def process_mrs_ge(pfile_dir):
# Make a top-level instance and hide in top left corner, get filepath
# root = Tkinter.Tk()
# root.geometry('0x0+0+0')
# root.attributes("-topmost", 1)
# pfile_dir = tkFileDialog.askdirectory(parent=root)
# root.attributes("-topmost", 0)
# root.destroy()
print('Working...')
root_dir = os.path.abspath(os.path.join(pfile_dir, os.pardir))
pat_name = os.path.split(os.path.dirname(pfile_dir))[1]
filelist = glob.glob(pfile_dir + '/*.7')
filelist.extend(glob.glob(pfile_dir + '/*.anon'))
for fname in filelist:
try:
print('')
print(fname)
fbin = open(fname, 'rb')
sw_ver = int(
str(struct.unpack('f', fbin.read(4))[0]).split('.')[0])
print('software version: ' + str(sw_ver))
s = fbin.read()
if s.find('PROBE-P') > 0:
seqtype = 'PRESS'
elif s.find('PROBE-S') > 0:
seqtype = 'STEAM'
else:
print('WARNING: acquisition type not recognised!')
if sw_ver < 26:
fbin.seek(82)
pt_sz = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(1468)
hdr_sz = struct.unpack('<i', fbin.read(4))[0]
if sw_ver == 15:
fbin.seek(144580) # FOR REV 15
elif sw_ver == 16:
fbin.seek(144628) # FOR REV 16
elif sw_ver > 19:
fbin.seek(148404) # FOR REV 20 & 24
te = struct.unpack('<i', fbin.read(4))[0]
if sw_ver == 15:
fbin.seek(144572) # FOR REV 15
elif sw_ver == 16:
fbin.seek(144620) # FOR REV 16
elif sw_ver > 19 and sw_ver < 26:
fbin.seek(148396) # FOR REV 20 & 24
tr = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(148712)
exam_no = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(148724)
series_no = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(424)
c_freq = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(200)
st_rcv = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(202)
en_rcv = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(70)
nechoes = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(72)
navs = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(74)
nframes = struct.unpack('<h', fbin.read(2))[0]
n_rcv = (en_rcv - st_rcv) + 1
# pfile=headers.Pfile.from_file(fname)
# frm_sz=pfile.acq_x_res*2*pt_sz
# echo_size=frm_sz*pfile.acq_y_Res
# slice_size=echo_size*nechoes
# data_sz=pfile.acq_x_res*2*(pfile.acq_y_Res-1)
fbin.seek(102)
acq_x_res = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(104)
acq_y_Res = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(216)
bandwidth = struct.unpack('<f', fbin.read(4))[0]
# if sw_ver == 15:
# fbin.seek(948) # FOR REV 15
# elif sw_ver == 16:
# fbin.seek(948) # FOR REV 16
# elif sw_ver > 19:
# fbin.seek(948) # FOR REV 20 & 24
fbin.seek(232)
ws_nsa = struct.unpack('<f', fbin.read(4))[0]
elif sw_ver == 26:
fbin.seek(158)
pt_sz = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(4)
hdr_sz = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(199244) # FOR REV 26
te = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(199236) # FOR REV 26
tr = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(196428)
exam_no = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(194068)
series_no = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(504)
c_freq = struct.unpack('<i', fbin.read(4))[0]
fbin.seek(264)
st_rcv = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(266)
en_rcv = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(146)
nechoes = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(148)
navs = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(2520)
nframes = struct.unpack('<h', fbin.read(2))[0]
n_rcv = (en_rcv - st_rcv) + 1
# pfile=headers.Pfile.from_file(fname)
# frm_sz=pfile.acq_x_res*2*pt_sz
# echo_size=frm_sz*pfile.acq_y_Res
# slice_size=echo_size*nechoes
# data_sz=pfile.acq_x_res*2*(pfile.acq_y_Res-1)
fbin.seek(156)
acq_x_res = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(180)
acq_y_Res = struct.unpack('<h', fbin.read(2))[0]
fbin.seek(280)
bandwidth = struct.unpack('<f', fbin.read(4))[0]
fbin.seek(296)
ws_nsa = struct.unpack('<f', fbin.read(4))[0]
ws_avgs = int(ws_nsa) / int(navs)
frm_sz = acq_x_res * 2 * pt_sz
echo_size = frm_sz * acq_y_Res
slice_size = echo_size * nechoes
data_sz = acq_x_res * 2 * (acq_y_Res - 1)
#off_raw is the offset in bytes to the start of the raw data
off_raw = hdr_sz + frm_sz
w_avgs = acq_y_Res - 1 - ws_avgs
print('Number of water ref frames : ' + str(w_avgs))
print('Number of data frames : ' + str(ws_avgs))
fbin.seek(off_raw)
raw_data = np.zeros(data_sz)
frame_data = np.zeros(shape=(int(acq_y_Res - 1), int(acq_x_res),
int(n_rcv)),
dtype=np.int64)
frame_data = frame_data.view(np.complex64)
rot_frame_data = np.zeros(shape=(int(acq_y_Res - 1),
int(acq_x_res), int(n_rcv)),
dtype=np.float64)
rot_frame_data = rot_frame_data.view(np.complex64)
fid_norm = np.zeros(shape=(int(acq_y_Res - 1), int(acq_x_res),
int(n_rcv)),
dtype=np.float64)
fid_norm = fid_norm.view(np.complex64)
summed_signal = np.zeros(shape=(int(acq_y_Res - 1) *
int(acq_x_res)),
dtype=np.float64)
summed_signal = summed_signal.view(np.complex64)
for receiver in range(0, n_rcv):
fbin.seek(off_raw + (slice_size * receiver))
for element in range(0, data_sz):
raw_data[element] = struct.unpack('<i', fbin.read(4))[0]
for frm in range(0, int(acq_y_Res - 1)):
row_off = frm * acq_x_res * 2
for pt in range(0, int(acq_x_res)):
frame_data[frm, pt, receiver] = raw_data[
(2 * pt) +
row_off] + 1j * raw_data[(2 * pt) + 1 + row_off]
fbin.close()
amp = np.zeros(n_rcv, dtype=np.int64)
amp = amp.view(np.complex64)
amp = frame_data[0, 5, :]
ang = np.zeros(n_rcv)
fid_w = np.zeros(n_rcv)
for n in range(0, n_rcv):
ang[n] = cmath.phase(list(frame_data[0, 5, :])[n])
# print(ang)
ph_dif = ang - ang[0]
for receiver in range(0, n_rcv):
rot_frame_data[:, :,
receiver] = frame_data[:, :, receiver] * np.exp(
-1j * ph_dif[receiver])
for receiver in range(0, n_rcv):
fid_w[receiver] = abs(amp[receiver]) / np.sqrt(
sum(abs(amp * amp)))
# fid_w[receiver]=abs(amp[receiver])/sum(abs(amp))
# fid_w=np.max(np.abs(frame_data[1,:,:]),axis=0)/np.sqrt(np.sum(np.max(np.abs(frame_data[1,:,:]),axis=0)*np.max(np.abs(frame_data[1,:,:]),axis=0)))
# fid_w=np.abs(np.max(frame_data[0,:,:],axis=0))/(np.sum(np.abs(np.max(frame_data[0,:,:],axis=0))))
fid_norm[:, :,
receiver] = rot_frame_data[:, :,
receiver] * fid_w[receiver]
summed_signal = np.sum(fid_norm, axis=2)
#fbin.close()
#raw_data=raw_data.astype(np.int64)
print('coil weightings:')
print(fid_w)
output_data_ws = []
for pt in range(w_avgs * int(acq_x_res), np.size(summed_signal)):
output_data_ws.append(
np.float64(np.imag(summed_signal.flatten()[pt])))
output_data_ws.append(
np.float64(np.real(summed_signal.flatten()[pt])))
output_data_w = []
for pt in range(0, w_avgs * int(acq_x_res)):
output_data_w.append(
np.float64(np.imag(summed_signal.flatten()[pt])))
output_data_w.append(
np.float64(np.real(summed_signal.flatten()[pt])))
file_meta = Dataset()
file_meta.MediaStorageSOPClassUID = 'MRSpectroscopyStorage'
file_meta.MediaStorageSOPInstanceUID = dicom.UID.generate_uid()
file_meta.ImplementationClassUID = dicom.UID.generate_uid()
ds_ws = FileDataset(fname + '_dcm', {},
file_meta=file_meta,
preamble="\0" * 128)
ds_ws.Modality = 'MR'
ds_ws.ContentDate = str(datetime.date.today()).replace('-', '')
ds_ws.ContentTime = str(time.time()) #milliseconds since the epoch
ds_ws.StudyInstanceUID = dicom.UID.generate_uid()
ds_ws.SeriesInstanceUID = dicom.UID.generate_uid()
ds_ws.SOPInstanceUID = dicom.UID.generate_uid()
ds_ws.SOPClassUID = 'MRSpectroscopyStorage'
ds_ws.SecondaryCaptureDeviceManufctur = 'Python 2.7.3'
ds_ws.VolumeLocalizationTechnique = seqtype
ds_ws.RepetitionTime = tr / 1000
## These are the necessary imaging components of the FileDataset object.
ds_ws.Columns = 1
ds_ws.Rows = 1
ds_ws.DataPointColumns = int(acq_x_res)
ds_ws.SpectroscopyAcquisitionDataColumns = ds_ws.DataPointColumns
ds_ws.DataPointRows = ws_avgs
ds_ws.SpectroscopyData = output_data_ws
ds_ws.ImageType = ['ORIGINAL', 'PRIMARY', 'SPECTROSCOPY', 'NONE']
ds_ws.SpectralWidth = bandwidth
ds_ws.SeriesNumber = series_no
ds_ws.ProtocolName = fname.split('\\')[-1].split(
'.')[0] + '_SVS_TR' + str(int(tr / 1000)) + '_TE' + str(
int(te / 1000))
c_freq2 = c_freq / 10000000.0
ds_ws.TransmitterFrequency = c_freq2
ds_ws.EffectiveEchoTime = te / 1000
ds_ws.EchoTime = te / 1000
ds_ws.WaterReferencedPhaseCorrection = 'N'
if w_avgs > 0:
file_meta_w = Dataset()
file_meta_w.MediaStorageSOPClassUID = 'MRSpectroscopyStorage'
file_meta_w.MediaStorageSOPInstanceUID = dicom.UID.generate_uid(
)
file_meta_w.ImplementationClassUID = dicom.UID.generate_uid()
ds_w = FileDataset(fname + '_dcm', {},
file_meta=file_meta,
preamble="\0" * 128)
ds_w.Modality = 'MR'
ds_w.ContentDate = str(datetime.date.today()).replace('-', '')
ds_w.ContentTime = str(
time.time()) #milliseconds since the epoch
ds_w.StudyInstanceUID = dicom.UID.generate_uid()
ds_w.SeriesInstanceUID = dicom.UID.generate_uid()
ds_w.SOPInstanceUID = dicom.UID.generate_uid()
ds_w.SOPClassUID = 'MRSpectroscopyStorage'
ds_w.SecondaryCaptureDeviceManufctur = 'Python 2.7.3'
ds_w.VolumeLocalizationTechnique = seqtype
## These are the necessary imaging components of the FileDataset object.
ds_w.Columns = 1
ds_w.Rows = 1
ds_w.DataPointColumns = int(acq_x_res)
ds_w.SpectroscopyAcquisitionDataColumns = ds_w.DataPointColumns
ds_w.DataPointRows = w_avgs
ds_w.SpectroscopyData = output_data_w
ds_w.ImageType = [
'ORIGINAL', 'PRIMARY', 'SPECTROSCOPY', 'NONE'
]
ds_w.SpectralWidth = bandwidth
ds_w.SeriesNumber = series_no
ds_w.ProtocolName = fname.split('\\')[-1].split(
'.')[0] + '_SVS_TR' + str(int(tr / 1000)) + '_TE' + str(
int(te / 1000))
c_freq2 = c_freq / 10000000.0
ds_w.TransmitterFrequency = c_freq2
ds_w.EffectiveEchoTime = te / 1000
ds_w.EchoTime = te / 1000
ds_w.RepetitionTime = tr / 1000
ds_w.WaterReferencedPhaseCorrection = 'N'
if not os.path.exists(
str(root_dir) + '/' + str(pat_name) + '_PREPROC_LCM'):
os.makedirs(
str(root_dir) + '/' + str(pat_name) + '_PREPROC_LCM')
# if not os.path.exists(str(root_dir)+'/'+str(pat_name)+'_LCM_Results'):
# os.makedirs(str(root_dir)+'/'+str(pat_name)+'_LCM_Results')
ds_ws.save_as(
str(root_dir) + '/' + str(pat_name) + '_PREPROC_LCM/' +
str(ds_ws.ProtocolName))
if w_avgs > 0:
ds_w.save_as(
str(root_dir) + '/' + str(pat_name) + '_PREPROC_LCM/' +
str(ds_w.ProtocolName) + '_WATER')
except AttributeError:
print('...')
except IOError:
print 'no such file'
print('DONE!')
