Abstract
Final lesion volume (FLV) is a surrogate outcome measure in anterior circulation stroke (ACS). In posterior circulation stroke (PCS), this relation is plausibly understudied due to a lack of methods that automatically quantify FLV. The applicability of deep learning approaches to PCS is limited due to its lower incidence compared to ACS. We evaluated strategies to develop a convolutional neural network (CNN) for PCS lesion segmentation by using image data from both ACS and PCS patients. We included follow-up non-contrast computed tomography scans of 1018 patients with ACS and 107 patients with PCS. To assess whether an ACS lesion segmentation generalizes to PCS, a CNN was trained on ACS data (ACS-CNN). Second, to evaluate the performance of only including PCS patients, a CNN was trained on PCS data. Third, to evaluate the performance when combining the datasets, a CNN was trained on both datasets. Finally, to evaluate the performance of transfer learning, the ACS-CNN was fine-tuned using PCS patients. The transfer learning strategy outperformed the other strategies in volume agreement with an intra-class correlation of 0.88 (95% CI: 0.83-0.92) vs. 0.55 to 0.83 and a lesion detection rate of 87% vs. 41-77 for the other strategies. Hence, transfer learning improved the FLV quantification and detection rate of PCS lesions compared to the other strategies.
Original language | English |
---|---|
Article number | 1621 |
Journal | Diagnostics |
Volume | 11 |
Issue number | 9 |
DOIs | |
Publication status | Published - 4 Sept 2021 |
Access to Document
Other files and links
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver
}
In: Diagnostics, Vol. 11, No. 9, 1621, 04.09.2021.
Research output: Contribution to journal › Article › Academic › peer-review
TY - JOUR
T1 - Automated final lesion segmentation in posterior circulation acute ischemic stroke using deep learning
AU - Zoetmulder, Riaan
AU - Konduri, Praneeta R
AU - Obdeijn, Iris V
AU - Gavves, Efstratios
AU - Išgum, Ivana
AU - Majoie, Charles B L M
AU - Dippel, Diederik W J
AU - Roos, Yvo B W E M
AU - Goyal, Mayank
AU - Mitchell, Peter J
AU - Campbell, Bruce C V
AU - Lopes, Demetrius K
AU - Reimann, Gernot
AU - Jovin, Tudor G
AU - Saver, Jeffrey L
AU - Muir, Keith W
AU - White, Phil
AU - Bracard, Serge
AU - Chen, Bailiang
AU - Brown, Scott
AU - Schonewille, Wouter J
AU - van der Hoeven, Erik
AU - Puetz, Volker
AU - Marquering, Henk A
N1 - Funding Information: Conflicts of Interest: P. Konduri is funded by INSIST (www.insist-h2020.eu): a European Union’s Horizon 2020 research and innovation program (grant agreement number:777072). H.A. Marquer-ing is a cofounder and shareholder of Nico.Lab. I. Išgum is a cofounder, scientific lead, and share-holderofQuantib-U.C.B.L.M.Majoieisashareholder ofNico.Labandhas receivedspeakers’ bureau fees from Stryker (paid to institution). T.G. Jovin is a consultant for Stryker Neurovascular (principal investigator, DAWN), Cerenovus, has an ownership interest in Anaconda, Corindus, Viz.ai, and is onthe Advisory Board and an investor in FreeOxBiotech,Route92 and Blockade Medical. J.L. Saver receivesconsulting fees from and serves on a steering committee for Stryker, Med-tronic, Cerenovus, and Rapid Medical and has an institutional conflict of interest with The Univer-sityofCalifornia for patentrights in retrieval devices for stroke. M.Goyal has a patent for systems and methods for diagnosing strokes (PCT/CA2013/000761) licensed to GE Healthcare. D.W.J. Dippel 12 of 15 received honoraria fromStryker (paid toinstitution). J.Mitchellreceivedgrant funding for theEX- TEND-IA trial to the Florey Institute ofNeuroscience and Mental Healthfrom Covidien (Med-tronic), has served asan unpaid consultant toCodman Johnson & Johnson,and his organization has receivedresearch fundingand grantsfromCodman Johnson & Johnson, Medtronic,and Stryker. B.C.V. Campbell has received research support from the National Health and Medical Research Council of Australia (GNT1043242, GNT1035688), Royal Australasian College of Physicians, Royal Melbourne Hospital Foundation, National Heart Foundation, National Stroke Foundation of Aus-Foundation of Australia, and unrestricted grant funding for the EXTEND-IA trial to the Florey tralia, and unrestricted grant funding for the EXTEND-IAtrial tothe Florey Institute of Neurosci-Institute of Neuroscience and Mental Health from Covidien (Medtronic). ence and Mental Health from Covidien (Medtronic). Funding Information: Conflicts of Interest: P. Konduri is funded by INSIST (www.insist-h2020.eu, accessed on 30 June 2021): a European Union’s Horizon 2020 research and innovation program (grant agreement number: 777072). H.A. Marquering is a cofounder and shareholder of Nico.Lab. I. Išgum is a cofounder, scientific lead, and shareholder of Quantib-U. C.B.L.M. Majoie is a shareholder of Nico.Lab and has received speakers’ bureau fees from Stryker (paid to institution). T.G. Jovin is a consultant for Stryker Neurovascular (principal investigator, DAWN), Cerenovus, has an ownership interest in Anaconda, Corindus, Viz.ai, and is on the Advisory Board and an investor in FreeOx Biotech, Route92 and Blockade Medical. J.L. Saver receives consulting fees from and serves on a steering committee for Stryker, Medtronic, Cerenovus, and Rapid Medical and has an institutional conflict of interest with The University of California for patent rights in retrieval devices for stroke. M. Goyal has a patent for systems and methods for diagnosing strokes (PCT/CA2013/000761) licensed to GE Healthcare. D.W.J. Dippel received honoraria from Stryker (paid to institution). J. Mitchell received grant funding for the EXTEND-IA trial to the Florey Institute of Neuroscience and Mental Health from Covidien (Medtronic), has served as an unpaid consultant to Codman Johnson & Johnson, Funding Information: Acknowledgments: We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Quadro P6000 GPU used for this research. Publisher Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland.
PY - 2021/9/4
Y1 - 2021/9/4
N2 - Final lesion volume (FLV) is a surrogate outcome measure in anterior circulation stroke (ACS). In posterior circulation stroke (PCS), this relation is plausibly understudied due to a lack of methods that automatically quantify FLV. The applicability of deep learning approaches to PCS is limited due to its lower incidence compared to ACS. We evaluated strategies to develop a convolutional neural network (CNN) for PCS lesion segmentation by using image data from both ACS and PCS patients. We included follow-up non-contrast computed tomography scans of 1018 patients with ACS and 107 patients with PCS. To assess whether an ACS lesion segmentation generalizes to PCS, a CNN was trained on ACS data (ACS-CNN). Second, to evaluate the performance of only including PCS patients, a CNN was trained on PCS data. Third, to evaluate the performance when combining the datasets, a CNN was trained on both datasets. Finally, to evaluate the performance of transfer learning, the ACS-CNN was fine-tuned using PCS patients. The transfer learning strategy outperformed the other strategies in volume agreement with an intra-class correlation of 0.88 (95% CI: 0.83-0.92) vs. 0.55 to 0.83 and a lesion detection rate of 87% vs. 41-77 for the other strategies. Hence, transfer learning improved the FLV quantification and detection rate of PCS lesions compared to the other strategies.
AB - Final lesion volume (FLV) is a surrogate outcome measure in anterior circulation stroke (ACS). In posterior circulation stroke (PCS), this relation is plausibly understudied due to a lack of methods that automatically quantify FLV. The applicability of deep learning approaches to PCS is limited due to its lower incidence compared to ACS. We evaluated strategies to develop a convolutional neural network (CNN) for PCS lesion segmentation by using image data from both ACS and PCS patients. We included follow-up non-contrast computed tomography scans of 1018 patients with ACS and 107 patients with PCS. To assess whether an ACS lesion segmentation generalizes to PCS, a CNN was trained on ACS data (ACS-CNN). Second, to evaluate the performance of only including PCS patients, a CNN was trained on PCS data. Third, to evaluate the performance when combining the datasets, a CNN was trained on both datasets. Finally, to evaluate the performance of transfer learning, the ACS-CNN was fine-tuned using PCS patients. The transfer learning strategy outperformed the other strategies in volume agreement with an intra-class correlation of 0.88 (95% CI: 0.83-0.92) vs. 0.55 to 0.83 and a lesion detection rate of 87% vs. 41-77 for the other strategies. Hence, transfer learning improved the FLV quantification and detection rate of PCS lesions compared to the other strategies.
UR - http://www.scopus.com/inward/record.url?scp=85114608266&partnerID=8YFLogxK
U2 - https://doi.org/10.3390/diagnostics11091621
DO - https://doi.org/10.3390/diagnostics11091621
M3 - Article
C2 - 34573963
SN - 2075-4418
VL - 11
JO - Diagnostics
JF - Diagnostics
IS - 9
M1 - 1621
ER -