128) library.
- A wrapper script that runs the stock elastix-impact binary through the glibc 2.35 loader with the appropriate library paths.
The setup ensures compatibility with hosts that do not meet the required glibc and LibTorch versions, allowing the pipeline to function correctly. The script also verifies the installation and provides usage instructions.
The no-mask CT→T1c FOV mismatch left only ~14% of spatial samples
inside the moving buffer at auto-init, below Elastix's default
RequiredRatioOfValidSamples of 0.25, so the IMPACT pipeline aborted
on every M6 case. Set the gate to 0.1 so the metric's native
in-buffer evaluation proceeds while still catching diverging
transforms.
Switch the default semantic model from TotalSegmentator (TS/M730.pt)
to MRSegmentator (MRSeg/MRSeg.pt), a multi-modality MRI+CT model
better suited to this cross-modality task. Update docstrings and
REGISTRATION_BENCHMARK.md with the new per-case MI results.
- Added `--clip-ct LO HI` argument to clip CT intensities before registration and scoring, improving A/B comparisons.
- Implemented `_clip_image` function to handle intensity clipping of images.
- Updated registration functions to utilize CT clipping where beneficial, specifically in `register_niftyreg` and `register_synthmorph`.
- Enhanced `fixed_moving` function to ensure CT is always treated as the fixed image.
- Introduced Docker support for NiftyReg v2.1.1, including a Dockerfile for building the image.
- Updated documentation and help messages to reflect new features and usage instructions.
- Adjusted evaluation criteria for registration quality to account for intensity clipping.
Add `register_niftyreg` as the sixth rigid CT→T1c pipeline using
`reg_aladin` (symmetric block matching + trimmed least squares, NCC
similarity). Runs natively via `conda install -c conda-forge niftyreg`;
no Docker required.
Change the benchmark's primary reported metric from geometric skull-ring
QA to the achieved Mattes MI (`mi_ct_mr`, nats) between the warped CT
and the original MR over the MR head. Geometric QA fields are still
recorded per case in the JSON output.
Raise the SimpleITK benchmark-mode optimizer budget from 150 to 2000
iterations per pyramid level so all toolkits are compared on an equal
iteration footing.
Update `REGISTRATION_BENCHMARK.md` (five→six pipelines, niftyreg row,
elastix bin count 10→32), `AGENTS.md` (toolkit list), and
`bench_registration.py` docstrings/usage accordingly.
Replace the shared CT head-box resampling and manual centroid pre-shift with
each toolkit's native automatic transform initialization (elastix/impact use
CenterOfGravity, antspyx uses center-of-mass, simpleitk uses identity). All
pipelines now register on the full volumes instead of a cropped head box.
Drop the _ct_head_box_mm helper and update the benchmark table to document
per-toolkit init methods in a dedicated "init" column.
Drop the _ct_head_box_resample utility that was only needed for centered
initialization, since centroid alignment is now scoped to local optimizers
(antspyx, impact) rather than all pipelines. Reorder the pipeline table and
module docstring alphabetically and add explicit fixed/moving and centroid-init
columns so the per-toolkit registration convention is unambiguous.
Replace the unavailable DeepReg benchmark path with the freesurfer/synthmorph
Docker pipeline for rigid CT-to-T1c registration. Add a shared CT head-box
resample helper for centered initialization and update benchmark docs and CLI
for the new SynthMorph options.
Replace the conda-forge-dependent Elastix CLI path with the pip-installable
ITK-Elastix Python components so the benchmark can run elastix in the main
environment. Add SimpleITK/ITK image and transform conversion helpers, and
use the elastix 5.x EulerTransform rigid pipeline with Mattes MI and ASGD.
Update REGISTRATION_BENCHMARK.md and bench_registration.py documentation to
reflect the new elastix availability, results, and multi-environment merge
workflow. Ignore local elastix output directories in .gitignore.
Update M6 manifest building to register T1c to CT with rigid
Mattes MI on a CT-space head box, including multi-start initial
tilts for large inter-session head motion. The inverse transform
warps CT and GTV into T1c native space, with acceptance based on
concentricity of the warped CT skull ring and MR head silhouette.
Rerun preprocessing when a row has a label but the warped label file
is missing. Normalize volume orientation in read_nii_img.
Add registration benchmark documentation, benchmark script, and
shared registration pipeline helpers.
Add tof, sub, and subtraction tokens to the Lee T1c exclusion regex.
This prevents time-of-flight MRA and post/pre subtraction sequences
from matching T1-based names.
Fiesta/CISS/SSFP are balanced-SSFP (T2-dominant) sequences whose
short TE fools the seq/TE fallback, causing T1c misclassification.
Add them to the exclusion regex and guard against empty candidates.
Expand write_excluded_notes to enumerate all series in a timepoint
(not just T1c candidates) and parse DICOM txt metadata for each
series' description and protocol, producing richer no-T1c reports.
- Updated regex patterns for improved matching of series names and tags.
- Added functionality to reject non-head series based on study and series descriptions.
- Implemented max voxel spacing check to filter out series with excessive spacing.
- Enhanced the reconstruction script to handle dynamic-frame series exclusions and artifact pruning.
- Modified output paths for reconstructed NIfTI files and added QA screenshot generation.
- Improved argument parsing in benchmark and pseudo-labeling scripts for better flexibility.
- Introduced a new script for generating QA screenshots from reconstructed volumes.
Add a new benchmarking script and documentation to facilitate
performance comparisons between the different implemented pipelines.
- Create `scripts/benchmark_pipelines.py` for automated evaluation
- Add `BENCHMARKING.md` to outline benchmarking procedures and metrics
Update the nnU-Net pipeline to operate on native (unprocessed) volumes
instead of preprocessed ones. This allows nnU-Net to utilize its own
`plan_and_preprocess` logic for resampling, cropping, and normalization,
ensuring Pipeline B remains distinct from Pipelines A and C.
- Update `scripts/05_build_splits.py` to include native `img` and `label`
paths in the manifest rows.
- Modify `scripts_nnu/01_nnu_prepare_dataset.py` to symlink native
volumes and implement a label fix-up mechanism for non-conforming grids.
- Update `scripts_nnu/04_nnu_pseudo_label.py` and `05_nnu_eval_test.py`
to use native image paths and perform selection/evaluation in physical
mm on the native grid.
- Refactor `scripts_nnu/nnu_common.py` to handle native-grid label
alignment and volume-based selection gates.
- Update `README.md` to document the preprocessing differences between
Pipelines A/C and Pipeline B.
Introduce the third segmentation pipeline using MONAI (1.6.x) to allow
direct comparison with Pipelines A and B. This includes the implementation
of the iterative pseudo-labeling workflow, training scripts, and
inference protocols.
- Add `scripts_monai/` directory containing the MONAI pipeline scripts.
- Update documentation in `README.md` and `AGENTS.md` to include MONAI
package requirements and pipeline details.
- Configure `.gitignore` to exclude MONAI-specific run directories.
- Update data directory descriptions to include MONAI pseudo-labels.
Update README.md to include detailed project overview, environment
requirements, data sources, and a comprehensive directory layout.
Add nnU-Net pipeline documentation and directory descriptions.
Update .gitignore to exclude nnU-Net specific directories and add
new scripts directory for nnU-Net pipeline.
Add initial nnU-Net pipeline scripts.
Enhance the training pipeline with stateful checkpointing and improve
the resilience of the data loading process against filesystem latency
and transient I/O errors.
- Implement auto-resuming in `train_ddp` by loading model, optimizer,
and scheduler states from `state.pt`.
- Add atomic state saving using temporary files to prevent corruption.
- Introduce `_read_nii` with exponential backoff retries to handle
transient NFS/filesystem failures during NIfTI reading.
- Add explicit error handling for missing or unreadable label files in
`PatchDataset`.
- Update `sliding_window_probs` to conditionally apply Test-Time
Augmentation (TTA) based on the `tta` parameter.
- Add `scripts/test_dataloader.py` for verifying dataset integrity.
Refactor the data loading and preprocessing pipeline to handle edge cases in
medical imaging data, including NaN/Inf values, shape mismatches, and
numerical instability during training.
- Update `01_build_ntuh_manifest.py` with improved regex for T1c detection,
spine exclusion, and deduplication logic based on acquisition timestamps.
- Enhance `PatchDataset` in `src/dataset.py` to handle NaN/Inf values,
clip intensity ranges, and ensure label/image shape alignment via
padding/trimming.
- Add a zero-gradient fallback in `src/training.py` to prevent DDP
synchronization failures when encountering NaN/Inf losses.
- Add `scripts/scan_procs.py` for process monitoring.
- Increase DataLoader timeout to prevent hangs during heavy I/O.
