
SPECIFICATION OF HEADERINFORMATION OF EEG-Files
-----------------------------------------------


$Id$
Copyright (C) Alois Schloegl 2003-2005,2008,2009 <a.schloegl@ieee.org>
This is part of the BIOSIG-toolbox http://biosig.sf.net/


CONTENT:
=========
(A)  Basic header definition (common for all file formats)
(B)  Header definition for data acquisition
(C)  Header definition of EDF/BDF/GDF/SDF-Files 



(A)========== Basic header definition ================================

Name_of_field		type		description
-------------------------------------------------------------------------
HDR.TYPE		string		type of data format,  
HDR.VERSION		string		[OPTIONAL] depends on data format 

HDR.FileName				full filename
HDR.FILE	# file handling information 
    FILE.Path [optional]		path 
    FILE.Name [optional]		filename
    FILE.Ext  [optional]		extension
    FILE.OPEN				status: 0 (close), 1 (read), 2(write),3(rewrite header when SCLOSE)
    FILE.POS				position of the file handle (units depend on data format)
    FILE.PERMISSION			'r': read; 'w': write; 'z': on-the-fly (de-)compression 
    FILE.stdout	[optional]		output stream
    FILE.stderr	[optional]		error stream 
    FILE.size	[optional]

HDR.T0			float[1..6]	start time, [yyyy mm dd hh MM ss.cc] see HELP CLOCK  
HDR.NS			integer		number of channels
HDR.SampleRate		integer		sampling frequency in [Hz]
HDR.NRec		integer		number of records or blocks; 1 for continous data 
HDR.SPR			integer		samples per record  
HDR.Dur			float		Duration=SPR/SampleRate (in [s]) of data block [depreciated]

HDR.Calib		 		Matrix for all kind of rereferencing, depends on scaling factors as well as input argument
HDR.InChanSelect			input channel selection, # read from Raw-data 

HDR.EVENT 	# two versions are possible. Version 1:  
    EVENT.SampleRate	float		how many units [sample] give one second
    EVENT.POS		uint32		Position of event [in samples]
    EVENT.TYP		uint16		Type of event/marker/annotation according to table biosig/t200/eventcodes.txt 
  		# or Version 3: 
    EVENT.SampleRate	float		how many units [sample] give one second
    EVENT.POS		uint32		Position of event [in samples]
    EVENT.TYP		uint16		Type of event/marker/annotation according to table biosig/t200/eventcodes.txt 
    EVENT.CHN		uint16		associated channel; is zero if event applies to no specific or all channels
    EVENT.DUR		uint32		event duration in samples
    EVENT.VAL		(*)		use for sparse (non-equidistant) sample values). The type corresponds to GDFTYP of the respective channel. 
                                        and must not exceed 32 bits (4 bytes). Currently, only integer types are supported. 
		# [OPTIONAL]
    EVENT.CodeDesc	text		description of (user-specified) Eventtypes
    EVENT.Desc		text		description of Event [depreciated, will become obsolete]
    EVENT.CodeIndex	int		corresponding index [depreciated, will become obsolete]

HDR.ELEC.
    ELEC.XYZ		float 		electrode positions

HDR.Label		char-array	e.g. '+C3a - C3p  '
HDR.LeadIdCode          uint16          Lead identification label according to SCP or FEF standard
HDR.PhysDim		string		physical dimension e.g. 'uV'
HDR.PhysDimCode		uint16		Code of physical dimension according to "File Exchange Format for Vital Signs - Annex A "
HDR.PhysMax		float		physical maximum (for calibration)
HDR.DigMax		integer		digital maximum (for calibration)
HDR.PhysMin		float		physical minimum (for calibration) 
HDR.DigMin		integer		digital minimum (for calibration) 
HDR.Bits		integer		number of bits used
HDR.THRESHOLD		numeric		digital minimum and maximum (for saturation detection);
					a third column indicates the encoding of missing data;  
HDR.CHANTYP		ascii		guess for signal type 
HDR.Impedance	        float32	        Electrode Impedance for Voltage data channels  
HDR.fZ                	float32	        Probe frequency of impedance channels  

HDR.FLAG.TRIGGERED	int		0=no, 1=yes
HDR.FLAG.REFERENCE	string          COM, CAR: common average reference; LOC,LAR local average ref; LAP Laplacian derivation, WGT weighted average
HDR.FLAG.UCAL		int		0: data is calibrated (scaled), 1: data is not calibrated (scaled)
HDR.FLAG.OVERFLOWDETECTION 	int	1: automated overflow detection if HDR.THRESHOLD is available; 0: no overflow detection 
HDR.FLAG.OUTPUT		ascii		'double': data is a double; 'single': data is single precision 	
	
HDR.Classlabel		int		0: left, 1: right, etc.
HDR.ArtifactSelection	int		[OPTIONAL] 0: ok, 1: artifact 
HDR.TriggerOffset       samples         [OPTIONAL] Offset time between TriggerOnset and begin of trial

HDR.Filter	% filter information, usually for each channel separatily
    Filter.LowPass	float		[Hz]
    Filter.HighPass	float		[Hz]
    Filter.Notch 	float		[Hz] NaN: unknown, -1(<0): off
HDR.PreFilt		string		[OPTIONAL] filter setting - used by GDF, BDF, EDF

HDR.Patient	# contains patient information [OPTIONAL]
    Patient.Name			Name of Patient: should not be used because of privacy protection
    Patient.Id				Patient identification code as used in hospital
    Patient.Age				Age of Patient
    Patient.Sex				Patient Gender  0: unknown, 1: male, 2: female
    Patient.Handedness			Patient Handedness  0:unknown, 1: right, 2:left, 3: equal
    Patient.Medication			Medication
    Patient.Classification		Classification of Patient
    Patient.Height			height in centimeters [cm]
    Patient.Weight			weight in kilogram [kg]
    Patient.BMI				Body mass index
    Patient.Headsize	3xfloat32	Distance Nasion-Inion, distance A1-A2, Circumference of head
    Patient.Birthday			[Y M D h m s]: approx. Birthday, to calculate patient's age. 
    Patient.Impairment.Heart		0: unknown 1: NO 2: YES 3: pacemaker 
    Patient.Impairment.Visual		0: unknown 1: NO 2: YES 3: corrected (with visual aid) 
    Patient.Smoking			0: unknown 1: NO 2: YES 
    Patient.AlcoholAbuse		0: unknown 1: NO 2: YES 
    Patient.DrugAbuse			0: unknown 1: NO 2: YES
     
HDR.Manufacturer	# Manufacturer information			
    Manufacturer.Name			name of manufacture
    Manufacturer.Model			device model 
    Manufacturer.Version		version of device
    Manufacturer.SerialNumber    	Serial number 
    
HDR.REC.Hospital			Identification of hospital
HDR.RID					Identification of Recording, StudyID, consecutive number
HDR.REC.Technician			identity of recording technician
# HDR.REC.Doctor			Identification of refering doctor
HDR.REC.Equipment			recording equipment
HDR.REC.LOC		16 byte		Latitude, Longitude and Altitude 3 x INT32 according to RFC1876
HDR.REC.IPaddr		4(16)xuint8	IP address of recording system
# HDR.REC.Reference			reference (monopolar | CAR | LAR | Laplacian | Linked ears | weighted )
HDR.REC.Impedance	NS*float32	[OBSOLETE] replaced with HDR.Impedance  
HDR.ELEC.XYZ		3xNSxfloat32	sensor position in cartesion coordinates
HDR.ELEC.Orientation	3xNSxfloat32	sensor orientation in cartesion coordinates
HDR.ELEC.Area		NSxfloat32	area [m^2] of (MEG) sensor
HDR.ELEC.REF		3xfloat32	position of reference electrode 
HDR.ELEC.GND		3xfloat32	position of ground electrode 

HDR.AS		# provides some internal information	
    AS.SPR				samples per record for each channel
    AS.bpb				bytes per block
    AS.spb				samples per block
    AS.bi				cumulative index
    AS.SampleRate                       samplerate of each channel 
    AS.EVENTTABLEPOS                    position of eventtable   


(B)======= FOR DATA ACQUISITION (T100) ONLY  =======================
[OBSOLETE - MIGHT BE REMOVED IN FUTURE]

Name_of_field		type		example
-------------------------------------------------------------------------
HDR.T100.Version	string		'BCI 0.10' testing state of RS's data aquisition module 
					'BCI 1.00' testing state of RS's data aquisition module 
HDR.T100.AmpSens	float		Sensitivity of g.Tec-Amplifier 
HDR.T100.Range		float		Maximum physical value in DAQ
HDR.T100.Bits		integer		Number of bits in ADC

HDR.T100.DAQ_ID		<any>		z.B. "NI 9876"
HDR.T100.AmpName	string		z.B. "g.tec: Ser.No: 01234567"
HDR.Filter.LowPass	float		[Hz]
HDR.Filter.HighPass	float		[Hz]
HDR.Filter.Notch 	int8		0=Off, 1=On

HDR.T0			float[1..6]	[yyyy mm dd hh MM ss.cc] see HELP CLOCK (localtime)
HDR.tzmin		int16		timezone information, number of minutes east of UTC 
HDR.NS			Integer		number of channels
HDR.SampleRate		Integer		sampling frequency in [Hz]

HDR.Label		char-array	z.B. '+C3a - C3p  '
HDR.PhysDim		char-array	'uV    '
-------------------------------------------------------------------------




(C) ================== Header of EDF-files ========================

This documents is based on the EDF-header as defined in the "EDF/GDF toolbox for Matlab"                                                            
It will be used as template to define the header information based on different formats   
                                                         
=================|=====|===============================|=======================================
 field name       level  type and size                   text	
=================|=1234|===============================|=======================================
EDF.
   .FILE.
      .FID	     *	file identifier		
 	.OPEN		2 | 1 | 0    			set to 1 by EDFOPEN and set to 0 by EDFCLOSE
 	.Ext		char(n)				file extension, usually 'edf' or 'rec'
 	.Name         #	char(n)				file name
 	.Path        	char(n)				path 
 	.POS	     *	int					Position in EDF-Blocks
    .FileName           char(n)				full file name, including path,filename and extension
    .VERSION	     *	char(8) 				EDF field: Version
    .PID              # char(n)			     	Patient ID: ##OBSOLETE## this field will be replaced with HDR.Patient.Id	
    .RID              # char(n)                       	Recording ID: ##OBSOLETE## this field will be replaced with HDR.ID.Recording 
    .T0                 int(6) [yyyy mm dd hh MM ss]	startdatetime of recording
    .HeadLen         *  int				Header Length
    .NRec            *  int                     	number of records
    .Dur             *  float                       	Duration of record
    .NS		     *# int                      	number of channels                               
    .Label	      #	char array [NS,:]         	Label
    .Transducer       #	char array [NS,80]        	Transducer
    .PhysDim          #	char array [NS,:]         	physical dimension
    .PhysMin         *  float array [NS,1]        	physical minimum
    .PhysMax         *  float array [NS,1]        	physical maximum
    .DigMin          *  int array  [NS,1]         	digital minimum
    .DigMax          *  int array  [NS,1]         	digital maximum
    .PreFilt	      #	char array [NS,80]        	Pre-filtering
    .SPR	     *	int array  [NS,1]         	Samples per Record 
    .GDFTYP 	     *	int [NS]				defines datatype of the channel  
    .Cal	      #	float array [NS,1]        
    .Off              # float array [NS,1]         
    .Calib		float array [1+NS,NS]      
    .SampleRate	      #	float array [NS,1]         
    .Chan_Select        bool array [NS,1]          
    .ChanTyp            char array [1,EDF.NS]		indicates EEG, EOG, ECG, EMG, channel
HDR.ID.
	.Recording					Recording Identification 
HDR.Patient...
	.Id						Patient Identification 
	.Sex						Gender: 1: Male 2: Female 
	.Weigth						
	.Heigth
	.Birth						Birth date and time 
	.Name							
    .AS.                					undocumented
       .bpb             int
       .spb             int
       .bi              int array [1+NS,1]      
       .GDFbi           int array [1+NS,1]      
       .INFO		char array
       .Method		char array
       .Date		int(6)
=================|=====|===============================|=======================================
EDF.AS.startrec		record number at beginning of last gdfread(...)
EDF.AS.numrec		number of records loaded with last gdfread(...)

EDF.SIE	-----------	fields for Version 0.50 and higher
EDF.SIE.RS		switch for Resampling (0 = none)
EDF.SIE.RR		switch for Calibration and ReReferinzing (0 = none)
EDF.SIE.TH		Overflow detection (1=yes)
EDF.SIE.RAW		Raw EDF format (1, each column contains one block)
EDF.SIE.AFIR		apply AFIR (1=yes)

EDF.SIE.InChanSelect	input channel selection, # read from Raw-EDF input file
EDF.SIE.ChanSelect	(output) channel selection, 

EDF.SIE.ReRefMx		ReReferenzing Matrix
EDF.SIE.REG		Regression Matrix


EDF.AFIR	-----------	fields for Adaptive Filter for ECG minimization

=================|=====|===============================|=======================================
     
#  level 4, application  exported, always available in all future version (strict definition).
*  level 3, transport	EDF/GDF specific properties, useful for lower level EDF access,
   level 2, access	matlab specific fields, only used within the EDF toolbox, can change ithout further notices.
   level 1, low level	Binary representation as defined in the EDF specification,         
                              
Status of this document: Finished, not proofread yet.
(C) 1998 by Alois Schloegl
13.Nov.1998

This document holds at least for Version 3 of the "EDF toolbox for the use with Matlab". 
All fields are case sensitive.

"not supported" means, you can not rely on that field.
char(n) is a char array, string
The types int, float, bool, are always stored in double. 
                                                                                       
The definition holds for Version #3.00 - #3.08 (current). Level 1,2 fields might change without further notice. 
Further changes will be added at the end of this document. There should be none.


AMENDMENTS:
=) Changes due to the consideration of GDF:
The definition above holds for the "EDF/GDF toolbox for Matlab" Ver 0.40, too.
The following changes were performed. 

=================|=====|===============================|=======================================
 field name       level  type and size                   text	
=================|=1234|===============================|=======================================
~ (modified) 11:09 05.08.99 V0.56 
 	.POS	     *	int					Position in EDF-Blocks

+ (added) 11:09 12.05.99 V0.46 
EDF.AS.startrec    	int				
EDF.AS.numrec		int				

* (changed)
EDF.FILE.OPEN	 	2 | 1 | 0    			set to 0,1 and 2 by EDFCLOSE, EDFOPEN(.,'r') and EDFOPEN(.,'w'), respectively

+ (added)
EDF.GDFTYP 	     *	int [NS]			defines datatype of the channel  
EDF.AS.bpb              int				bytes per block
EDF.AS.GDFbi            int				bytes per block

- (removed)
    .sPhysMin		char array [NS,8]         	EDF field: physical minimum (8*ASCII)
    .sPhysMax  		char array [NS,8]         	EDF field: physical maximum (8*ASCII)
=================|=====|===============================|=======================================
