Parameter reference

Signature defaults are listed separately from settings chosen in the tutorials. The tables below are generated from the current source. Follow each method link for prerequisites, returned values, side effects, and source evidence.

ImageCollection

bactoscoop.imagecollection.ImageCollection

ImageCollection(image_folder_path=None, px=0.065, error_policy='continue', log_cell_errors=False)

  • image_folder_path — default None. Dataset folder containing TIFFs and the masks subfolder; None delays path configuration.

  • px — default 0.065. Micrometres per pixel. Geometry arrays remain in pixel coordinates.

  • error_policy — default 'continue'. “continue” records independent item/cell failures; “raise” raises the recorded error. Setup failures can still stop work.

  • log_cell_errors — default False. Print each recorded cell error when true; recording is independent of this flag.

processing_summary

bactoscoop.imagecollection.ImageCollection.processing_summary()

processing_summary()

load_masks

bactoscoop.imagecollection.ImageCollection.load_masks()

load_masks()

load_phase_images

bactoscoop.imagecollection.ImageCollection.load_phase_images()

load_phase_images(phase_channel='')

  • phase_channel — default ''. TIFF suffix identifying the phase images, such as “C1”. An empty suffix is the signature default.

load_channel_images

bactoscoop.imagecollection.ImageCollection.load_channel_images()

load_channel_images(channel_list)

  • channel_list — required. Sequence of channel suffixes to load or attach.

create_image_objects

bactoscoop.imagecollection.ImageCollection.create_image_objects()

create_image_objects(phase_channel='')

  • phase_channel — default ''. TIFF suffix identifying the phase images, such as “C1”. An empty suffix is the signature default.

add_channels

bactoscoop.imagecollection.ImageCollection.add_channels()

add_channels(img_objects, channel_list, load_data=True)

  • img_objects — required. Image objects to attach channel data to.

  • channel_list — required. Sequence of channel suffixes to load or attach.

  • load_data — default True. Load channel TIFFs before attaching them when true; false uses already loaded field-keyed data.

get_mesh_dataframe

bactoscoop.imagecollection.ImageCollection.get_mesh_dataframe()

get_mesh_dataframe(image_name)

  • image_name — required. See the linked implementation.

create_image_object

bactoscoop.imagecollection.ImageCollection.create_image_object()

create_image_object(image, image_name, index, mask, mesh_df, px)

  • image — required. See the linked implementation.

  • image_name — required. See the linked implementation.

  • index — required. See the linked implementation.

  • mask — required. See the linked implementation.

  • mesh_df — required. See the linked implementation.

  • px — required. Micrometres per pixel. Geometry arrays remain in pixel coordinates.

segment_images

bactoscoop.imagecollection.ImageCollection.segment_images()

segment_images(mask_thresh, minsize, n, model_name='bact_phase_omni')

  • mask_thresh — required. Threshold passed as Omnipose mask_threshold.

  • minsize — required. Minimum accepted segmentation-mask size in pixels.

  • n — required. Selected image positions; wrapper accepts None for all. Indices must be nonempty, unique valid integers.

  • model_name — default 'bact_phase_omni'. Omnipose model identifier; default collection model is bact_phase_omni.

batch_detect_objects

bactoscoop.imagecollection.ImageCollection.batch_detect_objects()

batch_detect_objects(channels=None, reset_channels=True, align=False, smoothing=0.1, log_sigma=3, kernel_width=4, min_overlap_ratio=0.01, max_external_ratio=0.1)

  • channels — default None. Requested channel identifiers; for correlation, channel pairs are generated from this sequence.

  • reset_channels — default True. Clear loaded channel dictionaries and background/interpolation caches before loading requested channels.

  • align — default False. Use the object-detection alignment path for the selected signal.

  • smoothing — default 0.1. Smoothing passed to the contour/mesh fitting path.

  • log_sigma — default 3. Laplacian-of-Gaussian sigma for signal-object detection.

  • kernel_width — default 4. Width of the square dilation kernel used for candidate object masks.

  • min_overlap_ratio — default 0.01. Minimum accepted overlap ratio under the utility implementation; inspect its denominator in source.

  • max_external_ratio — default 0.1. Maximum accepted outside-cell ratio under the utility implementation.

batch_process_mesh

bactoscoop.imagecollection.ImageCollection.batch_process_mesh()

batch_process_mesh(pkl_name=None, object_list=None, save_data=True, phase_channel='', join_thresh=4, split_thresh=0.35, CD_width=False, smoothing=0.1, neighbor_filter_max_neighbors=None, neighbor_filter_connectivity=1, max_daughter_cell_mesh_rows=800)

  • pkl_name — default None. Mesh input filename or pickle output name. For feature output, .pkl means exact filename; a suffixless name is a legacy tag.

  • object_list — default None. Explicit image-object sequence to process; None selects collection targets.

  • save_data — default True. Write the combined mesh table when true.

  • phase_channel — default ''. TIFF suffix identifying the phase images, such as “C1”. An empty suffix is the signature default.

  • join_thresh — default 4. Maximum pole-to-pole joining distance in pixels.

  • split_thresh — default 0.35. Relative constriction threshold used by the splitting pipeline.

  • CD_width — default False. Use the width-based constriction path when true; otherwise use the pipeline phase-signal path.

  • smoothing — default 0.1. Smoothing passed to the contour/mesh fitting path.

  • neighbor_filter_max_neighbors — default None. Maximum number of touching distinct labels allowed before meshing; None disables the prefilter.

  • neighbor_filter_connectivity — default 1. 1 or 4 for edge contact; 2 or 8 for edge and diagonal contact.

  • max_daughter_cell_mesh_rows — default 800. Maximum rebuilt mesh rows per daughter during cell splitting (default 800; integer >= 4). Rejects oversized predicted counts before reconstruction; both final daughters must have 4..limit rows. Does not limit unsplit parents and is separate from feature extraction max_mesh_size.

batch_load_mesh

bactoscoop.imagecollection.ImageCollection.batch_load_mesh()

batch_load_mesh(pkl_name, pkl_path=None, phase_channel='')

  • pkl_name — required. Mesh input filename or pickle output name. For feature output, .pkl means exact filename; a suffixless name is a legacy tag.

  • pkl_path — default None. Directory containing an input mesh file; None selects the image folder.

  • phase_channel — default ''. TIFF suffix identifying the phase images, such as “C1”. An empty suffix is the signature default.

batch_mask2mesh

bactoscoop.imagecollection.ImageCollection.batch_mask2mesh()

batch_mask2mesh(pkl_name=None, object_list=None, save_data=True, phase_channel='', smoothing=0.1)

  • pkl_name — default None. Mesh input filename or pickle output name. For feature output, .pkl means exact filename; a suffixless name is a legacy tag.

  • object_list — default None. Explicit image-object sequence to process; None selects collection targets.

  • save_data — default True. Write the combined mesh table when true.

  • phase_channel — default ''. TIFF suffix identifying the phase images, such as “C1”. An empty suffix is the signature default.

  • smoothing — default 0.1. Smoothing passed to the contour/mesh fitting path.

batch_shift_correction

bactoscoop.imagecollection.ImageCollection.batch_shift_correction()

batch_shift_correction(shifted_channel, reset_channels=False, log_sigma=1.5, kernel_width=3, min_overlap_ratio=0.1, max_external_ratio=1, phase_log_sigma=0.5, phase_closing_level=2, signal_closing_level=12, max_shift_correction=15)

  • shifted_channel — required. See the linked implementation.

  • reset_channels — default False. Clear loaded channel dictionaries and background/interpolation caches before loading requested channels.

  • log_sigma — default 1.5. Laplacian-of-Gaussian sigma for signal-object detection.

  • kernel_width — default 3. Width of the square dilation kernel used for candidate object masks.

  • min_overlap_ratio — default 0.1. Minimum accepted overlap ratio under the utility implementation; inspect its denominator in source.

  • max_external_ratio — default 1. Maximum accepted outside-cell ratio under the utility implementation.

  • phase_log_sigma — default 0.5. See the linked implementation.

  • phase_closing_level — default 2. See the linked implementation.

  • signal_closing_level — default 12. See the linked implementation.

  • max_shift_correction — default 15. See the linked implementation.

batch_calculate_features

bactoscoop.imagecollection.ImageCollection.batch_calculate_features()

batch_calculate_features(channel_method_pairs, all_data=True, reset=True, use_shifted_contours=False, shift_signal=False, max_mesh_size=600, retain_contour_on_object_mesh_failure_channels=None)

  • channel_method_pairs — required. Pairs of (channel list, feature method), for example [([None], “morphological”), ([“C2”], “profiling”)].

  • all_data — default True. Feature-method-specific option; objects includes detailed per-object arrays when true.

  • reset — default True. Clear stored feature dataframes when true; false retains other pairs while replacing each recalculated pair.

  • use_shifted_contours — default False. Select stored shifted contour, midline, and mesh geometry; the correction must already exist.

  • shift_signal — default False. Enable the method-specific optional signal cropping/shift path; this is not a universal object-coordinate shift.

  • max_mesh_size — default 600. Maximum accepted mesh row count; cells above this limit are removed from the image cell population.

  • retain_contour_on_object_mesh_failure_channels — default None. Channels whose object rows keep contour measurements and NaN mesh aggregates when object geometry fails.

batch_calculate_signal_correlation_features

bactoscoop.imagecollection.ImageCollection.batch_calculate_signal_correlation_features()

batch_calculate_signal_correlation_features(df, channels, feature_method_tuples)

  • df — required. DataFrame to inspect, transform, or augment. Correlation mutates it in place and returns it.

  • channels — required. Requested channel identifiers; for correlation, channel pairs are generated from this sequence.

  • feature_method_tuples — required. Pairs of (feature-name list, correlation-selector list) applied to each channel pair.

curate_dataset

bactoscoop.imagecollection.ImageCollection.curate_dataset()

curate_dataset(path_to_model, cols=4, control=False, save_curated_data=True)

  • path_to_model — required. Path to a serialized SVM carrying the feature_names_in_ training-column schema.

  • cols — default 4. Legacy curation argument retained for compatibility; unused by the current curation preparation.

  • control — default False. Request visual examples of retained/rejected classes; random selection can be unavailable for small classes.

  • save_curated_data — default True. Write curated mesh pickle when true; in-memory curation always occurs.

add_meshdata_to_dataframe

bactoscoop.imagecollection.ImageCollection.add_meshdata_to_dataframe()

add_meshdata_to_dataframe()

export_curated_masks

bactoscoop.imagecollection.ImageCollection.export_curated_masks()

export_curated_masks(output_subfolder='curated_masks', overwrite=True)

  • output_subfolder — default 'curated_masks'. Destination for newly rasterized curated masks, relative to the dataset folder.

  • overwrite — default True. Replace existing curated mask TIFFs when true; false raises for a file already present.

merge_dataframes

bactoscoop.imagecollection.ImageCollection.merge_dataframes()

merge_dataframes(include_metadata_tag=False, discard_morphological_nan=False)

  • include_metadata_tag — default False. Rename identity columns with Metadata_ prefixes.

  • discard_morphological_nan — default False. Drop merged rows missing cell_area or cell_length; these columns must exist.

dataframe_to_pkl

bactoscoop.imagecollection.ImageCollection.dataframe_to_pkl()

dataframe_to_pkl(pkl_name=None)

  • pkl_name — default None. Mesh input filename or pickle output name. For feature output, .pkl means exact filename; a suffixless name is a legacy tag.

meshdata_to_parquet

bactoscoop.imagecollection.ImageCollection.meshdata_to_parquet()

meshdata_to_parquet(parquet_name=None)

  • parquet_name — default None. Output filename ending in .parquet within the image folder; None chooses the dataset-based default.

dataframe_to_parquet

bactoscoop.imagecollection.ImageCollection.dataframe_to_parquet()

dataframe_to_parquet(parquet_name=None)

  • parquet_name — default None. Output filename ending in .parquet within the image folder; None chooses the dataset-based default.