Considerations on accuracy, pattern and possible underlying factors of brain microbleed progression in older adults with absence or mild presence of vascular pathology

A new family of instance-level loss functions for improving instance-level segmentation and detection of white matter hyperintensities in routine clinical brain MRI

In this study, we introduce "instance loss functions", a new family of loss functions designed to enhance the training of neural networks in the instance-level segmentation and detection of objects in biomedical image data, particularly those of varied numbers and sizes. Intended to be utilized conjointly with traditional loss functions, these proposed functions, prioritize object instances over pixel-by-pixel comparisons. The specific functions, the instance segmentation loss (Linstance), the instance center loss (Lcenter), the false instance rate loss (Lfalse), and the instance proximity loss (Lproximity), serve distinct purposes. Specifically, Linstance improves instance-wise segmentation quality, Lcenter enhances segmentation quality of small instances, Lfalse minimizes the rate of false and missed detections across varied instance sizes, and Lproximity improves detection quality by pulling predicted instances towards the ground truth instances. Through the task of segmenting white matter hyperintensities (WMH) in brain MRI, we benchmarked our proposed instance loss functions, both individually and in combination via an ensemble inference models approach, against traditional pixel-level loss functions. Data were sourced from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the WMH Segmentation Challenge datasets, which exhibit significant variation in WMH instance sizes. Empirical evaluations demonstrate that combining two instance-level loss functions through ensemble inference models outperforms models using other loss function on both the ADNI and WMH Segmentation Challenge datasets for the segmentation and detection of WMH instances. Further, applying these functions to the segmentation of nuclei in histopathology images demonstrated their effectiveness and generalizability beyond WMH, improving performance even in contexts with less severe instance imbalance.

Prevalence of cerebral microbleeds in Alzheimer's disease, dementia with Lewy bodies and Parkinson's disease dementia: A systematic review and meta-analysis

Cerebral microbleeds (CMB) are often associated with vascular risk factors and/or cerebral amyloid angiopathy and are frequently identified in people with dementia. The present study therefore aimed to estimate the pooled prevalence and associations of CMB in Alzheimer's disease (AD), dementia with Lewy bodies (DLB) and Parkinson's disease dementia (PDD), using meta-analytic methods. Sixty-five MRI studies were included after a systematic search on major electronic databases. We found that the prevalence of CMB was comparable across the three dementia subtypes (31-36%) and was highly influenced by the MRI techniques used. CMB in AD were associated with a history of hypertension and amyloid-β burden. In contrast, CMB in DLB, despite being predominantly lobar, were associated with hypertension, but not amyloid-β burden. These findings suggest that the underlying pathophysiology of CMB in DLB might differ from that of AD. There was substantially larger number of AD studies identified and more studies evaluating CMB in Lewy body dementias are warranted.

Inherited risk of dementia and the progression of cerebral small vessel disease and inflammatory markers in cognitively healthy midlife adults: the PREVENT-Dementia study

Cerebral small vessel disease (SVD) and inflammation are increasingly recognized as key contributors to Alzheimer's disease (AD), although the timing, trajectory, and relation between them early in the disease process is unclear. Therefore, to investigate very early-stage changes, we compared 158 healthy midlife adults with and without inherited AD predisposition (APOE4 carriership (38% positive), parental family history (FH) of dementia (54% positive)) on markers of SVD (white matter hyperintensities (WMH), cerebral microbleeds), and inflammation (C-reactive protein (CRP), fibrinogen), cross-sectionally and longitudinally over two years. While WMH severity was comparable between groups at baseline, longitudinal progression of WMH was greater in at-risk groups (APOE4+ and FH+). Topographically, APOE4 was associated exclusively with deep, but not periventricular, WMH progression after adjusting for FH. Conversely, APOE4 carriers displayed lower CRP levels than noncarriers, but not fibrinogen. Furthermore, interaction analysis showed that FH moderated the effect of SVD and inflammation on reaction time, an early feature of SVD, but not episodic memory or executive function. Findings suggest that vascular and inflammatory changes could occur decades before dementia onset, and may be of relevance in predicting incipient clinical progression.

Limited One-time Sampling Irregularity Map (LOTS-IM) for Automatic Unsupervised Assessment of White Matter Hyperintensities and Multiple Sclerosis Lesions in Structural Brain Magnetic Resonance Images

We present the application of limited one-time sampling irregularity map (LOTS-IM): a fully automatic unsupervised approach to extract brain tissue irregularities in magnetic resonance images (MRI), for quantitatively assessing white matter hyperintensities (WMH) of presumed vascular origin, and multiple sclerosis (MS) lesions and their progression. LOTS-IM generates an irregularity map (IM) that represents all voxels as irregularity values with respect to the ones considered "normal". Unlike probability values, IM represents both regular and irregular regions in the brain based on the original MRI's texture information. We evaluated and compared the use of IM for WMH and MS lesions segmentation on T2-FLAIR MRI with the state-of-the-art unsupervised lesions' segmentation method, Lesion Growth Algorithm from the public toolbox Lesion Segmentation Toolbox (LST-LGA), with several well established conventional supervised machine learning schemes and with state-of-the-art supervised deep learning methods for WMH segmentation. In our experiments, LOTS-IM outperformed unsupervised method LST-LGA on WMH segmentation, both in performance and processing speed, thanks to the limited one-time sampling scheme and its implementation on GPU. Our method also outperformed supervised conventional machine learning algorithms (i.e., support vector machine (SVM) and random forest (RF)) and deep learning algorithms (i.e., deep Boltzmann machine (DBM) and convolutional encoder network (CEN)), while yielding comparable results to the convolutional neural network schemes that rank top of the algorithms developed up to date for this purpose (i.e., UResNet and UNet). LOTS-IM also performed well on MS lesions segmentation, performing similar to LST-LGA. On the other hand, the high sensitivity of IM on depicting signal change deems suitable for assessing MS progression, although care must be taken with signal changes not reflective of a true pathology.

Association between Striatal Brain Iron Deposition, Microbleeds and Cognition 1 Year After a Minor Ischaemic Stroke

Brain iron deposits (IDs) are inversely associated with cognitive function in community-dwelling older people, but their association with cognition after ischemic stroke, and whether that differs from microbleeds, is unknown. We quantified basal ganglia IDs (BGID) and microbleeds (BMBs) semi-automatically on brain magnetic resonance images from patients with minor stroke (NIHSS < 7), at presentation and 12 months after stroke. We administered the National Adult Reading Test (NART, estimates premorbid or peak adult cognition) and the Revised Addenbrooke's Cognitive Examination (ACE-R; current cognition) at 1 and 12 months after stroke. We adjusted analyses for baseline cognition, age, gender, white matter hyperintensity (WMH) volume and vascular risk factors. In 200 patients, mean age 65 years, striatal IDs and BMBs volumes did not change over the 12 months. Baseline BGID volumes correlated positively with NART scores at both times (ρ = 0.19, p < 0.01). Baseline and follow-up BGID volumes correlated positively with age (ρ = 0.248, p < 0.001 and ρ = 0.271, p < 0.001 respectively), but only baseline (and not follow-up) BMB volume correlated with age (ρ = 0.129, p < 0.05). Both smoking and baseline WMH burden predicted verbal fluency and visuospatial abilities scores (B = -1.13, p < 0.02 and B = -0.22, p = 0.001 respectively) at 12 months after stroke. BGIDs and BMBs are associated differently with cognition post-stroke; studies of imaging and post-stroke cognition should adjust for premorbid cognition. The positive correlation of BGID with NART may reflect the lower premorbid cognition in patients with stroke at younger vs older ages.