Potential of a statistical approach for the standardization of multicenter diffusion tensor data: A phantom study.

BACKGROUND: Diffusion tensor imaging (DTI) parameters, such as fractional anisotropy (FA), allow examining the structural integrity of the brain. However, the true value of these parameters may be confounded by variability in MR hardware, acquisition parameters, and image quality. PURPOSE: To examine the effects of confounding factors on FA and to evaluate the feasibility of statistical methods to model and reduce multicenter variability. STUDY TYPE: Longitudinal multicenter study. PHANTOM: DTI single strand phantom (HQ imaging). FIELD STRENGTH/SEQUENCE: 3T diffusion tensor imaging. ASSESSMENTS: Thirteen European imaging centers participated. DTI scans were acquired every 6 months and whenever maintenance or upgrades to the system were performed. A total of 64 scans were acquired in 2 years, obtained by three scanner vendors, using six individual head coils, and 12 software versions. STATISTICAL TESTS: The variability in FA was assessed by the coefficients of variation (CoV). Several linear mixed effects models (LMEM) were developed and compared by means of the Akaike Information Criterion (AIC). RESULTS: The CoV was 2.22% for mean FA and 18.40% for standard deviation of FA. The variables "site" (P = 9.26 × 10-5 ), "vendor" (P = 2.18 × 10-5 ), "head coil" (P = 9.00 × 10-4 ), "scanner drift," "bandwidth" (P = 0.033), "TE" (P = 8.20 × 10-6 ), "SNR" (P = 0.029) and "mean residuals" (P = 6.50 × 10-4 ) had a significant effect on the variability in mean FA. The variables "site" (P = 4.00 × 10-4 ), "head coil" (P = 2.00 × 10-4 ), "software" (P = 0.014), and "mean voxel outlier intensity count" (P = 1.10 × 10-4 ) had a significant effect on the variability in standard deviation of FA. The mean FA was best predicted by an LMEM that included "vendor" and the interaction term of "SNR" and "head coil" as model factors (AIC -347.98). In contrast, the standard deviation of FA was best predicted by an LMEM that included "vendor," "bandwidth," "TE," and the interaction term between "SNR" and "head coil" (AIC -399.81). DATA CONCLUSION: Our findings suggest that perhaps statistical models seem promising to model the variability in quantitative DTI biomarkers for clinical routine and multicenter studies. LEVEL OF EVIDENCE: 4 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:955-965.

Artificial Intelligence Assistive Software Tool for Automated Detection and Quantification of Amyloid-Related Imaging Abnormalities.

Importance: Amyloid-related imaging abnormalities (ARIA) are brain magnetic resonance imaging (MRI) findings associated with the use of amyloid-ß-directed monoclonal antibody therapies in Alzheimer disease (AD). ARIA monitoring is important to inform treatment dosing decisions and might be improved through assistive software. Objective: To assess the clinical performance of an artificial intelligence (AI)-based software tool for assisting radiological interpretation of brain MRI scans in patients monitored for ARIA. Design, Setting, and Participants: This diagnostic study used a multiple-reader multiple-case design to evaluate the diagnostic performance of radiologists assisted by the software vs unassisted. The study enrolled 16 US Board of Radiology-certified radiologists to perform radiological reading with (assisted) and without the software (unassisted). The study encompassed 199 retrospective cases, where each case consisted of a predosing baseline and a postdosing follow-up MRI of patients from aducanumab clinical trials PRIME, EMERGE, and ENGAGE. Statistical analysis was performed from April to July 2023. Exposures: Use of icobrain aria, an AI-based assistive software for ARIA detection and quantification. Main Outcomes and Measures: Coprimary end points were the difference in diagnostic accuracy between assisted and unassisted detection of ARIA-E (edema and/or sulcal effusion) and ARIA-H (microhemorrhage and/or superficial siderosis) independently, assessed with the area under the receiver operating characteristic curve (AUC). Results: Among the 199 participants included in this study of radiological reading performance, mean (SD) age was 70.4 (7.2) years; 105 (52.8%) were female; 23 (11.6%) were Asian, 1 (0.5%) was Black, 157 (78.9%) were White, and 18 (9.0%) were other or unreported race and ethnicity. Among the 16 radiological readers included, 2 were specialized neuroradiologists (12.5%), 11 were male individuals (68.8%), 7 were individuals working in academic hospitals (43.8%), and they had a mean (SD) of 9.5 (5.1) years of experience. Radiologists assisted by the software were significantly superior in detecting ARIA than unassisted radiologists, with a mean assisted AUC of 0.87 (95% CI, 0.84-0.91) for ARIA-E detection (AUC improvement of 0.05 [95% CI, 0.02-0.08]; P = .001]) and 0.83 (95% CI, 0.78-0.87) for ARIA-H detection (AUC improvement of 0.04 [95% CI, 0.02-0.07]; P = .001). Sensitivity was significantly higher in assisted reading compared with unassisted reading (87% vs 71% for ARIA-E detection; 79% vs 69% for ARIA-H detection), while specificity remained above 80% for the detection of both ARIA types. Conclusions and Relevance: This diagnostic study found that radiological reading performance for ARIA detection and diagnosis was significantly better when using the AI-based assistive software. Hence, the software has the potential to be a clinically important tool to improve safety monitoring and management of patients with AD treated with amyloid-ß-directed monoclonal antibody therapies.

A patch-based super resolution algorithm for improving image resolution in clinical mass spectrometry.

Mass spectrometry imaging (MSI) and histology are complementary analytical tools. Integration of the two imaging modalities can enhance the spatial resolution of the MSI beyond its experimental limits. Patch-based super resolution (PBSR) is a method where high spatial resolution features from one image modality guide the reconstruction of a low resolution image from a second modality. The principle of PBSR lies in image redundancy and aims at finding similar pixels in the neighborhood of a central pixel that are then used to guide reconstruction of the central pixel. In this work, we employed PBSR to increase the resolution of MSI. We validated the proposed pipeline by using a phantom image (micro-dissected logo within a tissue) and mouse cerebellum samples. We compared the performance of the PBSR with other well-known methods: linear interpolation (LI) and image fusion (IF). Quantitative and qualitative assessment showed advantage over the former and comparability with the latter. Furthermore, we demonstrated the potential applicability of PBSR in a clinical setting by accurately integrating structural (i.e., histological) and molecular (i.e., MSI) information from a case study of a dog liver.

Automatic Quantification of Computed Tomography Features in Acute Traumatic Brain Injury.

Traumatic brain injury is a complex and diverse medical condition with a high frequency of intracranial abnormalities. These can typically be visualized on a computed tomography (CT) scan, which provides important information for further patient management, such as the need for operative intervention. In order to quantify the extent of acute intracranial lesions and associated secondary injuries, such as midline shift and cisternal compression, visual assessment of CT images has limitations, including observer variability and lack of quantitative interpretation. Automated image analysis can quantify the extent of intracranial abnormalities and provide added value in routine clinical practice. In this article, we present icobrain, a fully automated method that reliably computes acute intracranial lesions volume based on deep learning, cistern volume, and midline shift on the noncontrast CT image of a patient. The accuracy of our method is evaluated on a subset of the multi-center data set from the CENTER-TBI (Collaborative European Neurotrauma Effectiveness Research in Traumatic Brain Injury) study for which expert annotations were used as a reference. Median volume differences between expert assessments and icobrain are 0.07 mL for acute intracranial lesions and -0.01 mL for cistern segmentation. Correlation between expert assessments and icobrain is 0.91 for volume of acute intracranial lesions and 0.94 for volume of the cisterns. For midline shift computations, median error is -0.22 mm, with a correlation of 0.93 with expert assessments.

Diffusion kurtosis imaging allows the early detection and longitudinal follow-up of amyloid-ß-induced pathology.

BACKGROUND: Alzheimer's disease (AD) is a progressive neurodegenerative disorder and the most common cause of dementia in the elderly population. In this study, we used the APP/PS1 transgenic mouse model to explore the feasibility of using diffusion kurtosis imaging (DKI) as a tool for the early detection of microstructural changes in the brain due to amyloid-ß (Aß) plaque deposition. METHODS: We longitudinally acquired DKI data of wild-type (WT) and APP/PS1 mice at 2, 4, 6 and 8 months of age, after which these mice were sacrificed for histological examination. Three additional cohorts of mice were also included at 2, 4 and 6 months of age to allow voxel-based co-registration between diffusion tensor and diffusion kurtosis  metrics and immunohistochemistry. RESULTS: Changes were observed in diffusion tensor (DT) and diffusion kurtosis (DK) metrics in many of the 23 regions of interest that were analysed. Mean and axial kurtosis were greatly increased owing to Aß-induced pathological changes in the motor cortex of APP/PS1 mice at 4, 6 and 8 months of age. Additionally, fractional anisotropy (FA) was decreased in APP/PS1 mice at these respective ages. Linear discriminant analysis of the motor cortex data indicated that combining diffusion tensor and diffusion kurtosis metrics permits improved separation of WT from APP/PS1 mice compared with either diffusion tensor or diffusion kurtosis metrics alone. We observed that mean kurtosis and FA are the critical metrics for a correct genotype classification. Furthermore, using a newly developed platform to co-register the in vivo diffusion-weighted magnetic resonance imaging with multiple 3D histological stacks, we found high correlations between DK metrics and anti-Aß (clone 4G8) antibody, glial fibrillary acidic protein, ionised calcium-binding adapter molecule 1 and myelin basic protein immunohistochemistry. Finally, we observed reduced FA in the septal nuclei of APP/PS1 mice at all ages investigated. The latter was at least partially also observed by voxel-based statistical parametric mapping, which showed significantly reduced FA in the septal nuclei, as well as in the corpus callosum, of 8-month-old APP/PS1 mice compared with WT mice. CONCLUSIONS: Our results indicate that DKI metrics hold tremendous potential for the early detection and longitudinal follow-up of Aß-induced pathology.

Reliable measurements of brain atrophy in individual patients with multiple sclerosis.

INTRODUCTION: As neurodegeneration is recognized as a major contributor to disability in multiple sclerosis (MS), brain atrophy quantification could have a high added value in clinical practice to assess treatment efficacy and disease progression, provided that it has a sufficiently low measurement error to draw meaningful conclusions for an individual patient. METHOD: In this paper, we present an automated longitudinal method based on Jacobian integration for measuring whole-brain and gray matter atrophy based on anatomical magnetic resonance images (MRI), named MSmetrix. MSmetrix is specifically designed to measure atrophy in patients with MS, by including iterative lesion segmentation and lesion filling based on FLAIR and T1-weighted MRI scans. RESULTS: MS metrix is compared with SIENA with respect to test-retest error and consistency, resulting in an average test-retest error on an MS data set of 0.13% (MS metrix) and 0.17% (SIENA) and a consistency error of 0.07% (MS metrix) and 0.05% (SIENA). On a healthy subject data set including physiological variability the test-retest is 0.19% (MS metrix) and 0.31% (SIENA). CONCLUSION: Therefore, we can conclude that MSmetrix could be of added value in clinical practice for the follow-up of treatment and disease progression in MS patients.