Pixel-Level Classification of Five Histologic Patterns of Lung Adenocarcinoma.

Shao, D; Su, F; Zou, X; Lu, J; Wu, S; Tian, R; Ran, D; Guo, Z; Jin, D

Pixel-Level Classification of Five Histologic Patterns of Lung Adenocarcinoma.

Shao, D Su, F Zou, X Lu, J

Wu, S Tian, R Ran, D Guo, Z Jin, D

Permalink

Publisher:: AMER CHEMICAL SOC
Publication Type:: Journal Article
Citation:: Anal Chem, 2023, 95, (5), pp. 2664-2670
Issue Date:: 2023-02-07

Closed Access

	Filename	Description	Size
	shao-et-al-2023-pixel-level-classification-of-five-histologic-patterns-of-lung-adenocarcinoma.pdf	Published version	5.32 MB		View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Shao, D
dc.contributor.author	Su, F
dc.contributor.author	Zou, X
dc.contributor.author	Lu, J https://orcid.org/0000-0003-0690-4732
dc.contributor.author	Wu, S
dc.contributor.author	Tian, R
dc.contributor.author	Ran, D
dc.contributor.author	Guo, Z
dc.contributor.author	Jin, D https://orcid.org/0000-0003-1046-2666
dc.date.accessioned	2024-02-08T23:28:18Z
dc.date.available	2024-02-08T23:28:18Z
dc.date.issued	2023-02-07
dc.identifier.citation	Anal Chem, 2023, 95, (5), pp. 2664-2670
dc.identifier.issn	0003-2700
dc.identifier.issn	1520-6882
dc.identifier.uri	http://hdl.handle.net/10453/175525
dc.description.abstract	Lung adenocarcinoma is the most common histologic type of lung cancer. The pixel-level labeling of histologic patterns of lung adenocarcinoma can assist pathologists in determining tumor grading with more details than normal classification. We manually annotated a dataset containing a total of 1000 patches (200 patches for each pattern) of 512 × 512 pixels and 420 patches (contains test sets) of 1024 × 1024 pixels according to the morphological features of the five histologic patterns of lung adenocarcinoma (lepidic, acinar, papillary, micropapillary, and solid). To generate an even large amount of data patches, we developed a data stitching strategy as a data augmentation for classification in model training. Stitched patches improve the Dice similarity coefficient (DSC) scores by 24.06% on the whole-slide image (WSI) with the solid pattern. We propose a WSI analysis framework for lung adenocarcinoma pathology, intelligently labeling lung adenocarcinoma histologic patterns at the pixel level. Our framework contains five branches of deep neural networks for segmenting each histologic pattern. We test our framework with 200 unclassified patches. The DSC scores of our results outpace comparing networks (U-Net, LinkNet, and FPN) by up to 10.78%. We also perform results on four WSIs with an overall accuracy of 99.6%, demonstrating that our network framework exhibits better accuracy and robustness in most cases.
dc.format	Print-Electronic
dc.language	eng
dc.publisher	AMER CHEMICAL SOC
dc.relation.ispartof	Anal Chem
dc.relation.isbasedon	10.1021/acs.analchem.2c03020
dc.rights	info:eu-repo/semantics/closedAccess
dc.subject	0301 Analytical Chemistry, 0399 Other Chemical Sciences
dc.subject.classification	Analytical Chemistry
dc.subject.classification	3205 Medical biochemistry and metabolomics
dc.subject.classification	3401 Analytical chemistry
dc.subject.classification	4004 Chemical engineering
dc.subject.mesh	Humans
dc.subject.mesh	Adenocarcinoma
dc.subject.mesh	Adenocarcinoma of Lung
dc.subject.mesh	Lung Neoplasms
dc.subject.mesh	Neoplasm Grading
dc.subject.mesh	Neural Networks, Computer
dc.subject.mesh	Humans
dc.subject.mesh	Adenocarcinoma
dc.subject.mesh	Lung Neoplasms
dc.subject.mesh	Neoplasm Grading
dc.subject.mesh	Adenocarcinoma of Lung
dc.subject.mesh	Neural Networks, Computer
dc.subject.mesh	Humans
dc.subject.mesh	Adenocarcinoma
dc.subject.mesh	Adenocarcinoma of Lung
dc.subject.mesh	Lung Neoplasms
dc.subject.mesh	Neoplasm Grading
dc.subject.mesh	Neural Networks, Computer
dc.title	Pixel-Level Classification of Five Histologic Patterns of Lung Adenocarcinoma.
dc.type	Journal Article
utslib.citation.volume	95
utslib.location.activity	United States
utslib.for	0301 Analytical Chemistry
utslib.for	0399 Other Chemical Sciences
pubs.organisational-group	University of Technology Sydney
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	University of Technology Sydney/Faculty of Science
pubs.organisational-group	University of Technology Sydney/Faculty of Science/School of Mathematical and Physical Sciences
pubs.organisational-group	University of Technology Sydney/Strength - AAII - Australian Artificial Intelligence Institute
pubs.organisational-group	University of Technology Sydney/Strength - IBMD - Initiative for Biomedical Devices
utslib.copyright.status	closed_access	*
dc.date.updated	2024-02-08T23:28:15Z
pubs.issue	5
pubs.publication-status	Published
pubs.volume	95
utslib.citation.issue	5

Abstract:

Lung adenocarcinoma is the most common histologic type of lung cancer. The pixel-level labeling of histologic patterns of lung adenocarcinoma can assist pathologists in determining tumor grading with more details than normal classification. We manually annotated a dataset containing a total of 1000 patches (200 patches for each pattern) of 512 × 512 pixels and 420 patches (contains test sets) of 1024 × 1024 pixels according to the morphological features of the five histologic patterns of lung adenocarcinoma (lepidic, acinar, papillary, micropapillary, and solid). To generate an even large amount of data patches, we developed a data stitching strategy as a data augmentation for classification in model training. Stitched patches improve the Dice similarity coefficient (DSC) scores by 24.06% on the whole-slide image (WSI) with the solid pattern. We propose a WSI analysis framework for lung adenocarcinoma pathology, intelligently labeling lung adenocarcinoma histologic patterns at the pixel level. Our framework contains five branches of deep neural networks for segmenting each histologic pattern. We test our framework with 200 unclassified patches. The DSC scores of our results outpace comparing networks (U-Net, LinkNet, and FPN) by up to 10.78%. We also perform results on four WSIs with an overall accuracy of 99.6%, demonstrating that our network framework exhibits better accuracy and robustness in most cases.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/175525