GlobusLearning precise segmentation of neurofibrillary tangles from rapid manual point annotations
Data files
Too many files!
You can download upto 1000 files at a time. For downloading a higher number of files, please use our alternative downloading methods.
Linked Information
Links Extracted from Files
-
-
-
-
-
-
-
-
-
-
-
-
- + 3 more
- 1 University of California, San Francisco
- 2 University of California, Davis
- 3 University of California, San Diego
- 4 Columbia University
AccessionS-BIAD1165
DescriptionAccumulation of abnormal tau protein into neurofibrillary tangles (NFTs) is a pathologic hallmark of Alzheimer disease. Accurate and efficient detection and quantification of NFTs in tissue samples aids in deeper phenotyping of Alzheimer disease and may reveal relationships with clinical, demographic, and genetic features. However, expert manual analysis can be time-consuming, subject to observer variability, and limited in handling the large amounts of data generated by modern imaging techniques. We present a scalable, open access, deep learning-based approach to quantify the NFT burden in digital whole slide images (WSIs) of post-mortem human brain tissue. We trained a UNet model on 45 annotated 2400µm by 1200µm regions of interest (ROIs) selected from 15 unique WSIs of temporal cortex from Alzheimer disease cases from three institutes (University of California (UC)-Davis, UC-San Diego, and Columbia University). We developed a method to generate detailed segmentation ground truth masks at the pixel level directly from simple point annotations. The model achieved a precision of 0.53, recall of 0.60, and F1 score of 0.53 on a held-out test set of 7 WSIs, providing researchers with an efficient and reliable tool for NFT burden quantification. We compared this to an object detection model on the same dataset, which achieved comparable but more coarse-grained performance. Both models correlated with expert semi-quantitative scores at the whole-slide level. Our approach provides an open deep learning pipeline for detailed and scalable NFT spatial distribution and morphology analysis across large cohorts, which is not feasible through manual assessment.
The dataset includes all WSIs and annotations used in the study.
The dataset includes all WSIs and annotations used in the study.
Keywordsdigital pathology,
neurofibrillary tangles,
deep learning,
NFT,
segmentation,
alzheimer disease,
tau,
object detection,
neuropathology
AcknowledgementsThe authors thank the families and participants of the University of California Davis, University of California San Diego, and Columbia University Alzheimer’s Disease Research Centers (ADRC) for their generous donations, as well as ADRC staff and faculty for their contributions. The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any public health agency or the US government. The authors would also like to thank Mikio Tada and Irene Wang for their contributions throughout the project, as well as David Gutman and JC Vizcarra for sharing ideas and data to strengthen this study.
LicenseCC0
Funding statementThis project was made possible by grant DAF2018-191905 (https://doi.org/10.37921/550142lkcjzw) from the Chan Zuckerberg Initiative DAF, an advised fund of the Silicon Valley Community Foundation (funder https://doi.org/10.13039/100014989) (M.J.K.) and grants from the National Institute on Aging (NIA) of the National Institutes of Health (NIH) under Award Numbers R01AG062517 (B.N.D.), P30AG072972 (C.D.), P30AG062429 (C.D.), P50AG008702 (A.F.T., Neuropathology Core), and P30AG066462 (A.F.T., Neuropathology Core).
Publication Sina Ghandian, Liane Albarghouthi, Kiana Nava, Shivam R. Rai Sharma, Lise Minaud, Laurel Beckett, Naomi Saito, Charles DeCarli, Robert A. Rissman, Andrew F. Teich, Lee-Way Jin, Brittany N. Dugger & Michael J. Keiser.
Learning precise segmentation of neurofibrillary tangles from rapid manual point annotations
Biosample 1
OrganismHomo sapiens (human)
DescriptionHuman brain tissue sliced, immunohistochemically stained with AT8 to target neurofibrillary tangles, and scanned with a Carl Zeiss scanner.
Biological entityHuman brain tissue, temporal lobe
Specimen-1
Sample preparation protocolSee paper for further details.
Image acquisition 1
Imaging instrumentZeiss Axio Scan Z.1
Image acquisition parameters40x magnification, 0.11 microns/px; stored at 60% quality via JPEG XR
Imaging methodmicroscopy with lenses,
light microscopy
Image analysis 1
Image analysis overviewUtilized a custom-written pipeline to carry out image analysis. Libraries used included Zarr, openCV, skimage, histomicsTK, libczi, czifile, imagecodecs, pytorch, pytorch-lightning, ultralytics, kornia, torchvision.
Study Component
NameExperiment A
DescriptionConvert point annotations to segmentation ground truth masks with which to train a deep learning model.
File Listfile_list.json
Associations