We collected phase-contrast MRI data (5-mm midline sagittal section, rostro-caudal direction) and obtained average velocity in certain Regions-0f-Interest (ROIs). The measured ROIs are cerebellar tonsil, ventral spinal subarachnoid space, dorsal spinal subarachnoid space, upper portion of syrinxes, upper cervical cord, and medulla. Data obtained from 18 Chiari-I patients (preoperative and postoperative) and 21 controls are included. Because the preoperative MRI data was missing in one patient, the preoperative studies included 17 MRI sessions. The unit of the data is cm/sec. The data are synchronized with pre-processing so that the rise of the caudal CSF movement be placed at the center of the cardiac cycle.

We performed phase-contrast MRI on 18 patients with Chiari-I malformationc(associated with syringomyelia) and 21 controls. The MRI on the patientscwas taken before andcafter surgery (foramen magnum decompression).

Phase-contrast MRI was takn on the midline sagittal section with 5-mmcthickness. The movement in the FH (foot-head) direction was encoded withcvelocity encoding (VENC) of 10 cm/sec.

On the data of each MRI session, six regions of interest (ROIs) were placed; namely, cerebellar tonsil (tonsil), ventral subarachnoid space just below the foramen magnum (sa), dorsal subarachnoid space just below the foramen magnum (cm), the upper portion of the syrinx (syrinx), the cord segment between the fouth ventricle and the syrinx (cord), the medulla (medulla). The average flow speed inside each ROI was measured at each time point in the cardiac cycle.

The acquired data were synchronized with post-processing so that the maximal caudal velocity of 'sa' will be at the center of the data of 50 bins encompassing a cardiac cycle.

The data file is a json-format dump of a python object. Because the original python object contained numpy array, we had to encode the object into the one with normal python lists. To restore the original python object, please use the function 'decode_'.

The file '57.mp4' is the movie of phase-contrast MRI of the session No. 57, which is a preoperative session of a Chiari-I patient. The data is shown with color coding with the rostral velocity encoded in yellow and the caudal velocity encoded in dark blue. This movie shows the paradoxical upward movement of the upper spinal cord during the phase where the CSF in the subarachnoid space is moving downward.

The 'encoded_data.json' file contains all the data of the MRI sessions processed as above. This file is was created by dumping a python object containing all the data using the 'json' module of python version 3. It can be imported to restore the original python object using the 'json' module as follows. A function in the 'analyze.py' (decode_data()) is needed for this restoration.

---------------python code----------------
import json
with open('encoded_data.json', 'rt') as f:
    encoded_data = json.load(f)
    datafile = decode_data(encoded_data)
------------------------------------------

# Usage of analyze.py

The 'datafile' is a python dictionary with the keys indicating the MRI
session number.

Various functions to manupulate this data file are provided in
'analyze.py'.

To obtain the list of session_id of preoperative studies.
syrinx_sessions_raw()
which gives the list [1, 5, 9, 14, 24, 28, 36, 41, 45, 52, 55, 57, 60, 62, 68, 80, 81]

To obtain the list of session_id of control studies.
control_sessions_raw()
which gives the list [70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93]

To obtain the list of session_id of postoperative studies.
postop_sessions()
which gives the list [4, 8, 13, 16, 23, 35, 40, 44, 47, 51, 54, 56, 59, 66, 67, 69, 82, 94]

The correspondence between the preoperative session_id and the postoperative sesion_id in a same patient is given below.

[[1,4], [5,8], [9,13], [14,16], [24,51], [28,35], [36,40], [41,44], [45,47], [52,54], [55,56], [57,59], [60,69], [62,66], [68,94], [80,67], [81,82]]

The figures in the article were created using the functions
'draw_mean_graph()' and 'draw_two_locations_with CI()'.