Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Aug 5;8(1):173.
doi: 10.1038/s41698-024-00664-0.

Deep learning-based detection and semi-quantitative model for spread through air spaces (STAS) in lung adenocarcinoma

Yipeng Feng #  1   2   3 , Hanlin Ding #  1   2   3 , Xing Huang #  4 , Yijian Zhang  1   2   3 , Mengyi Lu  5 , Te Zhang  1   2   3 , Hui Wang  1   2   3 , Yuzhong Chen  1   2   3 , Qixing Mao  1   2 , Wenjie Xia  1   2 , Bing Chen  1   2 , Yi Zhang  4 , Chen Chen  6 , Tianhao Gu  3 , Lin Xu  1   2   3   7 , Gaochao Dong  8   9   10 , Feng Jiang  11   12   13
Affiliations

Deep learning-based detection and semi-quantitative model for spread through air spaces (STAS) in lung adenocarcinoma

Yipeng Feng et al. NPJ Precis Oncol. .

Abstract

Tumor spread through air spaces (STAS) is a distinctive metastatic pattern affecting prognosis in lung adenocarcinoma (LUAD) patients. Several challenges are associated with STAS detection, including misdetection, low interobserver agreement, and lack of quantitative analysis. In this research, a total of 489 digital whole slide images (WSIs) were collected. The deep learning-based STAS detection model, named STASNet, was constructed to calculate semi-quantitative parameters associated with STAS density and distance. STASNet demonstrated an accuracy of 0.93 for STAS detection at the tiles level and had an AUC of 0.72-0.78 for determining the STAS status at the WSI level. Among the semi-quantitative parameters, T10S, combined with the spatial location information, significantly stratified stage I LUAD patients on disease-free survival. Additionally, STASNet was deployed into a real-time pathological diagnostic environment, which boosted the STAS detection rate and led to the identification of three easily misidentified types of occult STAS.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1. Dataset characteristics and pre-training preparation for model.
A The comprehensive study flowchart. B Patients and WSIs number of three cohorts. C Tiles of normal, tumor, and STAS at 128*128, 256*256, and 512*512 pixels.
Fig. 2
Fig. 2. Development of deep learning models and WSI workflow.
A Detailed architecture of STASNet. B Whole slide work flowchart based on STASNet.
Fig. 3
Fig. 3. STASNet was able to accurately identify the STAS.
A, D, G Represent digitated H&E of LUAD in three cohorts. B, E, H The result of STASNet at the represent digitated H&E. C, F, I The represent tiles (256*256 pixel) of the top 10 tiles of STAS score.
Fig. 4
Fig. 4. Spatial-related T10S is an excellent recurrence predictor.
A–C DFS curves for LUAD patients with STAS positive versus STAS negative; p-value reflects Log Rank testing. A All patients (n = 268, p = 0.0042). B Stage IA patients (n = 150, p = 0.0016). C Stage IB patients (n = 118, p = 0.42). D–F DFS curves for LUAD patients with T10S high versus T10S low. D All patients (n = 268, p = 0.00000028). E Stage IA patients (n = 150, p = 0.0010). F Stage IB patients (n = 118, p = 0.0000046). G C-index of T10S and STAS in the total cohort (T10S, C-index = 0.633 95% CI = 0.586–0.680), (STAS, C-index = 0.561, 95% CI = 0.513–0.609). H–J Time-dependent ROC curve analyses on the LUAD patients for predicting 1-, 2-, and 3-year DFS.
Fig. 5
Fig. 5. AI-assistance STAS detection.
A Workflow of STAS diagnosis combined with AI. B–D The characteristics of three types of mis-detection STAS (0.2418 μm/pixel).
See this image and copyright information in PMC

References

    1. Bray, F. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin.74, 229–263 (2024). 10.3322/caac.21834 - DOI - PubMed
    1. Howlader, N. et al. The effect of advances in lung-cancer treatment on population mortality. N. Engl. J. Med.383, 640–649 (2020). 10.1056/NEJMoa1916623 - DOI - PMC - PubMed
    1. Yanagawa, N. et al. New IASLC/ATS/ERS classification and invasive tumor size are predictive of disease recurrence in stage I lung adenocarcinoma. J. Thorac. Oncol.8, 612–618 (2013). 10.1097/JTO.0b013e318287c3eb - DOI - PubMed
    1. Kodama, T. et al. Cell incohesiveness and pattern of extension in a rare case of bronchioloalveolar carcinoma. Ultrastruct. Pathol.1, 177–188 (1980). 10.3109/01913128009141415 - DOI - PubMed
    1. Travis, W. D. et al. The 2015 World Health Organization Classification of Lung Tumors: impact of genetic, clinical and radiologic advances since the 2004 classification. J. Thorac. Oncol.10, 1243–1260 (2015). 10.1097/JTO.0000000000000630 - DOI - PubMed

LinkOut - more resources