MEA Instructions Version 3.

13b
© Fabian Ramseyer, Ph.D.
University of Bern
Department for Psychology and Psychotherapy
Gesellschaftsstrasse 49
CH – 3012 Bern, Switzerland

ramseyer@psync.ch | www.psync.ch
General remarks and conditions of use:

MEA may be used freely for educational and research

purposes.
By providing information about your research project
and your affiliation, you agree and certify that you will
not resell and/or modify MEA for profit.
The information you provide on the download form will
only be used for internal statistics and for the purpose
of acquiring funding for future MEA-related projects.

You agree that – when referring to MEA – to always

cite the following reference:

Ramseyer, F. & Tschacher, W. (2011). Nonverbal synchrony in psychotherapy:

Coordinated body-movement reflects relationship quality and outcome.
Journal of Consulting and Clinical Psychology, 79(3), 284-295.
doi:10.1037/a0023419

You are welcome, to also cite the source of the download:

Ramseyer, F. (2017). MEA: Motion Energy Analysis [Computer software].
Retrieved from http://www.psync.ch

Motion Energy Analysis (MEA)

1 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions Version 3.13a & 3.13b
Version 3 was modified to guarantee accurate
analyses across numerous hardware configurations
Version 3.13 only runs on OSX 10.9 >

In contrast to previous versions, the video-files may

now be stored on slower (e.g. external) hard-drives.
The speed of the analysis now depends on the
following factors:
hardware (processor, RAM, graphics card)
storage medium and storage location
(SSD, HD with 10k, 7.2k or 5.4k RPM)

A new, “serial-processing” mode has been added:

If you have a folder with multiple videos (as many as
you like), and if these videos all share similar (or
better: identical) regions of interest, then you may run
the analysis in an automated fashion. Your videos will
be processed one-by-one and the corresponding
text-files with the individual raw-data will be written to
a folder on your desktop (MEA_output).

MEA 3 should run in 64-bit mode. Depending on your

MAC, you have to manually define that the program
runs in 64-bit (see next page).

Motion Energy Analysis (MEA)

2 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions Version 3.13
64 bit Mode

in the finder, locate the program and “get info”

cmd-I (⌘+i)
then, switch off “Start in 32bit mode”

OS X 10.6.8

check by clicking:
About MAX

OS X 10.10

check by clicking:
About MAX

Motion Energy Analysis (MEA)

3 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
basic workflow
Follow green numbers 1 to 7

buttons / areas are clickable

(e.g. click on orange button labeled “8” to select ROI 8)

hover mouse over specific areas for additional

information
check video-output for plausibility of data

Motion Energy Analysis (MEA)

4 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
module overview (modules without numbers)
various information and setting areas are visible. They
may be left untouched; in some cases experimentation
may lead to better results

movie information

ROI – output (Motion Energy)

manual / auto MODE

informations on auto mode

Motion Energy Analysis (MEA)

5 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
(1) read original movie file

program indicates frames per second (fps)

of original video-recording and further info on movie

(2) control video playback

(3) choose resolution (VIDEO size)

Motion Energy Analysis (MEA)

6 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
(2) choose pen size for the definition of ROI’s:
smaller ROIs are better drawn with small pen size (2, 4)
bigger ROIs are better drawn with bigger pen size (8, 15)

before drawing, clear previous ROI drawings

(6) draw one ROI after each other:

– select ROI # to draw (red dot indicates active ROI),
then draw directly onto original movie (left screen)
– make sure to fully cover the entire ROI with color !

size and shape of

each ROI may be
chosen freely
ROIs cancel each
other out
individual ROIs may
be summed later

if different ROIs appear to have the same color, use

“Assign new colors” for better visible differentiation
Motion Energy Analysis (MEA)
7 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
One area in the video may only be
covered by one ROI
ROIs cancel each other out

Keep (logical) position of ROIs constant for all analyses

When analysing e.g. patients and therapists, always
assign ROI1 to e.g. head of Patient, ROI2 body of
patient, ROI3 head of therapist, ROI4 body of therapist,
irrespective of whether patients sat on the left or right
side of the video. A consistent ROI-assignment is
necessary for an automated analysis later on.

If ROIs have been mistakenly drawn too big,

they may be either cleared altogether (by clicking on
“clear all”), or by “erasing” excessive portions with
another – otherwise unused – ROI.

Resutls should always be double-checked for accuracy

Motion Energy Analysis (MEA)

8 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
the adequacy of the chosen region(s) can be assessed
quickly by “zooming” through the entire sequence:
Using the slider above the left movie-screen, it is
possible to get a picture whether the ROI remains
suitable for the duration of the entire sequence.

in the case of highly variable behavior / zones, please

consider splitting up the sequence into shorter (stable)
segments. (Time-series may be stacked together later
on).
Unfortunately, there’s no easy solution for overlapping
regions and for crossing of regions.

Motion Energy Analysis (MEA)

9 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
(4) adjust threshold for movement detection:
difference-image (right screen) indicates the fit of the
threshold. If the threshold is too low, then the signal will
contain too much noise. If the threshold is too high, then
small movements will not be registered.

too low (5) good (15) too high (20)

raw data should contain zeroes (0) when people do not
move, see raw data comparison on next page
Motion Energy Analysis (MEA)
10 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions
(7) the threshold setting determines the signal-to-noise
ratio of the raw data.

too low (5)

good (15)

The best way to determine the correct settings is to

explore various threshold values; then compare the
resulting raw data with the original video. One should see
peaks in times of a subject’s movement and no peaks
when the subject is immobile.
Continuous activation in a ROI (overlay plot 1, above)
usually results from thresholds settings that are too low

raw data may be later be filtered and cleaned (e.g. by

using a moving average) in other programs (e.g. SAS, R).
Motion Energy Analysis (MEA)
11 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions: MANUAL
(5) clear data accumulated up to this point (“clear”).
Or, alternatively, look at data generated (“open”).

(6) Begin analysis

(7) after analysis is done, save the MEA-file (name is

freely choosable), with the extension “.txt”

(1) redo steps 1 to 7 for next movie

Motion Energy Analysis (MEA)

12 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions: AUTO
Create a folder on your desktop with this exact! name:
MEA_output
move the program (MEA) directly into your video-folder!
(the program operates from any destination. For serial
processing, it has to reside inside the very same folder
that you are analyzing.
(6.b) switch ON serial-processing

drop folder of your videos on blue area

you will see all available videos and their count in the gray
areas
(6.c) click on red “START” circle
files will now be automatically saved to the source-folder
the extension _MEA.txt will be added
The programm will continue to execute this loop until
you stop it. When you see that the last movie has been
analyzed, click “stop”.
Motion Energy Analysis (MEA)
13 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 MEA Instructions: AUTO
serial processing is in beta-mode !

please be extra-careful to check whether your data do

correspond with your videos.
You may experience problems with the first and last few
videos in your folder – please double-check.

Make sure to delete/sort out the duplicate files

generated once all files have been run through, because
the program remains in “loop-mode”, i.e. keeps running
the last video as long until it is interrupted by the manual
“STOP” issued by you.

Motion Energy Analysis (MEA)

14 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 integrity of raw-data
(I) checking for other anomalies in the data:
some movie-codecs (e.g. Windows media) induce key-
frames or other anomalies. The suitability of a recording
(or a digitized video) depends on many factors.
Generally speaking, the Macintosh platform performs far
better with MEA.
Codecs from the Quicktime family should be chosen
NO keyframes should be used when digitizing or
transcoding material

good CODECs:
h.264, Apple Quicktime
example of
keyframes:
regular peaks

example of good raw data and good threshold-setting

Motion Energy Analysis (MEA)

15 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch
 frame-drops (an issue on older versions < 3.0)
(II) depending on your local configuration, your computer
may be too slow to accurately assess all video-frames.
if your computer cannot cope with the data from your
original video, then MEA will “lose” some of the frames.
This data-loss shows up as a characteristic “frame-drops”.
frame-drops are data-points where the program
(erroneously) writes zero movement (0) instead of the
correct value.
fortunately, this behavior can be easily detected in the
resulting raw-data: plotting the raw-data reveals the frame-
drop pattern:

if you detect frame-drops in your raw-data, please:

REDUCE video-size (to 320x240), frame-rate (to 10)
USE desktop computer, macintosh-version, h264 codec
SWITCH OFF other programs, virus-scanners, etc.
KEEP video on fast (internal) harddrive;
for external HDs, use firewire/thunderbolt instead of USB

Motion Energy Analysis (MEA)

16 Fabian Ramseyer | University of Bern, Switzerland | ramseyer@psync.ch