Dear Wiltrud,
What was the buffer length you used in tSSS? The large HPI fitting
errors may
be related to long buffers; the current algorithm may loose track of the
head
position if the buffer length is 8 s or more. Do you still get such
errors if you
use 4-second tSSS buffers with movement correction?
Currently, we are testing maxfilter release 2.2 where this problem, as
well as many
others, are solved. The aimed release is in September, and I will send
more detailed
information after summer holidays.
Best regards, Jukka
Wiltrud Fassbinder wrote:
> I am trying to make continuous HPI work, but I am not sure about the
> errors I am getting when running Maxfilter tsss. Right now my work flow
> is that after visually scanning the data and identifying bad channels, I
> also run Maxfilter to identify bad channels. Then I run Maxfilter with
> temporal extension on these data, and I use the bad channel utility to
> mark the bad channels. However, when I ran tsss with continuous HPI, I
> got error messages and a larger error (around 50 mm). If I just use sss
> with autobad, I don't get these errors. I looked at different runs with
> different numbers of bad channels, and it did not seem to be directly
> related to the number of bad channels. What do the errors mean? How
> should continuous HPI be integrated in the workflow?
> Wiltrud
>
> Wiltrud Fassbinder, Ph.D.
> Adjunct Associate Professor
> Communication Science & Disorders
> School of Health and Rehabilitation Sciences
> University of Pittsburgh
>
> Research Associate
> Department of Audiology and Speech Pathology
> VA Pittsburgh Healthcare System
>
>
> TSSS - bad channels provided with utility
>
> No processing records found.
> Using the default virtual transformation.
> Using the virtual transformation from default.
> Opened FIFF file fw_LD1_2.fif (169 data buffers).
> Signal Space Separation: Movement compensation for Elekta Neuromag data.
> Applying the fine-calibration from file /neuro/databases/sss/sss_cal.dat.
> Hpi coils used (4): 1 2 3 4 Total 313 channels (306 MEG coils).
> Composing 80 inside and 15 outside multipole components.
> Inside origin: (0.5 14.8 44.7 mm) in head frame.
> Using 2550 sensor points; min,max,ave distance to origin 104, 144, 119 mm.
> Using 64/80 inside and 15/15 outside harmonic components.
> CTC: CTime 52.000: HPI coils are off, data block is skipped!
> Warning: 2/4 acceptable hpi fits found, cannot determine the
> transformation! (t = 54.000)
> hannel lists of the (306 x 306) cross-talk matrix OK.
> Using the cross-talk matrix in /neuro/databases/ctc/ct_sparse.fif.
> Found 4 HPI coils: 293 307 314 321 Hz
> Line interference frequencies: 60 120 180 240 300 360 Hz
> Consistency of isotrak and HPI fit is OK.
> Storing the head positions to fw_LD1_2.fif_pos.txt
> --- Processing raw tag #b = 1/169 (#t = 52.000) ---
> Buffering raw data...
> --- Processing raw tag #b = 2/169 (#t = 53.000) ---
> Processed raw data buffer with 1000 samples.
> Skipped 1 raw data tags.
>
> --- Processing raw tag #b = 3/169 (#t = 54.000) ---
> HPI amplitude correlation 0.995 (174 chnls > 0.95)
> Static bad channels (7): 812 741 742 743 2442 1812 1842
> Using direct pseudo-inverse...
> 52.4 54.3 91.3 / 61.1 39.9 42.5 / g = 0.999 err = 51.6 mm /
> 1.000 0.000 0.000
> ( -56.2 -27.9 95.9 / -48.4 -32.2 46.7 / g = 0.913 err = 50.0 mm )
> 0.000 1.000 0.000
> -48.5 58.9 93.8 / -40.9 51.4 48.8 / g = 0.999 err = 46.3 mm /
> 0.000 0.000 1.000
> ( 58.4 -39.0 68.7 / 57.0 -52.2 19.9 / g = 0.Warning: 2/4
> acceptable hpi fits found, cannot determine the transformation! (t =
> 54.200)
> Warning: 2/4 acceptable hpi fits found, cannot determine the
> transformation! (t = 54.400)
> 948 err = 50.7 mm ) 0.0 0.0 0.0
> #t = 54.00, #e = 4.97 cm, #g = 0.96, #v = 0.00 cm/s, #r = 0.00 rad/s, #d
> = 0.00 cm
> 52.4 54.3 91.3 / 61.1 39.9 42.4 / g = 0.999 err = 51.7 mm /
> 1.000 0.000 0.000
> ( -56.2 -27.9 95.9 / -48.5 -32.2 46.7 / g = 0.912 err = 49.9 mm )
> 0.000 1.000 0.000
> -48.5 58.9 93.8 / -40.6 51.2 48.9 / g = 0.999 err = 46.2 mm /
> 0.000 0.000 1.000
> ( 58.4 -39.0 68.7 / 57.0 -52.2 19.8 / g = 0.949 err = 50.7 mm
> ) 0.0 0.0 0.0
> #t = 54.20, #e = 4.96 cm, #g = 0.96, #v = 0.00 cm/s, #r = 0.00 rad/s, #d
> = 0.00 cm
> 52.4 54.3 91.3 / 61.1 39.9 42.4 / g = 0.999 err = 51.8 mm /
> 1.000 0.000 0.000
> ( -56.2 -27.9 95.9 / -48.4 -32.2 46.7 / g = 0.913 err = 50.0 mm )
> 0.000 1.000 0.000
> -48.5 58.9 93.8 / -40.7 51.3 48.8 / g = 0.999 err = 46.3 mm /
> 0.000 0.000 1.000
> ( 58.4 -39.0 68.7 / 57.0 -52.2 19.8 / g = 0.948 err = 50.7 mm
> ) 0.0 0.0 0.0
> #t = 54.40, #e = 4.97 cm, #g = 0.96, #v = 0.00 cm/s, #r = 0.00 rad/s, #d
> = 0.00 cmWarning: 2/4 acceptable hpi fits found, cannot determine the
> transformation! (t = 54.600)
> Warning: 2/4 acceptable hpi fits found, cannot determine the
> transformation! (t = 54.800)
> Could not find optimal quaternion rotation.
>
>
> [...]
> Skipped 1 raw data tags.
>
> and so on
>
>
> SSS with AUTOBAD:
>
> No processing records found.
> Using the default virtual transformation.
> Using the virtual transformation from default.
> Opened FIFF file fw_LD1.fif (169 data buffers).
> Signal Space Separation: Movement compensation for Elekta Neuromag data.
> Applying the fine-calibration from file /neuro/databases/sss/sss_cal.dat.
> Hpi coils used (4): 1 2 3 4 Total 313 channels (306 MEG coils).
> Composing 80 inside and 15 outside multipole components.
> Inside origin: (0.5 14.8 44.7 mm) in head frame.
> Using 2550 sensor points; min,max,ave distance to origin 104, 144, 119 mm.
> Using 64/80 inside and 15/15 outside harmonic components.
> CTC: ChaTime 52.000: HPI coils are off, data block is skipped!
> nnel lists of the (306 x 306) cross-talk matrix OK.
> Using the cross-talk matrix in /neuro/databases/ctc/ct_sparse.fif.
> Found 4 HPI coils: 293 307 314 321 Hz
> Line interference frequencies: 60 120 180 240 300 360 Hz
> Consistency of isotrak and HPI fit is OK.
> Storing the head positions to fw_LD1.fif_pos.txt
> --- Processing raw tag #b = 1/169 (#t = 52.000) ---
> Buffering raw data...
> --- Processing raw tag #b = 2/169 (#t = 53.000) ---
> Processed raw data buffer with 1000 samples.
> Skipped 1 raw data tags.
>
> --- Processing raw tag #b = 3/169 (#t = 54.000) ---
> HPI amplitude correlation 0.995 (178 chnls > 0.95)
> All channels OK.
> Using direct pseudo-inverse...
> Hpi fit OK, movements [mm/s] = 7.5 / 6.5 / --- / 3.0 /
> 52.4 54.3 91.3 / 52.6 54.0 91.2 / g = 0.999 err = 0.4 mm /
> 0.997 -0.069 -0.032
> -56.2 -27.9 95.9 / -56.3 -28.0 95.9 / g = 1.000 err = 0.2 mm /
> 0.069 0.997 -0.016
> ( -48.5 58.9 93.8 / -50.1 58.2 94.2 / g = 0.999 err = 1.8 mm )
> 0.033 0.014 0.999
> 58.4 -39.0 68.7 / 58.2 -38.5 68.9 / g = 0.999 err = 0.5 mm /
> -4.2 10.7 46.2
> #t = 54.00, #e = 0.04 cm, #g = 1.00, #v = 0.09 cm/s, #r = 0.00 rad/s, #d
> = 0.00 cm
> Hpi fit OK, movements [mm/s] = 0.3 / 0.3 / --- / 0.2 /
> 52.4 54.3 91.3 / 52.6 54.0 91.2 / g = 0.999 err = 0.4 mm /
> 0.997 -0.069 -0.033
> -56.2 -27.9 95.9 / -56.4 -28.0 95.9 / g = 1.000 err = 0.2 mm /
> 0.069 0.998 -0.017
> ( -48.5 58.9 93.8 / -49.8 58.2 94.3 / g = 0.999 err = 1.5 mm )
> 0.034 0.014 0.999
> 58.4 -39.0 68.7 / 58.2 -38.5 68.9 / g = 0.999 err = 0.5 mm /
> -4.2 10.7 46.2
> #t = 54.20, #e = 0.04 cm, #g = 1.00, #v = 0.21 cm/s, #r = 0.00 rad/s, #d
> = 0.01 cm
> Hpi fit OK, movements [mm/s] = 0.4 / 0.2 / --- / 0.1 /
> 52.4 54.3 91.3 / 52.6 54.0 91.2 / g = 0.999 err = 0.4 mm /
> 0.997 -0.069 -0.032
> -56.2 -27.9 95.9 / -56.4 -28.0 95.9 / g = 1.000 err = 0.2 mm /
> 0.069 0.997 -0.017
> ( -48.5 58.9 93.8 / -49.9 58.2 94.3 / g = 0.999 err = 1.6 mm )
> 0.034 0.015 0.999
> 58.4 -39.0 68.7 / 58.2 -38.5 68.9 / g = 0.999 err = 0.5 mm /
> -4.2 10.7 46.3
> #t = 54.40, #e = 0.04 cm, #g = 1.00, #v = 0.02 cm/s, #r = 0.00 rad/s, #d
> = 0.01 cm
> Hpi fit OK, movements [mm/s] = 0.2 / 0.2 / --- / 0.1 /
> 52.4 54.3 91.3 / 52.6 54.0 91.2 / g = 0.999 err = 0.4 mm /
> 0.997 -0.069 -0.032
> -56.2 -27.9 95.9 / -56.3 -28.0 95.9 / g = 1.000 err = 0.2 mm /
> 0.069 0.997 -0.018
> ( -48.5 58.9 93.8 / -49.9 58.2 94.3 / g = 0.999 err = 1.7 mm )
> 0.033 0.015 0.999
> 58.4 -39.0 68.7 / 58.2 -38.5 68.9 / g = 0.999 err = 0.5 mm /
> -4.2 10.7 46.3
> #t = 54.60, #e = 0.04 cm, #g = 1.00, #v = 0.02 cm/s, #r = 0.00 rad/s, #d
> = 0.01 cm
> Hpi fit OK, movements [mm/s] = 0.1 / 0.2 / --- / 0.1 /
> 52.4 54.3 91.3 / 52.6 54.0 91.2 / g = 0.999 err = 0.4 mm /
> 0.997 -0.069 -0.033
> -56.2 -27.9 95.9 / -56.4 -28.0 95.9 / g = 1.000 err = 0.2 mm /
> 0.068 0.998 -0.017
> ( -48.5 58.9 93.8 / -49.7 58.1 94.3 / g = 0.999 err = 1.5 mm )
> 0.034 0.015 0.999
> 58.4 -39.0 68.7 / 58.2 -38.6 68.9 / g = 0.999 err = 0.5 mm /
> -4.2 10.7 46.3
> #t = 54.80, #e = 0.04 cm, #g = 1.00, #v = 0.02 cm/s, #r = 0.00 rad/s, #d
> = 0.01 cm
> Processed raw data buffer with 1000 samples.
--
================================
Dr. Jukka Nenonen
Manager, Method development
Elekta Oy
Street address: Siltasaarenkatu 18-20A, Helsinki, Finland
Mailing address: P.O. Box 34, FIN-00531 HELSINKI, Finland
Tel: +358 9 756 240 85 (office), +358 400 249 557 (mobile),
+358 9 756 240 11 (fax), +358 9 756 2400 (operator)
E-mail: [log in to unmask]
http://www.elekta.com/healthcare_international_elekta_neuromag.php
|