BrainLossNet: a fast, accurate and robust method to estimate brain volume loss from longitudinal MRI.

PURPOSE: MRI-derived brain volume loss (BVL) is widely used as neurodegeneration marker. SIENA is state-of-the-art for BVL measurement, but limited by long computation time. Here we propose "BrainLossNet", a convolutional neural network (CNN)-based method for BVL-estimation. METHODS: BrainLossNet uses CNN-based non-linear registration of baseline(BL)/follow-up(FU) 3D-T1w-MRI pairs. BVL is computed by non-linear registration of brain parenchyma masks segmented in the BL/FU scans. The BVL estimate is corrected for image distortions using the apparent volume change of the total intracranial volume. BrainLossNet was trained on 1525 BL/FU pairs from 83 scanners. Agreement between BrainLossNet and SIENA was assessed in 225 BL/FU pairs from 94 MS patients acquired with a single scanner and 268 BL/FU pairs from 52 scanners acquired for various indications. Robustness to short-term variability of 3D-T1w-MRI was compared in 354 BL/FU pairs from a single healthy men acquired in the same session without repositioning with 116 scanners (Frequently-Traveling-Human-Phantom dataset, FTHP). RESULTS: Processing time of BrainLossNet was 2-3 min. The median [interquartile range] of the SIENA-BrainLossNet BVL difference was 0.10% [- 0.18%, 0.35%] in the MS dataset, 0.08% [- 0.14%, 0.28%] in the various indications dataset. The distribution of apparent BVL in the FTHP dataset was narrower with BrainLossNet (p = 0.036; 95th percentile: 0.20% vs 0.32%). CONCLUSION: BrainLossNet on average provides the same BVL estimates as SIENA, but it is significantly more robust, probably due to its built-in distortion correction. Processing time of 2-3 min makes BrainLossNet suitable for clinical routine. This can pave the way for widespread clinical use of BVL estimation from intra-scanner BL/FU pairs.

Prediction of Alzheimer's Dementia in Patients with Amnestic Mild Cognitive Impairment in Clinical Routine: Incremental Value of Biomarkers of Neurodegeneration and Brain Amyloidosis Added Stepwise to Cognitive Status.

The aim of this study was to evaluate the incremental benefit of biomarkers for prediction of Alzheimer's disease dementia (ADD) in patients with mild cognitive impairment (MCI) when added stepwise in the order of their collection in clinical routine. The model started with cognitive status characterized by the ADAS-13 score. Hippocampus volume (HV), cerebrospinal fluid (CSF) phospho-tau (pTau), and the FDG t-sum score in an AD meta-region-of-interest were compared as neurodegeneration markers. CSF-Aß1-42 was used as amyloidosis marker. The incremental prognostic benefit from these markers was assessed by stepwise Kaplan-Meier survival analysis in 402 ADNI MCI subjects. Predefined cutoffs were used to dichotomize patients as 'negative' or 'positive' for AD characteristic alteration with respect to each marker. Among the neurodegeneration markers, CSF-pTau provided the best incremental risk stratification when added to ADAS-13. FDG PET outperformed HV only in MCI subjects with relatively preserved cognition. Adding CSF-Aß provided further risk stratification in pTau-positive subjects, independent of their cognitive status. Stepwise integration of biomarkers allows stepwise refinement of risk estimates for MCI-to-ADD progression. Incremental benefit strongly depends on the patient's status according to the preceding diagnostic steps. The stepwise Kaplan-Meier curves might be useful to optimize diagnostic workflow in individual patients.

Mental speed is associated with the shape irregularity of white matter MRI hyperintensity load.

Brain MRI white matter hyperintensities (WMHs) are common in elderly subjects. Their impact on cognition, however, appears highly variable. Complementing conventional scoring of WMH load (volume and location) by quantitative characterization of the shape irregularity of WMHs might improve the understanding of the relationship between WMH load and cognitive performance. Here we propose the "confluency sum score" (COSU) as a marker of the total shape irregularity of WMHs in the brain. The study included two independent patient samples: 87 cognitively impaired geriatric inpatients from a prospective neuroimaging study (iDSS) and 198 subjects from the National Alzheimer's Coordinating Center (NACC) database (132 with, 66 w/o cognitive impairment). After automatic segmentation and clustering of the WMHs on FLAIR (LST toolbox, SPM8), the confluency of the i-th contiguous WMH cluster was computed as confluencyi = [1/(36π)∙surfacei3/volumei2]1/3-1. The COSU was obtained by summing the confluency over all WMH clusters. COSU was tested for correlation with CERAD-plus subscores. Correlation analysis was restricted to subjects with at least moderate WMH load (≥ 13.5 ml; iDSS / NACC: n = 52 / 80). In the iDSS sample, among the 12 CERAD-plus subtests the trail making test A (TMT-A) was most strongly correlated with the COSU (Spearman rho = -0.345, p = 0.027). TMT-A performance was not associated with total WMH volume (rho = 0.147, p = 0.358). This finding was confirmed in the NACC sample (rho = -0.261, p = 0.023 versus rho = -0.040, p = 0.732). Cognitive performance in specific domains including mental speed and fluid abilities seems to be more strongly associated with the shape irregularity of white matter MRI hyperintensities than with their volume.

Hippocampal and cerebellar atrophy in patients with Cushing's disease.

OBJECT Cushing's disease (CD) may cause atrophy of different regions of the human brain, mostly affecting the hippocampus and the cerebellum. This study evaluates the use of 3-T MRI of newly diagnosed patients with CD to detect atrophic degeneration with voxel-based volumetry. METHODS Subjects with newly diagnosed, untreated CD were included and underwent 3-T MRI. Images were analyzed using a voxelwise statistical test to detect reduction of brain parenchyma. In addition, an atlas-based volumetric study for regions likely to be affected by CD was performed. RESULTS Nineteen patients with a mean disease duration of 24 months were included. Tumor markers included adrenocorticotropic hormone (median 17.5 pmol/L), cortisol (949.4 nmol/L), and dehydroepiandrosterone sulfate (5.4 µmol/L). The following values are expressed as the mean ± SD. The voxelwise statistical test revealed clusters of significantly reduced gray matter in the hippocampus and cerebellum, with volumes of 2.90 ± 0.26 ml (right hippocampus), 2.89 ± 0.28 ml (left hippocampus), 41.95 ± 4.67 ml (right cerebellar hemisphere), and 42.11 ± 4.59 ml (left cerebellar hemisphere). Healthy control volunteers showed volumes of 3.22 ± 0.25 ml for the right hippocampus, 3.23 ± 0.25 ml for the left hippocampus, 50.87 ± 4.23 ml for the right cerebellar hemisphere, and 50.42 ± 3.97 ml for the left cerebellar hemisphere. CONCLUSIONS Patients with untreated CD show significant reduction of gray matter in the cerebellum and hippocampus. These changes can be analyzed and objectified with the quantitative voxel-based method described in this study.

Metal artifact reduction in CT using tissue-class modeling and adaptive prefiltering.

High-density objects such as metal prostheses, surgical clips, or dental fillings generate streak-like artifacts in computed tomography images. We present a novel method for metal artifact reduction by in-painting missing information into the corrupted sinogram. The information is provided by a tissue-class model extracted from the distorted image. To this end the image is first adaptively filtered to reduce the noise content and to smooth out streak artifacts. Consecutively, the image is segmented into different material classes using a clustering algorithm. The corrupted and missing information in the original sinogram is completed using the forward projected information from the tissue-class model. The performance of the correction method is assessed on phantom images. Clinical images featuring a broad spectrum of metal artifacts are studied. Phantom and clinical studies show that metal artifacts, such as streaks, are significantly reduced and shadows in the image are eliminated. Furthermore, the novel approach improves detectability of organ contours. This can be of great relevance, for instance, in radiation therapy planning, where images affected by metal artifacts may lead to suboptimal treatment plans.

The impact of dental metal artifacts on head and neck IMRT dose distributions.

BACKGROUND AND PURPOSES: To quantify the cold or hot spot induced in IMRT treatment plans due to the presence of metal artifact in CT image data sets stemming from dental work. PATIENTS AND METHODS: Metal artifact corrected image data sets of five patients have been analyzed. IMRT plans were generated using five different planning image data sets: (a) uncorrected (UC) (b) homogeneous uncorrected (HUC), (c) sinogram completion corrected (SCC), (d) minimum value corrected (MVC), and (e) image set (d) subsequently corrected with a streak artifacts reduction algorithm (SAR-MVC). The SAR-MVC data set is assumed to be the closest approximation to the absence of metal artifacts and has therefore been taken as the reference image data set. An IMRT plan was generated for each of the image datasets (a)-(e). The resulting IMRT treatment plans for data sets (a)-(d) were then projected onto the reference data set (e) and recalculated. The reference dose distribution (e) was then subtracted from these recalculated dose distributions. Using dose difference analysis, the cold and hot spots in organs at risk (OARs) and the target volumes (TVs) were quantified. RESULTS: When compared to the reference dose distribution, the UC, HUC, and SCC plans exhibited hot spots showing on average more than 1.0 Gy hot dose in the left and right parotids. For the UC, HUC, and SCC recalculated plans, subvolumes of the clinical target volumes (CTV) were under dosed on average by more than 0.9 Gy. On the other hand, the MVC plan showed less than 0.3 Gy hot dose in both parotids, and the cold dose in the CTVs were reduced by up to 0.8 Gy. CONCLUSIONS: The presence of dental metal artifacts in head and neck planning CT data sets can lead to relative hot spots in OARs and relative cold spots in regions of the TVs when compared to the reference data set that more closely approximates the patient anatomy. This effect can be reduced if a simple minimum value correction (MVC) method for the dental metal artifacts is employed.