ecat

Read ECAT format images

An ECAT format image consists of:

  • a main header;

  • at least one matrix list (mlist);

ECAT thinks of memory locations in terms of blocks. One block is 512 bytes. Thus block 1 starts at 0 bytes, block 2 at 512 bytes, and so on.

The matrix list is an array with one row per frame in the data.

Columns in the matrix list are:

  • 0: Matrix identifier (frame number)

  • 1: matrix data start block number (subheader followed by image data)

  • 2: Last block number of matrix (image) data

  • 3: Matrix status

    • 1: hxists - rw

    • 2: exists - ro

    • 3: matrix deleted

There is one sub-header for each image frame (or matrix in the terminology above). A sub-header can also be called an image header. The sub-header is one block (512 bytes), and the frame (image) data follows.

There is very little documentation of the ECAT format, and many of the comments in this code come from a combination of trial and error and wild speculation.

XMedcon can read and write ECAT 6 format, and read ECAT 7 format: see http://xmedcon.sourceforge.net and the ECAT files in the source of XMedCon, currently libs/tpc/*ecat* and source/m-ecat*. Unfortunately XMedCon is GPL and some of the header files are adapted from CTI files (called CTI code below). It’s not clear what the licenses are for these files.

EcatHeader([binaryblock, endianness, check])

Class for basic Ecat PET header

EcatImage(dataobj, affine, header, ...[, ...])

Class returns a list of Ecat images, with one image(hdr/data) per frame

EcatImageArrayProxy(subheader)

Ecat implementation of array proxy protocol

EcatSubHeader(hdr, mlist, fileobj)

parses the subheaders in the ecat (.v) file there is one subheader for each frame in the ecat file

get_frame_order(mlist)

Returns the order of the frames stored in the file Sometimes Frames are not stored in the file in chronological order, this can be used to extract frames in correct order

get_series_framenumbers(mlist)

Returns framenumber of data as it was collected, as part of a series; not just the order of how it was stored in this or across other files

read_mlist(fileobj, endianness)

read (nframes, 4) matrix list array from fileobj

read_subheaders(fileobj, mlist, endianness)

Retrieve all subheaders and return list of subheader recarrays

EcatHeader

class nibabel.ecat.EcatHeader(binaryblock=None, endianness=None, check=True)

Bases: WrapStruct, SpatialHeader

Class for basic Ecat PET header

Sub-parts of standard Ecat File

  • main header

  • matrix list which lists the information for each frame collected (can have 1 to many frames)

  • subheaders specific to each frame with possibly-variable sized data blocks

This just reads the main Ecat Header, it does not load the data or read the mlist or any sub headers

Initialize Ecat header from bytes object

Parameters:
binaryblock{None, bytes} optional

binary block to set into header, By default, None in which case we insert default empty header block

endianness{None, ‘<’, ‘>’, other endian code}, optional

endian code of binary block, If None, guess endianness from the data

check{True, False}, optional

Whether to check and fix header for errors. No checks currently implemented, so value has no effect.

__init__(binaryblock=None, endianness=None, check=True)

Initialize Ecat header from bytes object

Parameters:
binaryblock{None, bytes} optional

binary block to set into header, By default, None in which case we insert default empty header block

endianness{None, ‘<’, ‘>’, other endian code}, optional

endian code of binary block, If None, guess endianness from the data

check{True, False}, optional

Whether to check and fix header for errors. No checks currently implemented, so value has no effect.

classmethod default_structarr(endianness=None)

Return header data for empty header with given endianness

get_data_dtype()

Get numpy dtype for data from header

get_filetype()

Type of ECAT Matrix File from code stored in header

get_patient_orient()

gets orientation of patient based on code stored in header, not always reliable

classmethod guessed_endian(hdr)

Guess endian from MAGIC NUMBER value of header data

template_dtype = dtype([('magic_number', 'S14'), ('original_filename', 'S32'), ('sw_version', '<u2'), ('system_type', '<u2'), ('file_type', '<u2'), ('serial_number', 'S10'), ('scan_start_time', '<u4'), ('isotope_name', 'S8'), ('isotope_halflife', '<f4'), ('radiopharmaceutical', 'S32'), ('gantry_tilt', '<f4'), ('gantry_rotation', '<f4'), ('bed_elevation', '<f4'), ('intrinsic_tilt', '<f4'), ('wobble_speed', '<u2'), ('transm_source_type', '<u2'), ('distance_scanned', '<f4'), ('transaxial_fov', '<f4'), ('angular_compression', '<u2'), ('coin_samp_mode', '<u2'), ('axial_samp_mode', '<u2'), ('ecat_calibration_factor', '<f4'), ('calibration_unitS', '<u2'), ('calibration_units_type', '<u2'), ('compression_code', '<u2'), ('study_type', 'S12'), ('patient_id', 'S16'), ('patient_name', 'S32'), ('patient_sex', 'S1'), ('patient_dexterity', 'S1'), ('patient_age', '<f4'), ('patient_height', '<f4'), ('patient_weight', '<f4'), ('patient_birth_date', '<u4'), ('physician_name', 'S32'), ('operator_name', 'S32'), ('study_description', 'S32'), ('acquisition_type', '<u2'), ('patient_orientation', '<u2'), ('facility_name', 'S20'), ('num_planes', '<u2'), ('num_frames', '<u2'), ('num_gates', '<u2'), ('num_bed_pos', '<u2'), ('init_bed_position', '<f4'), ('bed_position', '<f4', (15,)), ('plane_separation', '<f4'), ('lwr_sctr_thres', '<u2'), ('lwr_true_thres', '<u2'), ('upr_true_thres', '<u2'), ('user_process_code', 'S10'), ('acquisition_mode', '<u2'), ('bin_size', '<f4'), ('branching_fraction', '<f4'), ('dose_start_time', '<u4'), ('dosage', '<f4'), ('well_counter_corr_factor', '<f4'), ('data_units', 'S32'), ('septa_state', '<u2'), ('fill', 'S12')])

EcatImage

class nibabel.ecat.EcatImage(dataobj, affine, header, subheader, mlist, extra=None, file_map=None)

Bases: SpatialImage

Class returns a list of Ecat images, with one image(hdr/data) per frame

Initialize Image

The image is a combination of (array, affine matrix, header, subheader, mlist) with optional meta data in extra, and filename / file-like objects contained in the file_map.

Parameters:
dataobjarray-like

image data

affineNone or (4,4) array-like

homogeneous affine giving relationship between voxel coords and world coords.

headerNone or header instance

meta data for this image format

subheaderNone or subheader instance

meta data for each sub-image for frame in the image

mlistNone or array

Matrix list array giving offset and order of data in file

extraNone or mapping, optional

metadata associated with this image that cannot be stored in header or subheader

file_mapmapping, optional

mapping giving file information for this image format

Examples

>>> import os
>>> import nibabel as nib
>>> nibabel_dir = os.path.dirname(nib.__file__)
>>> from nibabel import ecat
>>> ecat_file = os.path.join(nibabel_dir,'tests','data','tinypet.v')
>>> img = ecat.load(ecat_file)
>>> frame0 = img.get_frame(0)
>>> frame0.shape == (10, 10, 3)
True
>>> data4d = img.get_fdata()
>>> data4d.shape == (10, 10, 3, 1)
True
__init__(dataobj, affine, header, subheader, mlist, extra=None, file_map=None)

Initialize Image

The image is a combination of (array, affine matrix, header, subheader, mlist) with optional meta data in extra, and filename / file-like objects contained in the file_map.

Parameters:
dataobjarray-like

image data

affineNone or (4,4) array-like

homogeneous affine giving relationship between voxel coords and world coords.

headerNone or header instance

meta data for this image format

subheaderNone or subheader instance

meta data for each sub-image for frame in the image

mlistNone or array

Matrix list array giving offset and order of data in file

extraNone or mapping, optional

metadata associated with this image that cannot be stored in header or subheader

file_mapmapping, optional

mapping giving file information for this image format

Examples

>>> import os
>>> import nibabel as nib
>>> nibabel_dir = os.path.dirname(nib.__file__)
>>> from nibabel import ecat
>>> ecat_file = os.path.join(nibabel_dir,'tests','data','tinypet.v')
>>> img = ecat.load(ecat_file)
>>> frame0 = img.get_frame(0)
>>> frame0.shape == (10, 10, 3)
True
>>> data4d = img.get_fdata()
>>> data4d.shape == (10, 10, 3, 1)
True
ImageArrayProxy

alias of EcatImageArrayProxy

property affine
files_types: tuple[tuple[str, str | None], ...] = (('image', '.v'), ('header', '.v'))
classmethod from_file_map(file_map, *, mmap=True, keep_file_open=None)

class method to create image from mapping specified in file_map

classmethod from_image(img)

Class method to create new instance of own class from img

Parameters:
imgspatialimage instance

In fact, an object with the API of spatialimage - specifically dataobj, affine, header and extra.

Returns:
cimgspatialimage instance

Image, of our own class

get_data_dtype(frame)
get_frame(frame, orientation=None)

Get full volume for a time frame

Parameters:
  • frame – Time frame index from where to fetch data

  • orientation – None (default), ‘neurological’ or ‘radiological’

Return type:

Numpy array containing (possibly oriented) raw data

get_frame_affine(frame)

returns 4X4 affine

get_mlist()

get access to the mlist

get_subheaders()

get access to subheaders

header_class

alias of EcatHeader

classmethod load(filespec)

Class method to create image from filename filename

Parameters:
filenamestr

Filename of image to load

mmap{True, False, ‘c’, ‘r’}, optional, keyword only

mmap controls the use of numpy memory mapping for reading image array data. If False, do not try numpy memmap for data array. If one of {‘c’, ‘r’}, try numpy memmap with mode=mmap. A mmap value of True gives the same behavior as mmap='c'. If image data file cannot be memory-mapped, ignore mmap value and read array from file.

keep_file_open{ None, True, False }, optional, keyword only

keep_file_open controls whether a new file handle is created every time the image is accessed, or a single file handle is created and used for the lifetime of this ArrayProxy. If True, a single file handle is created and used. If False, a new file handle is created every time the image is accessed. The default value (None) will result in the value of nibabel.arrayproxy.KEEP_FILE_OPEN_DEFAULT being used.

Returns:
imgDataobjImage instance
property shape
to_file_map(file_map=None)

Write ECAT7 image to file_map or contained self.file_map

The format consist of:

  • A main header (512L) with dictionary entries in the form

    [numAvail, nextDir, previousDir, numUsed]

  • For every frame (3D volume in 4D data) - A subheader (size = frame_offset) - Frame data (3D volume)

valid_exts: tuple[str, ...] = ('.v',)

EcatImageArrayProxy

class nibabel.ecat.EcatImageArrayProxy(subheader)

Bases: object

Ecat implementation of array proxy protocol

The array proxy allows us to freeze the passed fileobj and header such that it returns the expected data array.

__init__(subheader)
property is_proxy
property ndim
property shape

EcatSubHeader

class nibabel.ecat.EcatSubHeader(hdr, mlist, fileobj)

Bases: object

parses the subheaders in the ecat (.v) file there is one subheader for each frame in the ecat file

Parameters:
hdrEcatHeader

ECAT main header

mlistarray shape (N, 4)

Matrix list

fileobjECAT file <filename>.v fileholder or file object

with read, seek methods

__init__(hdr, mlist, fileobj)

parses the subheaders in the ecat (.v) file there is one subheader for each frame in the ecat file

Parameters:
hdrEcatHeader

ECAT main header

mlistarray shape (N, 4)

Matrix list

fileobjECAT file <filename>.v fileholder or file object

with read, seek methods

data_from_fileobj(frame=0, orientation=None)

Read scaled data from file for a given frame

Parameters:
  • frame – Time frame index from where to fetch data

  • orientation – None (default), ‘neurological’ or ‘radiological’

Return type:

Numpy array containing (possibly oriented) raw data

See also

raw_data_from_fileobj

get_frame_affine(frame=0)

returns best affine for given frame of data

get_nframes()

returns number of frames

get_shape(frame=0)

returns shape of given frame

get_zooms(frame=0)

returns zooms …pixdims

raw_data_from_fileobj(frame=0, orientation=None)

Get raw data from file object.

Parameters:
  • frame – Time frame index from where to fetch data

  • orientation – None (default), ‘neurological’ or ‘radiological’

Return type:

Numpy array containing (possibly oriented) raw data

See also

data_from_fileobj

get_frame_order

nibabel.ecat.get_frame_order(mlist)

Returns the order of the frames stored in the file Sometimes Frames are not stored in the file in chronological order, this can be used to extract frames in correct order

Returns:
id_dict: dict mapping frame number -> [mlist_row, mlist_id]
(where mlist id is value in the first column of the mlist matrix )

Examples

>>> import os
>>> import nibabel as nib
>>> nibabel_dir = os.path.dirname(nib.__file__)
>>> from nibabel import ecat
>>> ecat_file = os.path.join(nibabel_dir,'tests','data','tinypet.v')
>>> img = ecat.load(ecat_file)
>>> mlist = img.get_mlist()
>>> get_frame_order(mlist)
{0: [0, 16842758]}

get_series_framenumbers

nibabel.ecat.get_series_framenumbers(mlist)

Returns framenumber of data as it was collected, as part of a series; not just the order of how it was stored in this or across other files

For example, if the data is split between multiple files this should give you the true location of this frame as collected in the series (Frames are numbered starting at ONE (1) not Zero)

Returns:
frame_dict: dict mapping order_stored -> frame in series

where frame in series counts from 1; [1,2,3,4…]

Examples

>>> import os
>>> import nibabel as nib
>>> nibabel_dir = os.path.dirname(nib.__file__)
>>> from nibabel import ecat
>>> ecat_file = os.path.join(nibabel_dir,'tests','data','tinypet.v')
>>> img = ecat.load(ecat_file)
>>> mlist = img.get_mlist()
>>> get_series_framenumbers(mlist)
{0: 1}

read_mlist

nibabel.ecat.read_mlist(fileobj, endianness)

read (nframes, 4) matrix list array from fileobj

Parameters:
fileobjfile-like

an open file-like object implementing seek and read

Returns:
mlist(nframes, 4) ndarray

matrix list is an array with nframes rows and columns:

  • 0: Matrix identifier (frame number)

  • 1: matrix data start block number (subheader followed by image data)

  • 2: Last block number of matrix (image) data

  • 3: Matrix status

    • 1: hxists - rw

    • 2: exists - ro

    • 3: matrix deleted

Notes

A block is 512 bytes.

block_no in the code below is 1-based. block 1 is the main header, and the mlist blocks start at block number 2.

The 512 bytes in an mlist block contain 32 rows of the int32 (nframes, 4) mlist matrix.

The first row of these 32 looks like a special row. The 4 values appear to be (respectively):

  • not sure - maybe negative number of mlist rows (out of 31) that are blank and not used in this block. Called nfree but unused in CTI code;

  • block_no - of next set of mlist entries or 2 if no more entries. We also allow 1 or 0 to signal no more entries;

  • <no idea>. Called prvblk in CTI code, so maybe previous block no;

  • n_rows - number of mlist rows in this block (between ?0 and 31) (called nused in CTI code).

read_subheaders

nibabel.ecat.read_subheaders(fileobj, mlist, endianness)

Retrieve all subheaders and return list of subheader recarrays

Parameters:
fileobjfile-like

implementing read and seek

mlist(nframes, 4) ndarray

Columns are: * 0 - Matrix identifier. * 1 - subheader block number * 2 - Last block number of matrix data block. * 3 - Matrix status

endianness{‘<’, ‘>’}

little / big endian code

Returns:
subheaderslist

List of subheader structured arrays