allensdk.internal.brain_observatory.roi_filter_utils module

allensdk.internal.brain_observatory.roi_filter_utils.CRITERIA()

class allensdk.internal.brain_observatory.roi_filter_utils.TrainingLabelClassifier(criteria)

Bases: object

Very basic threshold_based classifier.

Has a decision function that is just the number of distinct criteria met by the classifier. Criteria are defined as a list of strings used with pandas.DataFrame.eval.

Parameters:

criteria : list

List of evaluation strings.

decision_function(self, X)

Get the distance from the decision boundary.

Parameters:

X : array-like

Features for each ROI.

Returns:

T : array-like

Distance for each sample from the decision boundary.

class allensdk.internal.brain_observatory.roi_filter_utils.TrainingMultiLabelClassifier(criteria=None)

Bases: object

Multilabel classifier using groups of TrainingLabelClassifiers.

This was used to generate labeling for training the original SVM for classification.

Parameters:

criteria : dictionary

Label names and criteria for each label.

get_eXcluded(self, X)

Get the calculated value of the eXcluded column.

This is useful for comparison with the original classifier implementation.

Parameters:

X : pandas.DataFrame

Object features from the object list file.

Returns:

numpy.ndarray

Calculated eXcluded score from the classifier.

label_data(self, X, as_columns=True)

Generate labels for each row in X.

Parameters:

X : pandas.DataFrame

Object features from the object list file.

Returns:

numpy.ndarray

Array of label codes representing the combination of labels found for each row.

allensdk.internal.brain_observatory.roi_filter_utils.calculate_max_border(motion_df, max_shift)

Calculate motion boundary from frame offsets.

When the motion correction algorithm fails to find sufficient matches, it generates very large frame offsets. The use of max_shift avoids filtering too many cells due to the large offsets, with the tradeoff that those frames will be noise.

Parameters:

motion_df : pandas.DataFrame

Dataframe containing the x, y offsets from motion correction.

max_shift : float

Maximum shift to allow when considering motion correction. Any larger shifts are considered outliers.

Returns:

list

[right_shift, left_shift, down_shift, up_shift]

allensdk.internal.brain_observatory.roi_filter_utils.get_indices_by_distance(object_list_points, mask_points)

Find indices of nearest neighbor matches.

Require a distance of 0 (perfect match) and a unique match between masks and object_list entries.

allensdk.internal.brain_observatory.roi_filter_utils.get_rois(segmentation_stack, border=None)

Extract a list of rois from the segmentation data array.

Parameters:

segmentation_stack : numpy.ndarray

The array from the maxInt_masks file showing the object masks.

border : list

[right_shift, left_shift, down_shift, up_shift] bounding box determined from motion correction.

Returns:

list

List of RoiMask objects.

allensdk.internal.brain_observatory.roi_filter_utils.order_rois_by_object_list(object_data, rois)

Reorder rois by matching bounding boxes to object list.

Parameters:

object_data : pandas.DataFrame

Object list data.

rois : list

List of RoiMasks.

Returns:

list

The list of rois reordered to index the same as object_data.