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
. 2019:11383:92-104.
doi: 10.1007/978-3-030-11723-8_9. Epub 2019 Jan 26.

Multi-Institutional Deep Learning Modeling Without Sharing Patient Data: A Feasibility Study on Brain Tumor Segmentation

Micah J Sheller  1 , G Anthony Reina  1 , Brandon Edwards  1 , Jason Martin  1 , Spyridon Bakas  2
Affiliations

Multi-Institutional Deep Learning Modeling Without Sharing Patient Data: A Feasibility Study on Brain Tumor Segmentation

Micah J Sheller et al. Brainlesion. 2019.

Abstract

Deep learning models for semantic segmentation of images require large amounts of data. In the medical imaging domain, acquiring sufficient data is a significant challenge. Labeling medical image data requires expert knowledge. Collaboration between institutions could address this challenge, but sharing medical data to a centralized location faces various legal, privacy, technical, and data-ownership challenges, especially among international institutions. In this study, we introduce the first use of federated learning for multi-institutional collaboration, enabling deep learning modeling without sharing patient data. Our quantitative results demonstrate that the performance of federated semantic segmentation models (Dice=0.852) on multimodal brain scans is similar to that of models trained by sharing data (Dice=0.862). We compare federated learning with two alternative collaborative learning methods and find that they fail to match the performance of federated learning.

Keywords: BraTS; Deep Learning; Federated; Glioma; Incremental; Machine Learning; Segmentation.

PubMed Disclaimer

Figures

Fig. 1.
Fig. 1.
System Architecture of Federated Learning.
Fig. 2.
Fig. 2.
U-Net network diagram. Numbers above each layer indicate the number of channels in that layer. Note that the channel count (purple circle) differs from the original design [20] by a factor of 2.
Fig. 3.
Fig. 3.
(A) Validation DC scores over training epochs. The model peaked at 12 epochs and achieves a validation DC score of greater than 0.86. (B) Model performance on two images from the test set MRI. The model predicted mask closely matches the ground truth labels. An overlay of the ground truth with the original MRI slice.
Fig. 4.
Fig. 4.
Comparing centralized learning, FL, IIL and CIIL for the actual BraTS data distribution. The x-axis in (B) shows passes over the full dataset (epochs). Epochs are not equivalent in wall-clock time. The shading in (B) is min/max.
Fig. 5.
Fig. 5.
CIIL catastrophic forgetting: first institution’s training DC.
Fig. 6.
Fig. 6.
Comparing FL, CIIL and IIL for collaborations of 4–32 institutions. Note that 32 simulations have a different y-axis range.
Fig. 7.
Fig. 7.
FL and CIIL over round/cycle (1 epoch per). Confidence intervals are 0–100%. Note that 16 and 32 institutions are shown to 30 and 50 rounds, respectively.
Fig. 8.
Fig. 8.
FL over rounds for various EpR (16 and 32 institutions). Min/max shading.
See this image and copyright information in PMC

References

    1. Bakas S, Akbari H, Pisapia J, Martinez-Lage M, Rozycki M, Rathore S, Dahmane N, O’Rourke DM, Davatzikos C: In vivo detection of EGFRvIII in glioblastoma via perfusion magnetic resonance imaging signature consistent with deep peritumoral infiltration: the ∅-index. Clinical Cancer Research 23(16), 4724–4734 (2017) - PMC - PubMed
    1. Chang K, Bai HX, Zhou H, Su C, Bi WL, Agbodza E, Kavouridis VK, Senders JT, Boaro A, Beers A, Zhang B, Capellini A, Liao W, Shen Q, Li X, Xiao B, Cryan J, Ramkissoon S, Ramkissoon L, Ligon K, Wen PY, Bindra RS, Woo J, Arnaout O, Gerstner ER, Zhang PJ, Rosen BR, Yang L, Huang RY, Kalpathy-Cramer J: Residual Convolutional Neural Network for the Determination of ¡em¿IDH¡/em¿ Status in Low- and High-Grade Gliomas from MR Imaging. Clinical Cancer Research 24(5), 1073–1081 (2018) - PMC - PubMed
    1. Korfiatis P, Kline TL, Lachance DH, Parney IF, Buckner JC, Erickson BJ: Residual Deep Convolutional Neural Network Predicts MGMT Methylation Status. Journal of Digital Imaging 30(5), 622–628 (2017) - PMC - PubMed
    1. Akbari H, Macyszyn L, Da X, Bilello M, Wolf RL, Martinez-Lage M, Biros G, Alonso-Basanta M, O’Rourke DM, Davatzikos C: Imaging Surrogates of Infiltration Obtained Via Multiparametric Imaging Pattern Analysis Predict Subsequent Location of Recurrence of Glioblastoma. Neurosurgery 78(4), 572–580 (2016) - PMC - PubMed
    1. Macyszyn L, Akbari H, Pisapia JM, Da X, Attiah M, Pigrish V, Bi Y, Pal S, Davuluri RV, Roccograndi L, Dahmane N, Martinez-Lage M, Biros G, Wolf RL, Bilello M, O’Rourke DM, Davatzikos C: Imaging patterns predict patient survival and molecular subtype in glioblastoma via machine learning techniques. Neuro-Oncology 18(3), 417–425 (2016) - PMC - PubMed

LinkOut - more resources