Abstract
Background: Cortical multiple sclerosis lesions are clinically relevant but inconspicuous at conventional clinical MRI. Double inversion recovery (DIR) and phase-sensitive inversion recovery (PSIR) are more sensitive but often unavailable. In the past 2 years, artificial intelligence (AI) was used to generate DIR and PSIR from standard clinical sequences (eg, T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery sequences), but multicenter validation is crucial for further implementation. Purpose: To evaluate cortical and juxtacortical multiple sclerosis lesion detection for diagnostic and disease monitoring purposes on AI-generated DIR and PSIR images compared with MRI-acquired DIR and PSIR images in a multicenter setting. Materials and Methods: Generative adversarial networks were used to generate AI-based DIR (n = 50) and PSIR (n = 43) images. The number of detected lesions between AI-generated images and MRI-acquired (reference) images was compared by randomized blinded scoring by seven readers (all with >10 years of experience in lesion assessment). Reliability was expressed as the intraclass correlation coefficient (ICC). Differences in lesion subtype were determined using Wilcoxon signed-rank tests. Results: MRI scans of 202 patients with multiple sclerosis (mean age, 46 years ± 11 [SD]; 127 women) were retrospectively collected from seven centers (February 2020 to January 2021). In total, 1154 lesions were detected on AI-generated DIR images versus 855 on MRI-acquired DIR images (mean difference per reader, 35.0% ± 22.8; P < .001). On AI-generated PSIR images, 803 lesions were detected versus 814 on MRI-acquired PSIR images (98.9% ± 19.4; P = .87). Reliability was good for both DIR (ICC, 0.81) and PSIR (ICC, 0.75) across centers. Regionally, more juxtacortical lesions were detected on AI-generated DIR images than on MRI-acquired DIR images (495 [42.9%] vs 338 [39.5%]; P < .001). On AI-generated PSIR images, fewer juxtacortical lesions were detected than on MRI-acquired PSIR images (232 [28.9%] vs 282 [34.6%]; P = .02). Conclusion: Artificial intelligence–generated double inversion-recovery and phase-sensitive inversion-recovery images performed well compared with their MRI-acquired counterparts and can be considered reliable in a multicenter setting, with good between-reader and between-center interpretative agreement.
| Original language | English |
|---|---|
| Article number | e221425 |
| Journal | Radiology |
| Volume | 307 |
| Issue number | 2 |
| DOIs | |
| Publication status | Published - 1 Apr 2023 |
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver
Bouman, P. M., Noteboom, S., Nobrega Santos, F. A., Beck, E. S., Bliault, G., Castellaro, M., Calabrese, M., Chard, D. T., Eichinger, P., Filippi, M., Inglese, M., Lapucci, C., Marciniak, A., Moraal, B., Pinzon, A. M., Mühlau, M., Preziosa, P., Reich, D. S., Rocca, M. A., ... Steenwijk, M. D. (2023). Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection. Radiology, 307(2), [e221425]. https://doi.org/10.1148/radiol.221425
Bouman, Piet M. ; Noteboom, Samantha ; Nobrega Santos, Fernando A. et al. / Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection. In: Radiology. 2023 ; Vol. 307, No. 2.
@article{7a83468a18ea4243857b0c2b6f4e7e31,
title = "Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection",
abstract = "Background: Cortical multiple sclerosis lesions are clinically relevant but inconspicuous at conventional clinical MRI. Double inversion recovery (DIR) and phase-sensitive inversion recovery (PSIR) are more sensitive but often unavailable. In the past 2 years, artificial intelligence (AI) was used to generate DIR and PSIR from standard clinical sequences (eg, T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery sequences), but multicenter validation is crucial for further implementation. Purpose: To evaluate cortical and juxtacortical multiple sclerosis lesion detection for diagnostic and disease monitoring purposes on AI-generated DIR and PSIR images compared with MRI-acquired DIR and PSIR images in a multicenter setting. Materials and Methods: Generative adversarial networks were used to generate AI-based DIR (n = 50) and PSIR (n = 43) images. The number of detected lesions between AI-generated images and MRI-acquired (reference) images was compared by randomized blinded scoring by seven readers (all with >10 years of experience in lesion assessment). Reliability was expressed as the intraclass correlation coefficient (ICC). Differences in lesion subtype were determined using Wilcoxon signed-rank tests. Results: MRI scans of 202 patients with multiple sclerosis (mean age, 46 years ± 11 [SD]; 127 women) were retrospectively collected from seven centers (February 2020 to January 2021). In total, 1154 lesions were detected on AI-generated DIR images versus 855 on MRI-acquired DIR images (mean difference per reader, 35.0% ± 22.8; P < .001). On AI-generated PSIR images, 803 lesions were detected versus 814 on MRI-acquired PSIR images (98.9% ± 19.4; P = .87). Reliability was good for both DIR (ICC, 0.81) and PSIR (ICC, 0.75) across centers. Regionally, more juxtacortical lesions were detected on AI-generated DIR images than on MRI-acquired DIR images (495 [42.9%] vs 338 [39.5%]; P < .001). On AI-generated PSIR images, fewer juxtacortical lesions were detected than on MRI-acquired PSIR images (232 [28.9%] vs 282 [34.6%]; P = .02). Conclusion: Artificial intelligence–generated double inversion-recovery and phase-sensitive inversion-recovery images performed well compared with their MRI-acquired counterparts and can be considered reliable in a multicenter setting, with good between-reader and between-center interpretative agreement.",
author = "Bouman, {Piet M.} and Samantha Noteboom and {Nobrega Santos}, {Fernando A.} and Beck, {Erin S.} and Gregory Bliault and Marco Castellaro and Massimiliano Calabrese and Chard, {Declan T.} and Paul Eichinger and Massimo Filippi and Matilde Inglese and Caterina Lapucci and Andrzej Marciniak and Bastiaan Moraal and Pinzon, {Alfredo Morales} and Mark M{\"u}hlau and Paolo Preziosa and Reich, {Daniel S.} and Rocca, {Maria A.} and Schoonheim, {Menno M.} and Twisk, {Jos W. R.} and Benedict Wiestler and Jonkman, {Laura E.} and Guttmann, {Charles R. G.} and Geurts, {Jeroen J. G.} and Steenwijk, {Martijn D.}",
note = "Funding Information: Supported by Stichting MS Research (Dutch MS Research Foundation) (grant 19-049). Development of the SPINE platform was supported in part by the International Progressive MS Alliance (award reference number PA-1603-08175), as well as the Bordeaux University Foundation through donations from Roche Pharmaceuticals and Talan. Publisher Copyright: {\textcopyright} 2023 Radiological Society of North America Inc.. All rights reserved.",
year = "2023",
month = apr,
day = "1",
doi = "10.1148/radiol.221425",
language = "English",
volume = "307",
journal = "Radiology Now",
issn = "0033-8419",
publisher = "Radiological Society of North America Inc.",
number = "2",
}
Bouman, PM, Noteboom, S, Nobrega Santos, FA, Beck, ES, Bliault, G, Castellaro, M, Calabrese, M, Chard, DT, Eichinger, P, Filippi, M, Inglese, M, Lapucci, C, Marciniak, A, Moraal, B, Pinzon, AM, Mühlau, M, Preziosa, P, Reich, DS, Rocca, MA, Schoonheim, MM, Twisk, JWR, Wiestler, B, Jonkman, LE, Guttmann, CRG, Geurts, JJG 2023, 'Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection', Radiology, vol. 307, no. 2, e221425. https://doi.org/10.1148/radiol.221425
Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection. / Bouman, Piet M.; Noteboom, Samantha; Nobrega Santos, Fernando A. et al.
In: Radiology, Vol. 307, No. 2, e221425, 01.04.2023.
Research output: Contribution to journal › Article › Academic › peer-review
TY - JOUR
T1 - Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection
AU - Bouman, Piet M.
AU - Noteboom, Samantha
AU - Nobrega Santos, Fernando A.
AU - Beck, Erin S.
AU - Bliault, Gregory
AU - Castellaro, Marco
AU - Calabrese, Massimiliano
AU - Chard, Declan T.
AU - Eichinger, Paul
AU - Filippi, Massimo
AU - Inglese, Matilde
AU - Lapucci, Caterina
AU - Marciniak, Andrzej
AU - Moraal, Bastiaan
AU - Pinzon, Alfredo Morales
AU - Mühlau, Mark
AU - Preziosa, Paolo
AU - Reich, Daniel S.
AU - Rocca, Maria A.
AU - Schoonheim, Menno M.
AU - Twisk, Jos W. R.
AU - Wiestler, Benedict
AU - Jonkman, Laura E.
AU - Guttmann, Charles R. G.
AU - Geurts, Jeroen J. G.
AU - Steenwijk, Martijn D.
N1 - Funding Information:Supported by Stichting MS Research (Dutch MS Research Foundation) (grant 19-049). Development of the SPINE platform was supported in part by the International Progressive MS Alliance (award reference number PA-1603-08175), as well as the Bordeaux University Foundation through donations from Roche Pharmaceuticals and Talan.Publisher Copyright:© 2023 Radiological Society of North America Inc.. All rights reserved.
PY - 2023/4/1
Y1 - 2023/4/1
N2 - Background: Cortical multiple sclerosis lesions are clinically relevant but inconspicuous at conventional clinical MRI. Double inversion recovery (DIR) and phase-sensitive inversion recovery (PSIR) are more sensitive but often unavailable. In the past 2 years, artificial intelligence (AI) was used to generate DIR and PSIR from standard clinical sequences (eg, T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery sequences), but multicenter validation is crucial for further implementation. Purpose: To evaluate cortical and juxtacortical multiple sclerosis lesion detection for diagnostic and disease monitoring purposes on AI-generated DIR and PSIR images compared with MRI-acquired DIR and PSIR images in a multicenter setting. Materials and Methods: Generative adversarial networks were used to generate AI-based DIR (n = 50) and PSIR (n = 43) images. The number of detected lesions between AI-generated images and MRI-acquired (reference) images was compared by randomized blinded scoring by seven readers (all with >10 years of experience in lesion assessment). Reliability was expressed as the intraclass correlation coefficient (ICC). Differences in lesion subtype were determined using Wilcoxon signed-rank tests. Results: MRI scans of 202 patients with multiple sclerosis (mean age, 46 years ± 11 [SD]; 127 women) were retrospectively collected from seven centers (February 2020 to January 2021). In total, 1154 lesions were detected on AI-generated DIR images versus 855 on MRI-acquired DIR images (mean difference per reader, 35.0% ± 22.8; P < .001). On AI-generated PSIR images, 803 lesions were detected versus 814 on MRI-acquired PSIR images (98.9% ± 19.4; P = .87). Reliability was good for both DIR (ICC, 0.81) and PSIR (ICC, 0.75) across centers. Regionally, more juxtacortical lesions were detected on AI-generated DIR images than on MRI-acquired DIR images (495 [42.9%] vs 338 [39.5%]; P < .001). On AI-generated PSIR images, fewer juxtacortical lesions were detected than on MRI-acquired PSIR images (232 [28.9%] vs 282 [34.6%]; P = .02). Conclusion: Artificial intelligence–generated double inversion-recovery and phase-sensitive inversion-recovery images performed well compared with their MRI-acquired counterparts and can be considered reliable in a multicenter setting, with good between-reader and between-center interpretative agreement.
AB - Background: Cortical multiple sclerosis lesions are clinically relevant but inconspicuous at conventional clinical MRI. Double inversion recovery (DIR) and phase-sensitive inversion recovery (PSIR) are more sensitive but often unavailable. In the past 2 years, artificial intelligence (AI) was used to generate DIR and PSIR from standard clinical sequences (eg, T1-weighted, T2-weighted, and fluid-attenuated inversion-recovery sequences), but multicenter validation is crucial for further implementation. Purpose: To evaluate cortical and juxtacortical multiple sclerosis lesion detection for diagnostic and disease monitoring purposes on AI-generated DIR and PSIR images compared with MRI-acquired DIR and PSIR images in a multicenter setting. Materials and Methods: Generative adversarial networks were used to generate AI-based DIR (n = 50) and PSIR (n = 43) images. The number of detected lesions between AI-generated images and MRI-acquired (reference) images was compared by randomized blinded scoring by seven readers (all with >10 years of experience in lesion assessment). Reliability was expressed as the intraclass correlation coefficient (ICC). Differences in lesion subtype were determined using Wilcoxon signed-rank tests. Results: MRI scans of 202 patients with multiple sclerosis (mean age, 46 years ± 11 [SD]; 127 women) were retrospectively collected from seven centers (February 2020 to January 2021). In total, 1154 lesions were detected on AI-generated DIR images versus 855 on MRI-acquired DIR images (mean difference per reader, 35.0% ± 22.8; P < .001). On AI-generated PSIR images, 803 lesions were detected versus 814 on MRI-acquired PSIR images (98.9% ± 19.4; P = .87). Reliability was good for both DIR (ICC, 0.81) and PSIR (ICC, 0.75) across centers. Regionally, more juxtacortical lesions were detected on AI-generated DIR images than on MRI-acquired DIR images (495 [42.9%] vs 338 [39.5%]; P < .001). On AI-generated PSIR images, fewer juxtacortical lesions were detected than on MRI-acquired PSIR images (232 [28.9%] vs 282 [34.6%]; P = .02). Conclusion: Artificial intelligence–generated double inversion-recovery and phase-sensitive inversion-recovery images performed well compared with their MRI-acquired counterparts and can be considered reliable in a multicenter setting, with good between-reader and between-center interpretative agreement.
UR - http://www.scopus.com/inward/record.url?scp=85153930451&partnerID=8YFLogxK
U2 - 10.1148/radiol.221425
DO - 10.1148/radiol.221425
M3 - Article
C2 - 36749211
SN - 0033-8419
VL - 307
JO - Radiology Now
JF - Radiology Now
IS - 2
M1 - e221425
ER -
Bouman PM, Noteboom S, Nobrega Santos FA, Beck ES, Bliault G, Castellaro M et al. Multicenter Evaluation of AI-generated DIR and PSIR for Cortical and Juxtacortical Multiple Sclerosis Lesion Detection. Radiology. 2023 Apr 1;307(2):e221425. doi: 10.1148/radiol.221425