Abstract
Machine learning approaches using structural magnetic resonance imaging (sMRI) can be informative for disease classification, although their ability to predict psychosis is largely unknown. We created a model with individuals at CHR who developed psychosis later (CHR-PS+) from healthy controls (HCs) that can differentiate each other. We also evaluated whether we could distinguish CHR-PS+ individuals from those who did not develop psychosis later (CHR-PS-) and those with uncertain follow-up status (CHR-UNK). T1-weighted structural brain MRI scans from 1165 individuals at CHR (CHR-PS+, n = 144; CHR-PS-, n = 793; and CHR-UNK, n = 228), and 1029 HCs, were obtained from 21 sites. We used ComBat to harmonize measures of subcortical volume, cortical thickness and surface area data and corrected for non-linear effects of age and sex using a general additive model. CHR-PS+ (n = 120) and HC (n = 799) data from 20 sites served as a training dataset, which we used to build a classifier. The remaining samples were used external validation datasets to evaluate classifier performance (test, independent confirmatory, and independent group [CHR-PS- and CHR-UNK] datasets). The accuracy of the classifier on the training and independent confirmatory datasets was 85% and 73% respectively. Regional cortical surface area measures-including those from the right superior frontal, right superior temporal, and bilateral insular cortices strongly contributed to classifying CHR-PS+ from HC. CHR-PS- and CHR-UNK individuals were more likely to be classified as HC compared to CHR-PS+ (classification rate to HC: CHR-PS+, 30%; CHR-PS-, 73%; CHR-UNK, 80%). We used multisite sMRI to train a classifier to predict psychosis onset in CHR individuals, and it showed promise predicting CHR-PS+ in an independent sample. The results suggest that when considering adolescent brain development, baseline MRI scans for CHR individuals may be helpful to identify their prognosis. Future prospective studies are required about whether the classifier could be actually helpful in the clinical settings.
Original language | English (US) |
---|---|
Journal | Molecular psychiatry |
DOIs | |
State | Accepted/In press - 2024 |
Bibliographical note
Publisher Copyright:
© The Author(s) 2024.
Publisher link
Other files and links
Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS
Zhu, Y., Maikusa, N., Radua, J., Sämann, P. G., Fusar-Poli, P., Agartz, I., Andreassen, O. A., Bachman, P., Baeza, I., Chen, X., Choi, S., Corcoran, C. M., Ebdrup, B. H., Fortea, A., Garani, R. R. G., Glenthøj, B. Y., Glenthøj, L. B., Haas, S. S., Hamilton, H. K., ... Yuan, L. (Accepted/In press). Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk. Molecular psychiatry. https://doi.org/10.1038/s41380-024-02426-7
Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk. / Zhu, Yinghan; Maikusa, Norihide; Radua, Joaquim et al.
In: Molecular psychiatry, 2024.
Research output: Contribution to journal › Article › peer-review
Zhu, Y, Maikusa, N, Radua, J, Sämann, PG, Fusar-Poli, P, Agartz, I, Andreassen, OA, Bachman, P, Baeza, I, Chen, X, Choi, S, Corcoran, CM, Ebdrup, BH, Fortea, A, Garani, RRG, Glenthøj, BY, Glenthøj, LB, Haas, SS, Hamilton, HK, Hayes, RA, He, Y, Heekeren, K, Kasai, K, Katagiri, N, Kim, M, Kristensen, TD, Kwon, JS, Lawrie, SM, Lebedeva, I, Lee, J, Loewy, RL, Mathalon, DH, McGuire, P, Mizrahi, R, Mizuno, M, Møller, P, Nemoto, T, Nordholm, D, Omelchenko, MA, Raghava, JM, Røssberg, JI, Rössler, W, Salisbury, DF, Sasabayashi, D, Smigielski, L, Sugranyes, G, Takahashi, T, Tamnes, CK, Tang, J, Theodoridou, A, Tomyshev, AS, Uhlhaas, PJ, Værnes, TG, van Amelsvoort, TAMJ, Waltz, JA, Westlye, LT, Zhou, JH, Thompson, PM, Hernaus, D, Jalbrzikowski, M, Koike, S, Allen, P, Baldwin, H, Catalano, S, Chee, MWL, Cho, KIK, de Haan, L, Horton, LE, Klaunig, MJ, Bin Kwak, Y, Ma, X, Nordentoft, M, Ouyang, L, Pariente, JC, Resch, F, Schiffman, J, Sørensen, ME, Suzuki, M, Vinogradov, S, Wenneberg, C, Yamasue, H & Yuan, L 2024, 'Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk', Molecular psychiatry. https://doi.org/10.1038/s41380-024-02426-7
Zhu Y, Maikusa N, Radua J, Sämann PG, Fusar-Poli P, Agartz I et al. Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk. Molecular psychiatry. 2024. doi: 10.1038/s41380-024-02426-7
Zhu, Yinghan ; Maikusa, Norihide ; Radua, Joaquim et al. / Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk. In: Molecular psychiatry. 2024.
@article{55ba85ec8dd1416eb5afccb7ca67e8b0,
title = "Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk",
abstract = "Machine learning approaches using structural magnetic resonance imaging (sMRI) can be informative for disease classification, although their ability to predict psychosis is largely unknown. We created a model with individuals at CHR who developed psychosis later (CHR-PS+) from healthy controls (HCs) that can differentiate each other. We also evaluated whether we could distinguish CHR-PS+ individuals from those who did not develop psychosis later (CHR-PS-) and those with uncertain follow-up status (CHR-UNK). T1-weighted structural brain MRI scans from 1165 individuals at CHR (CHR-PS+, n = 144; CHR-PS-, n = 793; and CHR-UNK, n = 228), and 1029 HCs, were obtained from 21 sites. We used ComBat to harmonize measures of subcortical volume, cortical thickness and surface area data and corrected for non-linear effects of age and sex using a general additive model. CHR-PS+ (n = 120) and HC (n = 799) data from 20 sites served as a training dataset, which we used to build a classifier. The remaining samples were used external validation datasets to evaluate classifier performance (test, independent confirmatory, and independent group [CHR-PS- and CHR-UNK] datasets). The accuracy of the classifier on the training and independent confirmatory datasets was 85% and 73% respectively. Regional cortical surface area measures-including those from the right superior frontal, right superior temporal, and bilateral insular cortices strongly contributed to classifying CHR-PS+ from HC. CHR-PS- and CHR-UNK individuals were more likely to be classified as HC compared to CHR-PS+ (classification rate to HC: CHR-PS+, 30%; CHR-PS-, 73%; CHR-UNK, 80%). We used multisite sMRI to train a classifier to predict psychosis onset in CHR individuals, and it showed promise predicting CHR-PS+ in an independent sample. The results suggest that when considering adolescent brain development, baseline MRI scans for CHR individuals may be helpful to identify their prognosis. Future prospective studies are required about whether the classifier could be actually helpful in the clinical settings.",
author = "Yinghan Zhu and Norihide Maikusa and Joaquim Radua and S{\"a}mann, {Philipp G.} and Paolo Fusar-Poli and Ingrid Agartz and Andreassen, {Ole A.} and Peter Bachman and Inmaculada Baeza and Xiaogang Chen and Sunah Choi and Corcoran, {Cheryl M.} and Ebdrup, {Bj{\o}rn H.} and Adriana Fortea and Garani, {Ranjini R.G.} and Glenth{\o}j, {Birte Yding} and Glenth{\o}j, {Louise Birkedal} and Haas, {Shalaila S.} and Hamilton, {Holly K.} and Hayes, {Rebecca A.} and Ying He and Karsten Heekeren and Kiyoto Kasai and Naoyuki Katagiri and Minah Kim and Kristensen, {Tina D.} and Kwon, {Jun Soo} and Lawrie, {Stephen M.} and Irina Lebedeva and Jimmy Lee and Loewy, {Rachel L.} and Mathalon, {Daniel H.} and Philip McGuire and Romina Mizrahi and Masafumi Mizuno and Paul M{\o}ller and Takahiro Nemoto and Dorte Nordholm and Omelchenko, {Maria A.} and Raghava, {Jayachandra M.} and R{\o}ssberg, {Jan I.} and Wulf R{\"o}ssler and Salisbury, {Dean F.} and Daiki Sasabayashi and Lukasz Smigielski and Gisela Sugranyes and Tsutomu Takahashi and Tamnes, {Christian K.} and Jinsong Tang and Anastasia Theodoridou and Tomyshev, {Alexander S.} and Uhlhaas, {Peter J.} and V{\ae}rnes, {Tor G.} and {van Amelsvoort}, {Therese A.M.J.} and Waltz, {James A.} and Westlye, {Lars T.} and Zhou, {Juan H.} and Thompson, {Paul M.} and Dennis Hernaus and Maria Jalbrzikowski and Shinsuke Koike and Paul Allen and Helen Baldwin and Sabrina Catalano and Chee, {Michael W.L.} and Cho, {Kang Ik K.} and {de Haan}, Lieuwe and Horton, {Leslie E.} and Klaunig, {Mallory J.} and {Bin Kwak}, Yoo and Xiaoqian Ma and Merete Nordentoft and Lijun Ouyang and Pariente, {Jose C.} and Franz Resch and Jason Schiffman and S{\o}rensen, {Mikkel E.} and Michio Suzuki and Sophia Vinogradov and Christina Wenneberg and Hidenori Yamasue and Liu Yuan",
note = "Publisher Copyright: {\textcopyright} The Author(s) 2024.",
year = "2024",
doi = "10.1038/s41380-024-02426-7",
language = "English (US)",
journal = "Molecular psychiatry",
issn = "1359-4184",
publisher = "Nature Publishing Group",
}
TY - JOUR
T1 - Using brain structural neuroimaging measures to predict psychosis onset for individuals at clinical high-risk
AU - Zhu, Yinghan
AU - Maikusa, Norihide
AU - Radua, Joaquim
AU - Sämann, Philipp G.
AU - Fusar-Poli, Paolo
AU - Agartz, Ingrid
AU - Andreassen, Ole A.
AU - Bachman, Peter
AU - Baeza, Inmaculada
AU - Chen, Xiaogang
AU - Choi, Sunah
AU - Corcoran, Cheryl M.
AU - Ebdrup, Bjørn H.
AU - Fortea, Adriana
AU - Garani, Ranjini R.G.
AU - Glenthøj, Birte Yding
AU - Glenthøj, Louise Birkedal
AU - Haas, Shalaila S.
AU - Hamilton, Holly K.
AU - Hayes, Rebecca A.
AU - He, Ying
AU - Heekeren, Karsten
AU - Kasai, Kiyoto
AU - Katagiri, Naoyuki
AU - Kim, Minah
AU - Kristensen, Tina D.
AU - Kwon, Jun Soo
AU - Lawrie, Stephen M.
AU - Lebedeva, Irina
AU - Lee, Jimmy
AU - Loewy, Rachel L.
AU - Mathalon, Daniel H.
AU - McGuire, Philip
AU - Mizrahi, Romina
AU - Mizuno, Masafumi
AU - Møller, Paul
AU - Nemoto, Takahiro
AU - Nordholm, Dorte
AU - Omelchenko, Maria A.
AU - Raghava, Jayachandra M.
AU - Røssberg, Jan I.
AU - Rössler, Wulf
AU - Salisbury, Dean F.
AU - Sasabayashi, Daiki
AU - Smigielski, Lukasz
AU - Sugranyes, Gisela
AU - Takahashi, Tsutomu
AU - Tamnes, Christian K.
AU - Tang, Jinsong
AU - Theodoridou, Anastasia
AU - Tomyshev, Alexander S.
AU - Uhlhaas, Peter J.
AU - Værnes, Tor G.
AU - van Amelsvoort, Therese A.M.J.
AU - Waltz, James A.
AU - Westlye, Lars T.
AU - Zhou, Juan H.
AU - Thompson, Paul M.
AU - Hernaus, Dennis
AU - Jalbrzikowski, Maria
AU - Koike, Shinsuke
AU - Allen, Paul
AU - Baldwin, Helen
AU - Catalano, Sabrina
AU - Chee, Michael W.L.
AU - Cho, Kang Ik K.
AU - de Haan, Lieuwe
AU - Horton, Leslie E.
AU - Klaunig, Mallory J.
AU - Bin Kwak, Yoo
AU - Ma, Xiaoqian
AU - Nordentoft, Merete
AU - Ouyang, Lijun
AU - Pariente, Jose C.
AU - Resch, Franz
AU - Schiffman, Jason
AU - Sørensen, Mikkel E.
AU - Suzuki, Michio
AU - Vinogradov, Sophia
AU - Wenneberg, Christina
AU - Yamasue, Hidenori
AU - Yuan, Liu
N1 - Publisher Copyright:© The Author(s) 2024.
PY - 2024
Y1 - 2024
N2 - Machine learning approaches using structural magnetic resonance imaging (sMRI) can be informative for disease classification, although their ability to predict psychosis is largely unknown. We created a model with individuals at CHR who developed psychosis later (CHR-PS+) from healthy controls (HCs) that can differentiate each other. We also evaluated whether we could distinguish CHR-PS+ individuals from those who did not develop psychosis later (CHR-PS-) and those with uncertain follow-up status (CHR-UNK). T1-weighted structural brain MRI scans from 1165 individuals at CHR (CHR-PS+, n = 144; CHR-PS-, n = 793; and CHR-UNK, n = 228), and 1029 HCs, were obtained from 21 sites. We used ComBat to harmonize measures of subcortical volume, cortical thickness and surface area data and corrected for non-linear effects of age and sex using a general additive model. CHR-PS+ (n = 120) and HC (n = 799) data from 20 sites served as a training dataset, which we used to build a classifier. The remaining samples were used external validation datasets to evaluate classifier performance (test, independent confirmatory, and independent group [CHR-PS- and CHR-UNK] datasets). The accuracy of the classifier on the training and independent confirmatory datasets was 85% and 73% respectively. Regional cortical surface area measures-including those from the right superior frontal, right superior temporal, and bilateral insular cortices strongly contributed to classifying CHR-PS+ from HC. CHR-PS- and CHR-UNK individuals were more likely to be classified as HC compared to CHR-PS+ (classification rate to HC: CHR-PS+, 30%; CHR-PS-, 73%; CHR-UNK, 80%). We used multisite sMRI to train a classifier to predict psychosis onset in CHR individuals, and it showed promise predicting CHR-PS+ in an independent sample. The results suggest that when considering adolescent brain development, baseline MRI scans for CHR individuals may be helpful to identify their prognosis. Future prospective studies are required about whether the classifier could be actually helpful in the clinical settings.
AB - Machine learning approaches using structural magnetic resonance imaging (sMRI) can be informative for disease classification, although their ability to predict psychosis is largely unknown. We created a model with individuals at CHR who developed psychosis later (CHR-PS+) from healthy controls (HCs) that can differentiate each other. We also evaluated whether we could distinguish CHR-PS+ individuals from those who did not develop psychosis later (CHR-PS-) and those with uncertain follow-up status (CHR-UNK). T1-weighted structural brain MRI scans from 1165 individuals at CHR (CHR-PS+, n = 144; CHR-PS-, n = 793; and CHR-UNK, n = 228), and 1029 HCs, were obtained from 21 sites. We used ComBat to harmonize measures of subcortical volume, cortical thickness and surface area data and corrected for non-linear effects of age and sex using a general additive model. CHR-PS+ (n = 120) and HC (n = 799) data from 20 sites served as a training dataset, which we used to build a classifier. The remaining samples were used external validation datasets to evaluate classifier performance (test, independent confirmatory, and independent group [CHR-PS- and CHR-UNK] datasets). The accuracy of the classifier on the training and independent confirmatory datasets was 85% and 73% respectively. Regional cortical surface area measures-including those from the right superior frontal, right superior temporal, and bilateral insular cortices strongly contributed to classifying CHR-PS+ from HC. CHR-PS- and CHR-UNK individuals were more likely to be classified as HC compared to CHR-PS+ (classification rate to HC: CHR-PS+, 30%; CHR-PS-, 73%; CHR-UNK, 80%). We used multisite sMRI to train a classifier to predict psychosis onset in CHR individuals, and it showed promise predicting CHR-PS+ in an independent sample. The results suggest that when considering adolescent brain development, baseline MRI scans for CHR individuals may be helpful to identify their prognosis. Future prospective studies are required about whether the classifier could be actually helpful in the clinical settings.
UR - http://www.scopus.com/inward/record.url?scp=85184404780&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85184404780&partnerID=8YFLogxK
U2 - 10.1038/s41380-024-02426-7
DO - 10.1038/s41380-024-02426-7
M3 - Article
C2 - 38332374
AN - SCOPUS:85184404780
SN - 1359-4184
JO - Molecular psychiatry
JF - Molecular psychiatry
ER -