Single subject prediction of brain disorders in neuroimaging: Promises and pitfalls.
Citations
917 citations
622 citations
Cites background or methods or result from "Single subject prediction of brain ..."
...[3] survey, a clear pattern emerged with higher prediction accuracy reported in the studies with small sample sizes....
[...]
...[3] conducted a survey of neuroimaging studies which used supervised ML methods to classify healthy individuals and individuals with brain disorders....
[...]
...Arbabshirani et al. [3] and Varoquaux [4] reviewed studies, which applied ML methods for prediction of various brain disorders and found that studies with smaller sample sizes tended to report higher classification performance....
[...]
...Most of the studies in our own and other surveys [3, 4] used feature selection....
[...]
...In their survey Arbabshirani et al. [3] noted that most of the neuroimaging studies consisted of two parts....
[...]
583 citations
460 citations
Cites background from "Single subject prediction of brain ..."
...Recent review papers (Arbabshirani et al., 2017; Falahati et al., 2014) reported studies on MRI- and multimodality-based classification of AD and MCI, limiting AD classification to MRI or its combination with other modalities only....
[...]
455 citations
Cites background or methods from "Single subject prediction of brain ..."
...In such cases, a high accuracy can be misleading, as it may reflect an overestimation of the algorithm’s performance (Arbabshirani et al., 2016)....
[...]
...73 Over the past decade, several ML methods have been applied to neuroimaging data from psychiatric and neurological patients with varying degrees of success (Arbabshirani et al., 2016; Wolfers et al., 2015)....
[...]
...Over the past decade, several ML methods have been applied to neuroimaging data from psychiatric and neurological patients with varying degrees of success (Arbabshirani et al., 2016; Wolfers et al., 2015)....
[...]
...In such cases, a high accuracy can be misleading, as it may reflect an overestimation of the algorithm‟s performance (Arbabshirani et al., 2016)....
[...]
...number of voxels) for a subject is much larger than the total number of subjects, resulting in high-dimensional data (Arbabshirani et al., 2016)....
[...]
References
16,538 citations
"Single subject prediction of brain ..." refers methods in this paper
...Examples of regularization algorithms used in embedded feature selection methods are LASSO, elastic net and ridge regression (Hastie et al., 2004; Ng, 2004; Park and Hastie, 2007; Zou and Hastie, 2005)....
[...]
15,478 citations
14,509 citations
"Single subject prediction of brain ..." refers background or methods in this paper
...(2008) t2:36 sMRI AD GM maps based on VBM 384,065 SVM HC = 137, AD = 108, MCI = 203, Total = 448 63.7–80.3% Adaszewski et al. (2013) t2:37 sMRI AD Gray matter probability maps 2E6 SVM HC = 226, AD = 91, Total = 417 87% Abdulkadir et al....
[...]
...Finally, there are embedded feature selection methods (Guyon and Elisseeff, 2003)....
[...]
...(2008) t2:36 sMRI AD GM maps based on VBM 384,065 SVM HC = 137, AD = 108, MCI = 203, Total = 448 63.7–80.3% Adaszewski et al. (2013) t2:37 sMRI AD Gray matter probability maps 2E6 SVM HC = 226, AD = 91, Total = 417 87% Abdulkadir et al. (2011) t2:38 sMRI and dMRI AD FA and GM volumes 142 SVM NC = 15, AD = 21, Total = 36 94....
[...]
...t2:7 rsfMRI AD Averaged voxel intensities of selected resting-state networks 4 Multivariate ROC HC = 16, AD = 15, Total = 31 100% Wu et al. (2013) t2:8 rsfMRI AD Graph measures based on FC analysis among ROIs 454 SVM HC = 20, AD = 20, Total = 40 100% Khazaee et al. (2015) t2:9 sMRI AD Eigen brains of key slices 10 SVM NC = 98, AD = 28, Total = 126 92.3% Zhang et al. (2015) t2:10 sMRI AD ODVBA of RAVENs maps N/A SVM HC = 50, AD = 50, Total = 100 90% Zhang and Davatzikos (2011) t2:11 sMRI AD Hippocampus shape measures using LDDMM and PCA 20 principal components (3–4 selected by the classifier) Logistic regression HC = 26, DAT = 18, Total = 44 81....
[...]
...t2:7 rsfMRI AD Averaged voxel intensities of selected resting-state networks 4 Multivariate ROC HC = 16, AD = 15, Total = 31 100% Wu et al. (2013) t2:8 rsfMRI AD Graph measures based on FC analysis among ROIs 454 SVM HC = 20, AD = 20, Total = 40 100% Khazaee et al....
[...]
13,710 citations
11,648 citations
"Single subject prediction of brain ..." refers background in this paper
...It is estimated that by the year 2020, depression will account for 15% of the disease burden in the world ranking second after heart disease (Kessler et al., 1994)....
[...]