Title: Deep segmentation networks predict survival of non-small cell lung cancer
Abstract Non-small-cell lung cancer (NSCLC) represents approximately 80–85% of lung cancer diagnoses and is the leading cause of cancer-related death worldwide. Recent studies indicate that image-based radiomics features from positron emission tomography/computed tomography (PET/CT) images have predictive power for NSCLC outcomes. To this end, easily calculated functional features such as the maximum and the mean of standard uptake value (SUV) and total lesion glycolysis (TLG) are most commonly used for NSCLC prognostication, but their prognostic value remains controversial. Meanwhile, convolutional neural networks (CNN) are rapidly emerging as a new method for cancer image analysis, with significantly enhanced predictive power compared to hand-crafted radiomics features. Here we show that CNNs trained to perform the tumor segmentation task, with no other information than physician contours, identify a rich set of survival-related image features with remarkable prognostic value. In a retrospective study on pre-treatment PET-CT images of 96 NSCLC patients before stereotactic-body radiotherapy (SBRT), we found that the CNN segmentation algorithm (U-Net) trained for tumor segmentation in PET and CT images, contained features having strong correlation with 2- and 5-year overall and disease-specific survivals. The U-Net algorithm has not seen any other clinical information (e.g. survival, age, smoking history, etc.) than the images and the corresponding tumor contours provided by physicians. In addition, we observed the same trend by validating the U-Net features against an extramural data set provided by Stanford Cancer Institute. Furthermore, through visualization of the U-Net, we also found convincing evidence that the regions of metastasis and recurrence appear to match with the regions where the U-Net features identified patterns that predicted higher likelihoods of death. We anticipate our findings will be a starting point for more sophisticated non-intrusive patient specific cancer prognosis determination. For example, the deep learned PET/CT features can not only predict survival but also visualize high-risk regions within or adjacent to the primary tumor and hence potentially impact therapeutic outcomes by optimal selection of therapeutic strategy or first-line therapy adjustment. more »« less
Zhong, Zisha; Kim, Yusung; Zhou, Leixin; Plichta, Kristin; Allen, Bryan; Buatti, John; Wu, Xiaodong
(, 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018))
null
(Ed.)
Positron emission tomography and computed tomography (PET-CT) plays a critically important role in modern cancer therapy. In this paper, we focus on automated tumor delineation on PET-CT image pairs. Inspired by co-segmentation model, we develop a novel 3D image co-matting technique making use of the inner-modality information of PET and CT for matting. The obtained co-matting results are then incorporated in the graph-cut based PET-CT co-segmentation framework. Our comparative experiments on 32 PET-CT scan pairs of lung cancer patients demonstrate that the proposed 3D image co-matting technique can significantly improve the quality of cost images for the co-segmentation, resulting in highly accurate tumor segmentation on both PET and CT scan pairs.
Zhong, Zisha; Kim, Yusung; Zhou, Leixin; Plichta, Kristin; Allen, Bryan; Buatti, John; Wu, Xiaodong
(, 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018))
null
(Ed.)
Positron emission tomography and computed tomography (PET-CT) dual-modality imaging provides critical diagnostic information in modern cancer diagnosis and therapy. Automated accurate tumor delineation is essentially important in computer-assisted tumor reading and interpretation based on PET-CT. In this paper, we propose a novel approach for the segmentation of lung tumors that combines the powerful fully convolutional networks (FCN) based semantic segmentation framework (3D-UNet) and the graph cut based co-segmentation model. First, two separate deep UNets are trained on PET and CT, separately, to learn high level discriminative features to generate tumor/non-tumor masks and probability maps for PET and CT images. Then, the two probability maps on PET and CT are further simultaneously employed in a graph cut based co-segmentation model to produce the final tumor segmentation results. Comparative experiments on 32 PET-CT scans of lung cancer patients demonstrate the effectiveness of our method.
In NSCLC, there is a pressing need for immunotherapy predictive biomarkers. The processes underlying B-cell dysfunction, as well as their prognostic importance in NSCLC, are unknown. Tumor-specific B-cell gene co-expression networks were constructed by comparing the Boolean implication modeling of single-cell RNA sequencing of NSCLC tumor B cells and normal B cells. Proliferation genes were selected from the networks using in vitro CRISPR-Cas9/RNA interfering (RNAi) screening data in more than 92 human NSCLC epithelial cell lines. The prognostic and predictive evaluation was performed using public NSCLC transcriptome and proteome profiles. A B cell proliferation and prognostic gene co-expression network was present only in normal lung B cells and missing in NSCLC tumor B cells. A nine-gene signature was identified from this B cell network that provided accurate prognostic stratification using bulk NSCLC tumor transcriptome (n = 1313) and proteome profiles (n = 103). Multiple genes (HLA-DRA, HLA-DRB1, OAS1, and CD74) differentially expressed in NSCLC B cells, peripheral blood lymphocytes, and tumor T cells had concordant prognostic indications at the mRNA and protein expression levels. The selected genes were associated with drug sensitivity/resistance to 10 commonly used NSCLC therapeutic regimens. Lestaurtinib was discovered as a potential repositioning drug for treating NSCLC.
Hida, Tomoyuki; Hata, Akinori; Lu, Junwei; Valtchinov, Vladimir I.; Hino, Takuya; Nishino, Mizuki; Honda, Hiroshi; Tomiyama, Noriyuki; Christiani, David C.; Hatabu, Hiroto
(, Cancer Imaging)
null
(Ed.)
Abstract Background Interstitial lung abnormalities (ILA) can be detected on computed tomography (CT) in lung cancer patients and have an association with mortality in advanced non-small cell lung cancer (NSCLC) patients. The aim of this study is to demonstrate the significance of ILA for mortality in patients with stage I NSCLC using Boston Lung Cancer Study cohort. Methods Two hundred and thirty-one patients with stage I NSCLC from 2000 to 2011 were investigated in this retrospective study (median age, 69 years; 93 males, 138 females). ILA was scored on baseline CT scans prior to treatment using a 3-point scale (0 = no evidence of ILA, 1 = equivocal for ILA, 2 = ILA) by a sequential reading method. ILA score 2 was considered the presence of ILA. The difference of overall survival (OS) for patients with different ILA scores were tested via log-rank test and multivariate Cox proportional hazards models were used to estimate hazard ratios (HRs) including ILA score, age, sex, smoking status, and treatment as the confounding variables. Results ILA was present in 22 out of 231 patients (9.5%) with stage I NSCLC. The presence of ILA was associated with shorter OS (patients with ILA score 2, median 3.85 years [95% confidence interval (CI): 3.36 – not reached (NR)]; patients with ILA score 0 or 1, median 10.16 years [95%CI: 8.65 - NR]; P < 0.0001). In a Cox proportional hazards model, the presence of ILA remained significant for increased risk for death (HR = 2.88, P = 0.005) after adjusting for age, sex, smoking and treatment. Conclusions ILA was detected on CT in 9.5% of patients with stage I NSCLC. The presence of ILA was significantly associated with a shorter OS and could be an imaging marker of shorter survival in stage I NSCLC.
Stevens, Jack B; Riley, Breylon A; Je, Jihyeon; Gao, Yuan; Wang, Chunhao; Mowery, Yvonne M; Brizel, David M; Yin, Fang‐Fang; Liu, Jian‐Guo; Lafata, Kyle J
(, Medical Physics)
Abstract BackgroundDelta radiomics is a high‐throughput computational technique used to describe quantitative changes in serial, time‐series imaging by considering the relative change in radiomic features of images extracted at two distinct time points. Recent work has demonstrated a lack of prognostic signal of radiomic features extracted using this technique. We hypothesize that this lack of signal is due to the fundamental assumptions made when extracting features via delta radiomics, and that other methods should be investigated. PurposeThe purpose of this work was to show a proof‐of‐concept of a new radiomics paradigm for sparse, time‐series imaging data, where features are extracted from a spatial‐temporal manifold modeling the time evolution between images, and to assess the prognostic value on patients with oropharyngeal cancer (OPC). MethodsTo accomplish this, we developed an algorithm to mathematically describe the relationship between two images acquired at time and . These images serve as boundary conditions of a partial differential equation describing the transition from one image to the other. To solve this equation, we propagate the position and momentum of each voxel according to Fokker–Planck dynamics (i.e., a technique common in statistical mechanics). This transformation is driven by an underlying potential force uniquely determined by the equilibrium image. The solution generates a spatial‐temporal manifold (3 spatial dimensions + time) from which we define dynamic radiomic features. First, our approach was numerically verified by stochastically sampling dynamic Gaussian processes of monotonically decreasing noise. The transformation from high to low noise was compared between our Fokker–Planck estimation and simulated ground‐truth. To demonstrate feasibility and clinical impact, we applied our approach to18F‐FDG‐PET images to estimate early metabolic response of patients (n = 57) undergoing definitive (chemo)radiation for OPC. Images were acquired pre‐treatment and 2‐weeks intra‐treatment (after 20 Gy). Dynamic radiomic features capturing changes in texture and morphology were then extracted. Patients were partitioned into two groups based on similar dynamic radiomic feature expression via k‐means clustering and compared by Kaplan–Meier analyses with log‐rank tests (p < 0.05). These results were compared to conventional delta radiomics to test the added value of our approach. ResultsNumerical results confirmed our technique can recover image noise characteristics given sparse input data as boundary conditions. Our technique was able to model tumor shrinkage and metabolic response. While no delta radiomics features proved prognostic, Kaplan–Meier analyses identified nine significant dynamic radiomic features. The most significant feature was Gray‐Level‐Size‐Zone‐Matrix gray‐level variance (p = 0.011), which demonstrated prognostic improvement over its corresponding delta radiomic feature (p = 0.722). ConclusionsWe developed, verified, and demonstrated the prognostic value of a novel, physics‐based radiomics approach over conventional delta radiomics via data assimilation of quantitative imaging and differential equations.
Baek, Stephen, He, Yusen, Allen, Bryan G., Buatti, John M., Smith, Brian J., Tong, Ling, Sun, Zhiyu, Wu, Jia, Diehn, Maximilian, Loo, Billy W., Plichta, Kristin A., Seyedin, Steven N., Gannon, Maggie, Cabel, Katherine R., Kim, Yusung, and Wu, Xiaodong. Deep segmentation networks predict survival of non-small cell lung cancer. Retrieved from https://par.nsf.gov/biblio/10390829. Scientific Reports 9.1 Web. doi:10.1038/s41598-019-53461-2.
Baek, Stephen, He, Yusen, Allen, Bryan G., Buatti, John M., Smith, Brian J., Tong, Ling, Sun, Zhiyu, Wu, Jia, Diehn, Maximilian, Loo, Billy W., Plichta, Kristin A., Seyedin, Steven N., Gannon, Maggie, Cabel, Katherine R., Kim, Yusung, & Wu, Xiaodong. Deep segmentation networks predict survival of non-small cell lung cancer. Scientific Reports, 9 (1). Retrieved from https://par.nsf.gov/biblio/10390829. https://doi.org/10.1038/s41598-019-53461-2
Baek, Stephen, He, Yusen, Allen, Bryan G., Buatti, John M., Smith, Brian J., Tong, Ling, Sun, Zhiyu, Wu, Jia, Diehn, Maximilian, Loo, Billy W., Plichta, Kristin A., Seyedin, Steven N., Gannon, Maggie, Cabel, Katherine R., Kim, Yusung, and Wu, Xiaodong.
"Deep segmentation networks predict survival of non-small cell lung cancer". Scientific Reports 9 (1). Country unknown/Code not available. https://doi.org/10.1038/s41598-019-53461-2.https://par.nsf.gov/biblio/10390829.
@article{osti_10390829,
place = {Country unknown/Code not available},
title = {Deep segmentation networks predict survival of non-small cell lung cancer},
url = {https://par.nsf.gov/biblio/10390829},
DOI = {10.1038/s41598-019-53461-2},
abstractNote = {Abstract Non-small-cell lung cancer (NSCLC) represents approximately 80–85% of lung cancer diagnoses and is the leading cause of cancer-related death worldwide. Recent studies indicate that image-based radiomics features from positron emission tomography/computed tomography (PET/CT) images have predictive power for NSCLC outcomes. To this end, easily calculated functional features such as the maximum and the mean of standard uptake value (SUV) and total lesion glycolysis (TLG) are most commonly used for NSCLC prognostication, but their prognostic value remains controversial. Meanwhile, convolutional neural networks (CNN) are rapidly emerging as a new method for cancer image analysis, with significantly enhanced predictive power compared to hand-crafted radiomics features. Here we show that CNNs trained to perform the tumor segmentation task, with no other information than physician contours, identify a rich set of survival-related image features with remarkable prognostic value. In a retrospective study on pre-treatment PET-CT images of 96 NSCLC patients before stereotactic-body radiotherapy (SBRT), we found that the CNN segmentation algorithm (U-Net) trained for tumor segmentation in PET and CT images, contained features having strong correlation with 2- and 5-year overall and disease-specific survivals. The U-Net algorithm has not seen any other clinical information (e.g. survival, age, smoking history, etc.) than the images and the corresponding tumor contours provided by physicians. In addition, we observed the same trend by validating the U-Net features against an extramural data set provided by Stanford Cancer Institute. Furthermore, through visualization of the U-Net, we also found convincing evidence that the regions of metastasis and recurrence appear to match with the regions where the U-Net features identified patterns that predicted higher likelihoods of death. We anticipate our findings will be a starting point for more sophisticated non-intrusive patient specific cancer prognosis determination. For example, the deep learned PET/CT features can not only predict survival but also visualize high-risk regions within or adjacent to the primary tumor and hence potentially impact therapeutic outcomes by optimal selection of therapeutic strategy or first-line therapy adjustment.},
journal = {Scientific Reports},
volume = {9},
number = {1},
author = {Baek, Stephen and He, Yusen and Allen, Bryan G. and Buatti, John M. and Smith, Brian J. and Tong, Ling and Sun, Zhiyu and Wu, Jia and Diehn, Maximilian and Loo, Billy W. and Plichta, Kristin A. and Seyedin, Steven N. and Gannon, Maggie and Cabel, Katherine R. and Kim, Yusung and Wu, Xiaodong},
}
Warning: Leaving National Science Foundation Website
You are now leaving the National Science Foundation website to go to a non-government website.
Website:
NSF takes no responsibility for and exercises no control over the views expressed or the accuracy of
the information contained on this site. Also be aware that NSF's privacy policy does not apply to this site.
An official website of the United States government
