Quantitative MR Image Reconstruction Using Parameter-Specific Dictionary Learning With Adaptive Dictionary-Size and Sparsity-Level Choice.

OBJECTIVE: We propose a method for the reconstruction of parameter-maps in Quantitative Magnetic Resonance Imaging (QMRI). METHODS: Because different quantitative parameter-maps differ from each other in terms of local features, we propose a method where the employed dictionary learning (DL) and sparse coding (SC) algorithms automatically estimate the optimal dictionary-size and sparsity level separately for each parameter-map. We evaluated the method on a T1-mapping QMRI problem in the brain using the BrainWeb data as well as in-vivo brain images acquired on an ultra-high field 7 T scanner. We compared it to a model-based acceleration for parameter mapping (MAP) approach, other sparsity-based methods using total variation (TV), Wavelets (Wl), and Shearlets (Sh) to a method which uses DL and SC to reconstruct qualitative images, followed by a non-linear (DL+Fit). RESULTS: Our algorithm surpasses MAP, TV, Wl, and Sh in terms of RMSE and PSNR. It yields better or comparable results to DL+Fit by additionally significantly accelerating the reconstruction by a factor of approximately seven. CONCLUSION: The proposed method outperforms the reported methods of comparison and yields accurate T1-maps. Although presented for T1-mapping in the brain, our method's structure is general and thus most probably also applicable for the the reconstruction of other quantitative parameters in other organs. SIGNIFICANCE: From a clinical perspective, the obtained T1-maps could be utilized to differentiate between healthy subjects and patients with Alzheimer's disease. From a technical perspective, the proposed unsupervised method could be employed to obtain ground-truth data for the development of data-driven methods based on supervised learning.

Macromolecule modelling for improved metabolite quantification using short echo time brain 1 H MRS at 3 T and 7 T: The PRaMM Model.

Purpose: To improve reliability of metabolite quantification at both, 3 T and 7 T, we propose a novel parametrized macromolecules quantification model (PRaMM) for brain 1 H MRS, in which the ratios of macromolecule peak intensities are used as soft constraints. Methods: Full- and metabolite-nulled spectra were acquired in three different brain regions with different ratios of grey and white matter from six healthy volunteers, at both 3 T and 7 T. Metabolite-nulled spectra were used to identify highly correlated macromolecular signal contributions and estimate the ratios of their intensities. These ratios were then used as soft constraints in the proposed PRaMM model for quantification of full spectra. The PRaMM model was validated by comparison with a single component macromolecule model and a macromolecule subtraction technique. Moreover, the influence of the PRaMM model on the repeatability and reproducibility compared to those other methods was investigated. Results: The developed PRaMM model performed better than the two other approaches in all three investigated brain regions. Several estimates of metabolite concentration and their Cramér-Rao lower bounds were affected by the PRaMM model reproducibility, and repeatability of the achieved concentrations were tested by evaluating the method on a second repeated acquisitions dataset. While the observed effects on both metrics were not significant, the fit quality metrics were improved for the PRaMM method (p≤0.0001). Minimally detectable changes are in the range 0.5 - 1.9 mM and percent coefficients of variations are lower than 10% for almost all the clinically relevant metabolites. Furthermore, potential overparameterization was ruled out. Conclusion: Here, the PRaMM model, a method for an improved quantification of metabolites was developed, and a method to investigate the role of the MM background and its individual components from a clinical perspective is proposed.

7T amygdala and hippocampus subfields in volumetry-based associations with memory: A 3-year follow-up study of early Alzheimer's disease.

INTRODUCTION: The hippocampus is the most prominent single region of interest (ROI) for the diagnosis and prediction of Alzheimer's disease (AD). However, its suitability in the earliest stages of cognitive decline, i.e., subjective cognitive decline (SCD), remains uncertain which warrants the pursuit of alternative or complementary regions. The amygdala might be a promising candidate, given its implication in memory as well as other psychiatric disorders, e.g. depression and anxiety, which are prevalent in SCD. In this 7 tesla (T) magnetic resonance imaging (MRI) study, we aimed to compare the contribution of volumetric measurements of the hippocampus, the amygdala, and their respective subfields, for early diagnosis and prediction in an AD-related study population. METHODS: Participants from a longitudinal study were grouped into SCD (n = 29), mild cognitive impairment (MCI, n = 23), AD (n = 22) and healthy control (HC, n = 31). All participants underwent 7T MRI at baseline and extensive neuropsychological testing at up to three visits (baseline n = 105, 1-year n = 78, 3-year n = 39). Analysis of covariance (ANCOVA) was used to assess group differences of baseline volumes of the amygdala and the hippocampus and their subfields. Linear mixed models were used to estimate the effects of baseline volumes on yearly changes of a z-scaled memory score. All models were adjusted to age, sex and education. RESULTS: Compared to the HC group, individuals with SCD showed smaller amygdala ROI volumes (range across subfields -11% to -1%), but not hippocampus ROI volumes (-2% to 1%) except for the hippocampus-amygdala-transition-area (-7%). However, cross-sectional associations between baseline memory and volumes were smaller for amygdala ROIs (std. ß [95% CI] ranging between 0.16 [0.08; 0.25] and 0.46 [0.31; 0.60]) than hippocampus ROIs (between 0.32 [0.19; 0.44] and 0.53 [0.40; 0.67]). Further, the association of baseline volumes with yearly memory change in the HC and SCD groups was similarly weak for amygdala ROIs and hippocampus ROIs. In the MCI group, volumes of amygdala ROIs were associated with a relevant yearly memory decline [95% CI] ranging between -0.12 [-0.24; 0.00] and -0.26 [-0.42; -0.09] for individuals with 20% smaller volumes than the HC group. However, effects were stronger for hippocampus ROIs with a corresponding yearly memory decline ranging between -0.21 [-0.35; -0.07] and -0.31 [-0.50; -0.13]. CONCLUSION: Volumes of amygdala ROIs, as determined by 7T MRI, might contribute to objectively and non-invasively identify patients with SCD, and thus aid early diagnosis and treatment of individuals at risk to develop dementia due to AD, however associations with other psychiatric disorders should be evaluated in further studies. The amygdala's value in the prediction of longitudinal memory changes in the SCD group remains questionable. Primarily in patients with MCI, memory decline over 3 years appears to be more strongly associated with volumes of hippocampus ROIs than amygdala ROIs.

Forward and Backward Recalling Sequences in Spatial and Verbal Memory Tasks: What Do We Measure?

There are different views in the literature about the number and inter-relationships of cognitive domains (such as memory and executive function) and a lack of understanding of the cognitive processes underlying these domains. In previous publications, we demonstrated a methodology for formulating and testing cognitive constructs for visuo-spatial and verbal recall tasks, particularly for working memory task difficulty where entropy is found to play a major role. In the present paper, we applied those insights to a new set of such memory tasks, namely, backward recalling block tapping and digit sequences. Once again, we saw clear and strong entropy-based construct specification equations (CSEs) for task difficulty. In fact, the entropy contributions in the CSEs for the different tasks were of similar magnitudes (within the measurement uncertainties), which may indicate a shared factor in what is being measured with both forward and backward sequences, as well as visuo-spatial and verbal memory recalling tasks more generally. On the other hand, the analyses of dimensionality and the larger measurement uncertainties in the CSEs for the backward sequences suggest that caution is needed when attempting to unify a single unidimensional construct based on forward and backward sequences with visuo-spatial and verbal memory tasks.

The Role of Entropy in Construct Specification Equations (CSE) to Improve the Validity of Memory Tests: Extension to Word Lists.

Metrological methods for word learning list tests can be developed with an information theoretical approach extending earlier simple syntax studies. A classic Brillouin entropy expression is applied to the analysis of the Rey's Auditory Verbal Learning Test RAVLT (immediate recall), where more ordered tasks-with less entropy-are easier to perform. The findings from three case studies are described, including 225 assessments of the NeuroMET2 cohort of persons spanning a cognitive spectrum from healthy older adults to patients with dementia. In the first study, ordinality in the raw scores is compensated for, and item and person attributes are separated with the Rasch model. In the second, the RAVLT IR task difficulty, including serial position effects (SPE), particularly Primacy and Recency, is adequately explained (Pearson's correlation R=0.80) with construct specification equations (CSE). The third study suggests multidimensionality is introduced by SPE, as revealed through goodness-of-fit statistics of the Rasch analyses. Loading factors common to two kinds of principal component analyses (PCA) for CSE formulation and goodness-of-fit logistic regressions are identified. More consistent ways of defining and analysing memory task difficulties, including SPE, can maintain the unique metrological properties of the Rasch model and improve the estimates and understanding of a person's memory abilities on the path towards better-targeted and more fit-for-purpose diagnostics.