def _load_ds_if_none(self):
hilbert_volume = self.hilbert_volume
if hilbert_volume is None:
hilbert_volume = load_nib(
pjoin(
DATA_DIRS['views_36'],
f'{self.filename[:-7]}_{self.plane.name.lower()}.nii.gz',
))
hilbert_volume = MyDataset._reduce_zdim(hilbert_volume)
sparse_volume = self.sparse_volume
if sparse_volume is None:
sparse_volume = load_nib(
pjoin(
DATA_DIRS['fdk_360'],
self.filename,
))[..., None]
sparse_volume = MyDataset._reduce_zdim(sparse_volume)
image_volume = self.image_volume
if image_volume is None:
image_volume = load_nib(
pjoin(
DATA_DIRS['datasets'],
self.filename,
))[..., None]
image_volume = MyDataset._reduce_zdim(image_volume)
return hilbert_volume, sparse_volume, image_volume
def blend_method(method: str, blur: int = 90, save: bool = True):
assert method in ['fdkconv', 's2f_inv', 's2f_inv3', 'inv_sp', 'inv_sp3']
reco_coronal = load_nib(f'testing/{method}_coronal.nii.gz')
reco_sagittal = load_nib(f'testing/{method}_sagittal.nii.gz')
# non-negativity constraint
reco_coronal[reco_coronal < 0] = 0
reco_sagittal[reco_sagittal < 0] = 0
# [x, y, z] -> [z, y, x]
reco_coronal = reco_coronal.transpose()
reco_sagittal = reco_sagittal.transpose()
wedge = create_wedge((512, 512), blur)[None, ...]
# spectral blending
reco_coronal_fft = np.fft.fftshift(np.fft.fft2(reco_coronal))
reco_sagittal_fft = np.fft.fftshift(np.fft.fft2(reco_sagittal))
blended_fft = wedge * reco_sagittal_fft + (1 - wedge) * reco_coronal_fft
blended = np.real(np.fft.ifft2(np.fft.fftshift(blended_fft)))
if save:
img = nib.Nifti1Image(blended.transpose(), np.eye(4))
nib.save(img, f'testing/{method}_blended.nii.gz')
return blended
def calculate_z_depending_metrics():
with open('train_valid.json', 'r') as json_file:
json_dict = json.load(json_file)
test_files = json_dict['test_files']
gt = load_nib(pjoin(DATA_DIRS['datasets'], test_files[0]))
gt = reduce_zdim(gt).transpose()
prediction = load_nib('testing/inv_sp3_blended.nii.gz')
prediction = prediction.transpose()
fdkconv = load_nib('testing/fdkconv_blended.nii.gz')
fdkconv = reduce_zdim(fdkconv).transpose()
pred_nmse = nmse(prediction, gt, lambda _: _)
pred_psnr = psnr(prediction, gt, lambda _: _)
pred_ssim = ssim(prediction, gt, lambda _: _)
fdkconv_nmse = nmse(fdkconv, gt, lambda _: _)
fdkconv_psnr = psnr(fdkconv, gt, lambda _: _)
fdkconv_ssim = ssim(fdkconv, gt, lambda _: _)
from matplotlib import pyplot as plt
plt.plot(pred_nmse / fdkconv_nmse), plt.figure()
plt.plot(pred_psnr / fdkconv_psnr), plt.figure()
plt.plot(pred_ssim / fdkconv_ssim), plt.show()
def calculate_sparse_fdk_metrics():
with open('train_valid.json', 'r') as json_file:
json_dict = json.load(json_file)
test_files = json_dict['test_files']
gt = load_nib(pjoin(DATA_DIRS['datasets'], test_files[0]))
gt = reduce_zdim(gt).transpose()
sparse_fdk = load_nib('sparse_fdk.nii.gz')
sparse_fdk = reduce_zdim(sparse_fdk).transpose()
return {
'nmse': nmse(sparse_fdk, gt),
'psnr': psnr(sparse_fdk, gt),
'ssim': ssim(sparse_fdk, gt),
}
def calculate_metrics(method: str, plane: HilbertPlane):
with open('train_valid.json', 'r') as json_file:
json_dict = json.load(json_file)
test_files = json_dict['test_files']
gt = load_nib(pjoin(DATA_DIRS['datasets'], test_files[0]))
gt = reduce_zdim(gt).transpose()
prediction = load_nib(f'testing/{method}_{plane.name.lower()}.nii.gz')
prediction = prediction.transpose()
return {
'nmse': nmse(prediction, gt),
'psnr': psnr(prediction, gt),
'ssim': ssim(prediction, gt),
}
def blend_sweep_blur():
with open('train_valid.json', 'r') as json_file:
json_dict = json.load(json_file)
test_files = json_dict['test_files']
gt = load_nib(pjoin(DATA_DIRS['datasets'], test_files[0]))
gt = reduce_zdim(gt).transpose()
reco_coronal = load_nib('testing/inv_sp3_coronal.nii.gz')
reco_sagittal = load_nib('testing/inv_sp3_sagittal.nii.gz')
# non-negativity constraint
reco_coronal[reco_coronal < 0] = 0
reco_sagittal[reco_sagittal < 0] = 0
# [x, y, z] -> [z, y, x]
reco_coronal = reco_coronal.transpose()
reco_sagittal = reco_sagittal.transpose()
all_blurs = np.linspace(1, 90, 10)
all_metrics = {'nmse': [], 'psnr': [], 'ssim': []}
for blur in all_blurs:
wedge = create_wedge((512, 512), blur)[None, ...]
# spectral blending
reco_coronal_fft = np.fft.fftshift(np.fft.fft2(reco_coronal))
reco_sagittal_fft = np.fft.fftshift(np.fft.fft2(reco_sagittal))
blended_fft = (wedge) * reco_sagittal_fft + (1 -
wedge) * reco_coronal_fft
blended = np.real(np.fft.ifft2(np.fft.fftshift(blended_fft)))
nmse_value = nmse(blended, gt)
psnr_value = psnr(blended, gt)
ssim_value = ssim(blended, gt)
print(f'{blur} {nmse_value} {psnr_value} {ssim_value}')
all_metrics['nmse'].append(nmse_value)
all_metrics['psnr'].append(psnr_value)
all_metrics['ssim'].append(ssim_value)
with open('spectral_blur_sweep.csv', 'w', newline='') as csv_file:
writer = csv.writer(csv_file)
writer.writerow(['blur'] + all_blurs)
writer.writerow(['nmse'] + all_metrics['nmse'])
writer.writerow(['psnr'] + all_metrics['psnr'])
writer.writerow(['ssim'] + all_metrics['ssim'])
def _load_ds_if_none(self):
sparse_volume = self.sparse_volume
if sparse_volume is None:
sparse_volume = load_nib(
pjoin(
DATA_DIRS['fdk_36'],
self.filename,
))[..., None]
sparse_volume = MyDataset._reduce_zdim(sparse_volume)
image_volume = self.image_volume
if image_volume is None:
image_volume = load_nib(
pjoin(DATA_DIRS['datasets'], self.filename), )[..., None]
image_volume = MyDataset._reduce_zdim(image_volume)
return sparse_volume, image_volume
def _load_ds_if_none(self):
sparse_hilbert = self.sparse_hilbert
if sparse_hilbert is None:
sparse_hilbert = load_nib(
pjoin(
DATA_DIRS['views_36'], f'{self.filename[:-7]}'
f'_{self.plane.name.lower()}.nii.gz'))
sparse_hilbert = MyDataset._reduce_zdim(sparse_hilbert)
full_hilbert = self.full_hilbert
if full_hilbert is None:
full_hilbert = load_nib(
pjoin(
DATA_DIRS['views_360'], f'{self.filename[:-7]}'
f'_{self.plane.name.lower()}.nii.gz'))
full_hilbert = MyDataset._reduce_zdim(full_hilbert)
return sparse_hilbert, full_hilbert