-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathidr0042-study.txt
69 lines (59 loc) · 6 KB
/
idr0042-study.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
# STUDY DESCRIPTION SECTION
"# Section with generic information about the study including title, description, publication details (if applicable) and contact details"
Comment[IDR Study Accession] idr0042
Study Title A Deep-learning classifier identifies patients with clinical heart failure using whole-slide images of H&E tissue
Study Type histology
Study Type Term Source REF
Study Type Term Accession
Study Description Cardiac histopathology from human patients with clinical heart failure or cadaveric donor hearts from patients without clinical heart failure. Human tissue research: Human heart tissue was procured from two separate groups of subjects: heart transplant or LVAD recipients with severe heart failure (Fal), and brain dead, organ donors with no history of heart failure (non-failing, NF). Tissue from patients with ischemic cardiomyopathy sampled infarct-free regions. No organs or tissue were procured from prisoners. Prospective informed consent for research use of heart tissue was obtained from all transplant or LVAD recipients and next-of-kin in the case of organ donors. All patient data and images were de-identified, and all protocols were performed in accordance with relevant guidelines for research involving tissue from human subjects. Tissue used in this study was collected and processed at the Cardiovascular Research Institute and the Department of Pathology and Laboratory Medicine at the University of Pennsylvania between 2008 and 2013. All patients were from the same institutional cohort. All study procedures were approved or waived by the University of Pennsylvania Institutional Review Board
Study Organism Homo sapiens
Study Organism Term Source REF NCBITaxon
Study Organism Term Accession NCBITaxon_9606
Study Experiments Number 1
Study External URL
Study Public Release Date 2018-04-12
# Study Publication
Study PubMed ID 29614076
Study Publication Title A deep-learning classifier identifies patients with clinical heart failure using whole-slide images of H&E tissue.
Study Author List Nirschl JJ, Janowczyk A, Peyster EG, Frank R, Margulies KB, Feldman MD, Madabhushi A
Study PMC ID PMC5882098
Study DOI https://doi.org/10.1371/journal.pone.0192726
# Study License and Data DOI
Study License CC BY 4.0
Study License URL https://creativecommons.org/licenses/by/4.0/
Study Copyright Nirschl et al
Study Data Publisher University of Dundee
Study Data DOI https://doi.org/10.17867/10000113
# Study Contacts
Study Person Last Name Madabhushi
Study Person First Name Anant
Study Person Email axm788@case.edu
Study Person Address Department of Biomedical Engineering, Case Western Reserve University, Cleveland, OH, 44106, USA
Study Person Roles Corresponding author
Term Source Name NCBITaxon CVDO FBbi PATO UBERON SNOMED
Term Source URI https://www.ncbi.nlm.nih.gov/taxonomy https://www.ebi.ac.uk/ols/ontologies/cvdo https://www.ebi.ac.uk/ols/ontologies/fbbi https://www.ebi.ac.uk/ols/ontologies/pato https://www.ebi.ac.uk/ols/ontologies/uberon http://bioportal.bioontology.org/ontologies/SNOMEDCT
# EXPERIMENT SECTION
"# Experiment Section containing all information relative to each experiment in the study including materials used, protocols names and description, phenotype names and description. For multiple experiments this section should be repeated. Copy and paste the whole section below and fill out for the next experiment"
Experiment Number 1
Comment[IDR Experiment Name] idr0042-nirschl-wsideeplearning/experimentA
Experiment Sample Type tissue
Experiment Description Histopathology sub-image from patient whole-slide images from patients with end-stage clinical heart failure or cadaveric donor hearts from patients without heart failure.
Experiment Size Datasets: 3 2D Images: 2299 Average Image Dimension (XYZCT): 250 x 250 x 1 x 3 x 1 Total Mb: 314 Mb
Experiment Example Images https://idr.openmicroscopy.org/webclient/?show=image-3428111 https://idr.openmicroscopy.org/webclient/img_detail/3428111
Experiment Imaging Method bright-field microscopy
Experiment Imaging Method Term Source REF FBbi
Experiment Imaging Method Term Accession FBbi_00000243
Experiment Comments Detection of clinical heart failure or severe tissue pathology from images of cardiac histopathology using convolutional neural networks.
# assay files
Experiment Assay File idr0042-experimentA-assays.txt
Experiment Assay File Format tab-delimited text
Assay Experimental Conditions Disease
Assay Experimental Conditions Term Source REF EFO
Assay Experimental Conditions Term Accession EFO_0000408
Quality Control Description Training and held-out test datasets were randomly allocated prior to image analysis. The training dataset was split into three folds for cross-validation to monitor training progress. The held-out test dataset was tested after all training, fine-tuning, and optimization was performed.
# Protocols
Protocol Name treatment protocol image acquisition
Protocol Type treatment protocol image acquisition
Protocol Type Term Source REF EFO EFO
Protocol Type Term Accession EFO_0003969
Protocol Description Dataset collection and histological processing: Both failing and non-failing hearts received in situ cold cardioplegia in the operating room and were immediately placed on wet ice in 4°C Krebs-Henseleit buffer. Within 4 hours of cardiectomy, transmural tissue from the left ventricular free wall were fixed in 4% paraformaldehyde and later processed, embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E) for morphologic analysis. Whole-slide images were acquired at 20x magnification using an Aperio ScanScope slide scanner. Images were down-sampled to 5x magnification for image analysis, a magnification sufficient for expert assessment of gross tissue pathology. The allocation to the training and held-out test cohort was random and performed prior to image analysis.