Hi Pierre,

Please note that the head position estimation has been enhanced in 
version 2.1.
See the User's Guide section 4.4. and Appendix D for details.

In 2.0, option hpistep did set the head position estimation interval and 
cHPI amplitudes
were estimated in successive (non-overlapping) 1-sec windows.

Version 2.1 extracts the cHPI amplitudes 100 times per second (as default)
by sliding a 200 ms window so that it partly overlaps with the previous one.
Now hpistep sets the sliding interval, as default it is 10 ms. 
Thereafter, the program
estimates from these cHPI amplitudes when a new head position is needed.
Thus, setting hpistep 200 ms sets successive amplitude estimation window but
may decrease the head position estimation accuracy.

Version 2.1 subtracts automatically the cHPI signals, and thus option 
'hpisubt'
was removed. However, the program also subtracts the line frequency 
terms and
thus affects as a notch filter (for 50,100,150,200,250 etc Hz).

Your options are however formally correct, except that I would remove 
'hpistep'.
Thus, difficult to tell what causes the crash. Please check:
1) Type /neuro/bin/util/show_fiff -v <file.fif>  and send me the output.
2) What was the sampling frequency?
3) Was continuous HPI was switched off after 652 sec?
4) Please try to redo the processing without 'hpistep' and without 'ds'.


I may need to get your files here to analyze the cause of crash in more 
detail.

With best regards,
Jukka Nenonen


Pierre Gagnepain wrote:
> Dear all
> 
> Apologize for the rather long email. Please found bellow description of 
> a bug that I found on my data using Max filter 2.1.
> 
> I first run autobad and it works fine:
> /neuro/bin/util/maxfilter-2.1 -f /SubjectDirectory/pd_1_raw.fif -o 
> pd_1_raw_sss.fif -v | tee pd_1_raw_autobad.log
> 
> Then I try just SSSt and movement detection (every 1000ms; NB no 
> movement correction):
> 
> /neuro/bin/util/maxfilter-2.1 -f /SubjectDirectory/pd_1_raw.fif -o 
> pd_1_raw_ssst.fif -ctc /neuro/databases/ctc/ct_sparse.fif -cal 
> /neuro/databases/sss/sss_cal.dat -frame head -origin 0 0 40 -format 
> float -autobad off -bad 1031 1111 -st 4 -headpos -hpistep 1000 -hp 
> pd_1_raw_mvpos.txt -ds 4 -v | tee pd_1_raw_ssst.log
> 
> Though I have 3 coils accepted out of 5, the program seems again to 
> work, a *.fif is output (though I note that I had to remove the option 
> "-hpisubt amp", which doesn't seem to be present in maxfilter 2.1?)
> 
> HOWEVER, when I try the critical movement compensation (every 200ms):
> 
> /neuro/bin/util/maxfilter-2.1 -f /SubjectDirectory/pd_1_raw.fif -o 
> pd_1_raw_mvcomp.fif -ctc /neuro/databases/ctc/ct_sparse.fif -cal 
> /neuro/databases/sss/sss_cal.dat -frame head -origin 0 0 40 -format 
> float -autobad off -bad 1031 1111 -st 4 -movecomp -hpistep 200 -ds 4 -v 
> | tee pd_1_raw_movecomp.log
> 
> I got a major crash:
> 
> [...]
> --- Reading raw tag #b = 569/569 (#t = 666.000) ---
> Static bad channels (2): 1031 1111
> HPI amplitude correlation 642.000: 0.999 (275 chnls > 0.95)
> HPI amplitude correlation 643.000: 0.999 (278 chnls > 0.95)
> HPI amplitude correlation 644.000: 0.999 (257 chnls > 0.95)
> HPI amplitude correlation 645.000: 0.999 (285 chnls > 0.95)
> HPI amplitude correlation 646.000: 0.998 (284 chnls > 0.95)
> HPI amplitude correlation 647.000: 0.997 (283 chnls > 0.95)
> HPI amplitude correlation 648.000: 0.996 (281 chnls > 0.95)
> HPI amplitude correlation 649.000: 0.995 (281 chnls > 0.95)
> HPI amplitude correlation 650.000: 0.994 (280 chnls > 0.95)
> HPI amplitude correlation 651.000: 0.993 (279 chnls > 0.95)
> HPI amplitude correlation 652.000: 0.991 (277 chnls > 0.95)
> HPI amp*** glibc detected *** 
> /neuro/bin/util/x86_64-pc-linux-gnu/maxfilter-2.1: free(): invalid 
> pointer: 0x000000000c7cfdc0 ***
> ======= Backtrace: =========
> /lib64/libc.so.6[0x37c8e722ef]
> /lib64/libc.so.6(cfree+0x4b)[0x37c8e7273b]
> [...]
> Continuous HPI was active on 547/569 buffers.
> Abort
> 
> Please note that I have a similar bug with another file from the same 
> participant but this time during ssst (with the following warning 
> message: "Program died due to signal 11, all output files were deleted!").
> 
> Note also that I have again another file from the same participant, 
> where there are only 3 coils accepted out of 5, but for which all 
> functions (autobad, ssst, movecomp, trans) seem to work perfectly fine.
> 
> Last information - When I used the previous version of Max Filter on 
> those problematic files, the program finds that the fit of the 5 Hpi 
> coils is bad (all are in parentheses).
> 
> Any suggestions would be greatly appreciated.
> 
> Many thanks
> 
> Pierre
> 


-- 

================================
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