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
. 2020 Nov 23;10(1):20331.
doi: 10.1038/s41598-020-77389-0.

Discriminating pseudoprogression and true progression in diffuse infiltrating glioma using multi-parametric MRI data through deep learning

Joonsang Lee  1 , Nicholas Wang  1 , Sevcan Turk  2 , Shariq Mohammed  1 , Remy Lobo  2 , John Kim  2 , Eric Liao  2 , Sandra Camelo-Piragua  3 , Michelle Kim  4 , Larry Junck  5 , Jayapalli Bapuraj  2 , Ashok Srinivasan  2 , Arvind Rao  6
Affiliations

Discriminating pseudoprogression and true progression in diffuse infiltrating glioma using multi-parametric MRI data through deep learning

Joonsang Lee et al. Sci Rep. .

Abstract

Differentiating pseudoprogression from true tumor progression has become a significant challenge in follow-up of diffuse infiltrating gliomas, particularly high grade, which leads to a potential treatment delay for patients with early glioma recurrence. In this study, we proposed to use a multiparametric MRI data as a sequence input for the convolutional neural network with the recurrent neural network based deep learning structure to discriminate between pseudoprogression and true tumor progression. In this study, 43 biopsy-proven patient data identified as diffuse infiltrating glioma patients whose disease progressed/recurred were used. The dataset consists of five original MRI sequences; pre-contrast T1-weighted, post-contrast T1-weighted, T2-weighted, FLAIR, and ADC images as well as two engineered sequences; T1post-T1pre and T2-FLAIR. Next, we used three CNN-LSTM models with a different set of sequences as input sequences to pass through CNN-LSTM layers. We performed threefold cross-validation in the training dataset and generated the boxplot, accuracy, and ROC curve, AUC from each trained model with the test dataset to evaluate models. The mean accuracy for VGG16 models ranged from 0.44 to 0.60 and the mean AUC ranged from 0.47 to 0.59. For CNN-LSTM model, the mean accuracy ranged from 0.62 to 0.75 and the mean AUC ranged from 0.64 to 0.81. The performance of the proposed CNN-LSTM with multiparametric sequence data was found to outperform the popular convolutional CNN with a single MRI sequence. In conclusion, incorporating all available MRI sequences into a sequence input for a CNN-LSTM model improved diagnostic performance for discriminating between pseudoprogression and true tumor progression.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
ROC curves for VGG16 deep learning models with each MRI sequence for prediction of PsP from TP. Each ROC curve for each modality obtained from threefold cross-validation (CV1, CV2, and CV3). The x-axis is the true negative rate (TNR) or specificity and the y-axis is true positive rate (TPR) or sensitivity. The mean AUC, area under ROC curve, values were (a) 0.53 for pre T1-weighted, (b) 0.49 for post T1-weighted, (c) 0.51 for T2-weighted, (d) 0.55 for FLAIR, (e) 0.47 for ADC map, (f) 0.59 for post T1–pre T1, (g) 0.54 for T2–FLAIR images.
Figure 2
Figure 2
ROC curves for CNN-LSTM deep learning models with a different set of sequences for prediction of PsP from TP. Each ROC curve for each modality obtained from threefold cross-validation (CV1, CV2, and CV3). The x-axis is the true negative rate (TNR) or specificity and the y-axis is true positive rate (TPR) or sensitivity. The mean AUC, area under ROC curve, values were (a) 0.64 for a set of 3 modalities, (b) 0.69 for a set of 5 modalities, and (c) 0.81 for a set of 7 modalities.
Figure 3
Figure 3
Boxplots of AUC for VGG16 and CNN-LSTM models. The first seven box plots are for VGG16 with an individual MRI sequence, and last three box plots are for CNN-LSTM with multiparametric MRI sequences. Statistics were collected from threefold cross validation and the red lines in the boxplots represent the median values.
Figure 4
Figure 4
A GBM patient (PsP) with various MRI scans such as (A) pre-contrast T1-weighted, (B) post-contrast T1-weighted, (C) T2-weighted, (D) Fluid-Attenuated Inversion Recovery (FLAIR), and (E) ADC images. In this study, we included two engineered modalities (F) T1post–T1pre and (G) T2–FLAIR. And (H) the region of interest (ROI) is shown on a post-contrast T1 image.
Figure 5
Figure 5
CNN-LSTM architecture with a set of multiparametric MRI sequence input consisting of (1) T1pre, (2) T1post, (3) T2, (4) FLAIR, (5) ADC map, (6) T1post–T1pre, and (7) T2–FLAIR.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Vigneswaran K, Neill S, Hadjipanayis CG. Beyond the World Health Organization grading of infiltrating gliomas: advances in the molecular genetics of glioma classification. Ann. Transl. Med. 2015;3:95. doi: 10.3978/j.issn.2305-5839.2015.03.57. - DOI - PMC - PubMed
    1. Bleeker FE, Molenaar RJ, Leenstra S. Recent advances in the molecular understanding of glioblastoma. J. Neurooncol. 2012;108:11–27. doi: 10.1007/s11060-011-0793-0. - DOI - PMC - PubMed
    1. Ostrom QT, et al. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2009–2013. Neuro Oncol. 2016;18:v1–v75. doi: 10.1093/neuonc/now207. - DOI - PMC - PubMed
    1. Ghosh M, et al. Survival and prognostic factors for glioblastoma multiforme: retrospective single-institutional study. Indian J. Cancer. 2017;54:362–367. doi: 10.4103/ijc.IJC_157_17. - DOI - PubMed
    1. van den Bent MJ, Smits M, Kros JM, Chang SM. Diffuse infiltrating oligodendroglioma and astrocytoma. J. Clin. Oncol. 2017;35:2394–2401. doi: 10.1200/JCO.2017.72.6737. - DOI - PubMed

Publication types

MeSH terms