Configuration Reference

All tuneable parameters live in core/config.py. Edit that file before running a registration pipeline.

File paths

Parameter	Description
`SOURCE_WSI_PATH`	Absolute path to the moving (source) whole-slide image (`.tiff`).
`TARGET_WSI_PATH`	Absolute path to the fixed (target) whole-slide image (`.tiff`).
`FIXED_NUCLEI_CSV`	Path to a pre-computed nuclei CSV for the fixed image (optional).
`MOVING_NUCLEI_CSV`	Path to a pre-computed nuclei CSV for the moving image (optional).

Each CSV must contain at least the columns global_x and global_y (nucleus centroid coordinates in pixels at REGISTRATION_RESOLUTION). An optional area column enables shape-aware registration.

Resolution parameters

Parameter	Description
`PREPROCESSING_RESOLUTION`	Objective power (magnification) at which WSIs are loaded for coarse rigid and elastic registration. Default: `0.625`.
`REGISTRATION_RESOLUTION`	Objective power (magnification) at which nuclei are detected and fine shape-aware registration is performed. Default: `40`.

REGISTRATION_RESOLUTION must be strictly greater than PREPROCESSING_RESOLUTION.

Patch parameters

Parameter	Description
`PATCH_SIZE`	`(height, width)` of each patch in pixels. Default: `(1000, 1000)`.
`PATCH_STRIDE`	Stride between patches. Default: `(1000, 1000)`.
`VISUALIZATION_SIZE`	Canvas size for visualisation output. Default: `(5000, 5000)`.

Nuclei detection parameters

Parameter	Description
`FIXED_THRESHOLD`	Binarisation threshold for the fixed image (0–255). Default: `100`.
`MOVING_THRESHOLD`	Binarisation threshold for the moving image (0–255). Default: `50`.
`MIN_NUCLEI_AREA`	Minimum nucleus area in pixels². Default: `200`.
`GAMMA_CORRECTION`	Gamma value applied before nuclei detection. Default: `0.4`.

`RegistrationParams` class

class core.config.RegistrationParams[source]

Bases: object

DISPLACEMENT_SIGMA = 10.0

INTERPOLATION_METHOD = 'linear'

MAX_DISPLACEMENT = 500.0

MNN_SAMPLE_SIZE = 5000

`VisualizationParams` class

class core.config.VisualizationParams[source]