Automatic change detection in multimodal serial MRI: application to multiple sclerosis lesion evolution

Detection of Focal Longitudinal Changes in the Brain by Subtraction of MR Images

“…Although several other studies have previously used subtraction imaging, [13][14][15][16][17][18] the procedure outlined here has several advantages. First, it can be used to analyze MR imaging data acquired serially over many time points in an unbiased fashion.…”

A non-parametric approach to automatic change detection in MRI images of the brain

“…The proposed 1 It is worth mentioning that performance of our change detection system is not particularly sensitive to the choice of LSK size because the local covariance matrix C i plays a role in automatically determining the shape and size of kernels. 2 , where Agt represents the ground truth, which is regions with true lesions (i.e., simulated lesions in this paper). A dt represents detected lesions.…”

“…For example, there are at least five different MRI modalities including T1 weighed, inversion recovery (IR), proton-density-weighted (PD), T2-weighted, and fluid attenuation inversion recovery (FLAIR). For statistical change detection in multispectral MRI scans, Bosc et al [2] used the Generalized Likelihood Ratio Test (GLRT) followed by nonlinear joint histogram normalization. However, their approach tends to fail when noise is non-stationary.…”

On-line anomaly detection and resilience in classifier ensembles

“…GLR maximizes the likelihood ratio over possible values of the parameter of the changed signal, assuming the normal data follows a Gaussian distribution. It is widely used for change detection in brain imaging (Bosc et al, 2003), diffusion tensor imaging for monitoring neuro-degenerative diseases (Boisgontier et al, 2009), for detecting land mines using multi-spectral images (Anderson, 2008), and target detection and parameter estimation of MIMO radar (Xu and Li, 2007). In the same class, Adaptive CUSUM is able to detect changes by suggesting a distribution for the unknown change model based on the distribution of the known model of unaltered data Roveri, 2006a,b, 2008).…”