Command line usage - Alignment¶

visualqc_alignment: rate quality of alignment between two images.

usage: visualqc_alignment [-h] -d IN_DIR -i1 IMAGE1 -i2 IMAGE2 [-l ID_LIST] [-o OUT_DIR]
                          [-vt {Checkerboard,Edges_Thinner,Voxelwise_diff,Animate,Color_mix,Edges_Diffused}]
                          [-dl DELAY_IN_ANIMATION] [-olm OUTLIER_METHOD] [-olf OUTLIER_FRACTION] [-olt OUTLIER_FEAT_TYPES]
                          [-old] [-w VIEWS [VIEWS ...]] [-s NUM_SLICES] [-r NUM_ROWS] [-so] [-p]

Input and output¶

`-d, --in_dir`	Absolute path to an input folder containing the MRI scan. Each subject will be queried after its ID in the metadata file, and is expected to have the MRI (specified `--mri_name`), in its own folder under –user_dir. E.g. `--user_dir /project/images_to_QC`
`-i1, --image1`	Specifies the name of the first 3d image to serve as the reference image.
`-i2, --image2`	Specifies the name of second 3d image to serve as the comparison image Order of the two images does not typically matter.
`-l, --id_list`	Abs path to file containing list of subject IDs to be processed. If not provided, all the subjects with required files will be processed. E.g. sub001 sub002 cn_003 cn_004
`-o, --out_dir`	Output folder to store the visualizations & ratings. Default: a new folder called `visualqc` will be created inside the input folder

Visualization¶

Customize behaviour of comparisons

`-vt, --vis_type`
	Possible choices: Checkerboard, Edges_Thinner, Voxelwise_diff, Animate, Color_mix, Edges_Diffused Specifies the visualization type to start with. You can change this via radio buttons as you go along. Default: Edges_Thinner.
`-dl, --delay_in_animation`
	Specifies the delay in animation of the display of two images (like in a GIF). Default: 0.6 (units in seconds).

Outlier detection¶

options related to automatically detecting possible outliers

`-olm, --outlier_method`
	Method used to detect the outliers. For more info, read http://scikit-learn.org/stable/modules/outlier_detection.html Default: isolation_forest.
`-olf, --outlier_fraction`
	Fraction of outliers expected in the given sample. Must be >= 1/n and <= (n-1)/n, where n is the number of samples in the current sample. For more info, read http://scikit-learn.org/stable/modules/outlier_detection.html Default: 0.2.
`-olt, --outlier_feat_types`
	Type of features to be employed in training the outlier detection method. It could be one of . Default: (‘MSE’,).
`-old, --disable_outlier_detection`
	This flag disables outlier detection and alerts altogether.

Layout options¶

`-w, --views`	Specifies the set of views to display - could be just 1 view, or 2 or all 3. Example: –views 0 (typically sagittal) or –views 1 2 (axial and coronal) Default: 0 1 2 (show all the views in the selected segmentation)
`-s, --num_slices`
	Specifies the number of slices to display per each view. This must be even to facilitate better division. Default: 12.
`-r, --num_rows`	Specifies the number of rows to display per each axis. Default: 2.

Workflow¶

Options related to workflow e.g. to pre-compute resource-intensive features, and pre-generate all the visualizations required

`-so, --screenshot_only`
	This flag enables the batch generation of screenshots of the visualizations generated, for archival purposes. This would skip the interactive and deeper review of the visualizations, and directly saves the screenshots to the output folder. Hence, only some static options would work, and dynamic animations would not. This is NOT recommended as a QC procedure as the generated screenshots can become out of sync with the actual data for a number of reasons, and for reliable and accurate QC, we recommend interactive review which presents fresh visualizations based on latest version of data on disk.
`-p, --prepare_first`
	This flag does the heavy preprocessing first, prior to starting any review and rating operations. Heavy processing can include computation of registration quality metrics and outlier detection etc. This makes the switch from one subject to the next, even more seamless (saving few seconds :) ). Default: False.