pvs_quantification
Current SHiVAi pipeline (2025-06) seems to have problem handling html/pdf reports (for example, when facing NaN value). A workaround is to modify the code in post.py to simply skip the report generation step, which is not critical for the pipeline to run.
class SummaryReport(BaseInterface):
"""Make a summary report of preprocessing and prediction"""
input_spec = SummaryReportInputSpec
output_spec = SummaryReportOutputSpec
# def _run_interface(self, runtime):
# """
# Build the report for the whole workflow. It contains segmentation statistics and
# quality control figures.
# """
# if self.inputs.anonymized:
# subject_id = ''
# else:
# subject_id = self.inputs.subject_id
# brain_vol_vox = nib.load(self.inputs.brainmask).get_fdata().astype(bool).sum() # in voxels
# pred_metrics_dict = {} # Will contain the stats dataframe for each biomarker
# pred_census_im_dict = {} # Will contain the path to the swarm plot for each biomarker
# pred_overlay_im_dict = {} # Will contain the path to the figure with biomarkers overlaid on the brain
# models_uid = {} # Will contain the md5 hash for each file of each predictive model
# pred_and_acq = self.inputs.pred_and_acq
# # Generate the distribution figures for each prediction and fill models_uid
# for pred in pred_and_acq:
# lpred = pred.lower() # "pred" is uppercase, so we also need a lowercase version
# if pred == 'LAC':
# name_in_plot = 'Lacuna'
# else:
# name_in_plot = pred
# models_uid[pred] = {}
# pred_metrics_dict[pred] = pd.read_csv(getattr(self.inputs, f'{lpred}_metrics_csv'))
# if pred_metrics_dict[pred]['Number of clusters'].sum() == 0: # No biomarker detected
# pred_census_im_dict[pred] = None
# else:
# pred_census_im_dict[pred] = violinplot_from_census(getattr(self.inputs, f'{lpred}_census_csv'),
# self.inputs.resolution,
# name_in_plot)
# pred_overlay_im_dict[pred] = getattr(self.inputs, f'{lpred}_overlay')
# ids, url = get_md5_from_json(getattr(self.inputs, f'{lpred}_model_descriptor'), get_url=True)
# models_uid[pred]['id'] = ids
# if url:
# models_uid[pred]['url'] = url
# # set optional inputs to None if undefined
# if isdefined(self.inputs.overlayed_brainmask_1):
# overlayed_brainmask_1 = self.inputs.overlayed_brainmask_1
# else:
# overlayed_brainmask_1 = None
# if isdefined(self.inputs.crop_brain_img):
# crop_brain_img = self.inputs.crop_brain_img
# else:
# crop_brain_img = None
# if isdefined(self.inputs.isocontour_slides_FLAIR_T1):
# isocontour_slides_FLAIR_T1 = self.inputs.isocontour_slides_FLAIR_T1
# else:
# isocontour_slides_FLAIR_T1 = None
# if isdefined(self.inputs.overlayed_brainmask_2):
# overlayed_brainmask_2 = self.inputs.overlayed_brainmask_2
# else:
# overlayed_brainmask_2 = None
# if isdefined(self.inputs.wf_graph):
# wf_graph = self.inputs.wf_graph
# else:
# wf_graph = None
# if isdefined(self.inputs.db):
# db = self.inputs.db
# else:
# db = ''
# # process
# html_report = make_report(
# pred_metrics_dict=pred_metrics_dict,
# pred_census_im_dict=pred_census_im_dict,
# pred_overlay_im_dict=pred_overlay_im_dict,
# pred_and_acq=pred_and_acq,
# brain_vol_vox=brain_vol_vox,
# thr_cluster_vals=self.inputs.thr_cluster_vals,
# min_seg_size=self.inputs.min_seg_size,
# models_uid=models_uid,
# bounding_crop=crop_brain_img,
# overlayed_brainmask_1=overlayed_brainmask_1,
# overlayed_brainmask_2=overlayed_brainmask_2,
# isocontour_slides_FLAIR_T1=isocontour_slides_FLAIR_T1,
# subject_id=subject_id,
# image_size=self.inputs.image_size,
# resolution=self.inputs.resolution,
# percentile=self.inputs.percentile,
# threshold=self.inputs.threshold,
# wf_graph=wf_graph
# )
# with open('Shiva_report.html', 'w', encoding='utf-8') as fid:
# fid.write(html_report)
# # Convert the HTML file to PDF using CSS
# # Creating custom CSSin addition to the main one for the pages header and the logos
# postproc_dir = os.path.dirname(postproc_init)
# css = CSS(os.path.join(postproc_dir, 'printed_styling.css'))
# now = datetime.now(timezone.utc).strftime("%Y/%m/%d - %H:%M (UTC)")
# content_sub_id = f'Patient ID: {subject_id} \A ' if subject_id else ''
# header = (
# '@page {'
# ' @top-left {'
# f' content: "{content_sub_id}Data-base: {db}";'
# ' font-size: 10pt;'
# ' white-space: pre;'
# ' }'
# ' @top-center {'
# f' content: "{now}";'
# ' font-size: 10pt;'
# ' }'
# '}'
# )
# css_header = CSS(string=header)
# shiva_logo_file = os.path.join(postproc_dir, 'logo_shiva.png')
# other_logos_file = os.path.join(postproc_dir, 'logos_for_shiva.png')
# with open(shiva_logo_file, 'rb') as f:
# image_data = f.read()
# shiva_logo = base64.b64encode(image_data).decode()
# with open(other_logos_file, 'rb') as f:
# image_data = f.read()
# other_logo = base64.b64encode(image_data).decode()
# logo = (
# '@page {'
# ' @bottom-left {'
# f' background-image: url(data:image/png;base64,{other_logo});'
# ' background-size: 552px 45px;'
# ' display: inline-block;'
# ' width: 560px; '
# ' height: 60px;'
# ' content:"";'
# ' background-repeat: no-repeat;'
# ' }'
# ' @top-right-corner {'
# f' background-image: url(data:image/png;base64,{shiva_logo});'
# ' background-size: 70px 70px;'
# ' display: inline-block;'
# ' width: 70px; '
# ' height: 70px;'
# ' content:"";'
# ' background-repeat: no-repeat;'
# ' }'
# '}'
# )
# logo_css = CSS(string=logo)
# HTML('Shiva_report.html').write_pdf('Shiva_report.pdf',
# stylesheets=[css, css_header, logo_css])
# setattr(self, 'html_report', os.path.abspath('Shiva_report.html'))
# setattr(self, 'pdf_report', os.path.abspath('Shiva_report.pdf'))
# return runtime
def _run_interface(self, runtime):
"""Skip actual report generation but create placeholder files."""
print("SummaryReport disabled: generating empty HTML and PDF files to avoid workflow errors.")
# Create empty HTML file
html_path = os.path.abspath('Shiva_report.html')
with open(html_path, 'w', encoding='utf-8') as f:
f.write('<html><body><p>Summary report generation skipped.</p></body></html>')
# Create empty (but valid) PDF file header to avoid downstream rendering issues
pdf_path = os.path.abspath('Shiva_report.pdf')
with open(pdf_path, 'wb') as f:
f.write(b'%PDF-1.4\n%EOF\n') # minimal valid PDF structure
setattr(self, 'html_report', html_path)
setattr(self, 'pdf_report', pdf_path)
return runtime
def _list_outputs(self):
"""Fill in the output structure."""
outputs = self.output_spec().trait_get()
outputs['html_report'] = getattr(self, 'html_report')
outputs['pdf_report'] = getattr(self, 'pdf_report')
return outputs