Indexing a multi-dimensional Numpy array Accessing a multi-dimensional Numpy array by indices is useful for many tasks, such as parsing the contours in a DICOM-RTSTRUCT file.
As a simple example, we are given a 5 x 6 array arr, and a list of the coordinates of 3 points coords:
>>> import numpy as np >>> arr = np.arange(30).reshape(5,6) >>> coords = np.array([[0, 1], [3, 4], [3, 2]]) >>> arr array([[ 0, 1, 2, 3, 4, 5], [ 6, 7, 8, 9, 10, 11], [12, 13, 14, 15, 16, 17], [18, 19, 20, 21, 22, 23], [24, 25, 26, 27, 28, 29]]) >>> coords array([[0, 1], [3, 4], [3, 2]]) To obtain the values of the targeted pixels as a list [arr[0,1], arr[3,4], arr[3,2]], that is [1, 22, 20], it is tempting to write “arr[coords]“.

DICOM-RTSTRUCT Contour Data Contours drawn for radiotherapy are saved as DICOM RT Structure Set (“RT” stands for radiotherapy.) in DICOM Standard, and usually as a single file. You can locate this file among CT or MRI data sets quite reliably, by traversing recursively through the directories and looking for MODALITY of “RTSTRUCT”.
In this modality, the contours are saved as 2D polygons slice by slice under the Contour Data (3006,0050) tag, which species the data format as:

© 2015–2019, Zheng Zhang · Powered by the Academic theme for Hugo.