Search

Introduction: Early markers of neurodegeneration provide an opportunity to detect, monitor, and initiate interventions in individuals who have an increased risk of developing dementia. Here, we investigated whether the Timed Up and Go (TUG) test is associated with early brain neurodegeneration and whether the TUG test could be a marker of cognitive decline in people with subjective cognitive decline (SCD). Methods: This is a longitudinal analysis of the Dementia Disease Initiation Study, a prospective, community-based, cohort study from Norway, designed to investigate early markers of cognitive impairment and dementia. Participants were classified as SCD and healthy controls (HC). The main studied variables were the TUG test and cognition as measured by the Mini-Mental State Examination and the Consortium to Establish a Registry for Alzheimer’s Disease memory composite score. Additionally, we investigated the cross-sectional association of brain morphology with the TUG using 1.5T-MRI. Results: The sample included 45 participants (SCD = 21, HC = 24) followed during a mean time of 1.50 ± 0.70 years. At baseline, the cognitive performance did not differ between the groups, but TUG was longer in SCD. Slower baseline TUG was associated with a faster cognitive decline in both groups and it was also associated with reduced cortical thickness especially in motor, executive, associative, and somatosensory cortical regions in people with SCD. Discussion/Conclusion: TUG predicted cognitive change in individuals with SCD, and there was a negative association between TUG and cortical thickness. TUG is a promising cheap and noninvasive marker of early cognitive decline and may help initiate interventions in individuals who have an increased risk of dementia.

Objectives Sex differences permeate many aspects of dementia with Lewy bodies (DLB), including epidemiology, pathogenesis, disease progression, and symptom manifestation. However, less is known about potential sex differences in patterns of neurodegeneration in DLB. Here, we test whether grey matter networks also differ between female and male DLB patients. To assess the specificity of these sex differences to DLB, we additionally investigate sex differences in healthy controls (HCs). Methods A total of 119 (68.7 ± 8.4 years) male and 45 female (69.9 ± 9.1 years) DLB patients from three European centres and the Mayo Clinic were included in this study. Additionally, we included 119 male and 45 female age-matched HCs from the Mayo Clinic. Grey matter volumes of 58 cortical, subcortical, cerebellar, and pontine brain regions derived from structural magnetic resonance images were corrected for age, intracranial volume, and centre. Sex-specific grey matter networks for DLB patients and HCs were constructed by correlating each pair of brain regions. Network properties of the correlation matrices were compared between sexes and groups. Additional analyses were conducted on W-scored data to identify DLB-specific findings. Results Networks of male HCs and male DLB patients were characterised by a lower nodal strength compared to their respective female counterparts. In comparison to female HCs, the grey matter networks of male HCs showed a higher global efficiency, modularity, and a lower number of modules. None of the global and nodal network measures showed significant sex differences in DLB. Conclusions The disappearance of sex differences in the structural grey matter networks of DLB patients compared to HCs may indicate a sex-dependent network vulnerability to the alpha-synuclein pathology in DLB. Future studies might investigate whether the differences in structural network measures are associated with differences in cognitive scores and clinical symptoms between the sexes.

Background and objectives. Sex is an important contributing factor to neuroimaging phenotypes in brain disorders. However, little is known about the contribution of sex differences to the neurodegeneration in dementia with Lewy bodies (DLB). We investigated sex differences in probable DLB patients by using both visual rating scales of lobar atrophy and automated estimations of regional atrophy. Methods. We included 442 probable DLB patients from the European-DLB consortium and the Mayo Clinic who have magnetic resonance imaging (MRI) data available. We assessed sex differences and the sex-by-age interaction in two largely independent samples through visual rating scales of lobar atrophy (n = 333; mean age 73 ± 8 years, 62% males) and automated regional estimations of gray matter (GM) volume and mean cortical thickness (CTh) (n = 165; mean age 69 ± 9 years, 72% males). We used binary logistic regression and ANOVA for statistical analysis. Results. We found a statistically significantly higher likelihood of frontal atrophy measured by the global cortical atrophy-frontal subscale (GCA-F) in males (40% of males had an abnormal GCA-F score versus 29% of females, P-value = 0.006). Using automated estimations, we found smaller GM volumes in 6 cortical regions in males compared with females, as well as smaller GM volume in the entorhinal cortex and thinner olfactory cortices in females, compared with males. The sex-by-age interaction showed statistically significant results in 6 cortical volumes and 7 mean CTh estimations (P-value ≤ 0.05), accentuated in the right middle frontal gyrus (FDR-adjusted P-value = 0.047). These cross-sectional interactions indicated that while females have statistically significantly less atrophy than males at younger ages, differences become non-significant at older ages, with females showing the same level of atrophy than males around the age of 75. Conclusions. This study demonstrates sex differences on brain atrophy in probable DLB. While male DLB patients have a more widespread pattern of cortical atrophy at younger ages, these sex differences tend to disappear with increasing age. Longitudinal studies will help establish these cross-sectional findings and inform on sex and age considerations to the use of MRI in clinical routine, as the field moves towards precision medicine.

We investigated whether cerebrovascular disease contributes to neurodegeneration and clinical phenotype in dementia with Lewy bodies (DLB). Regional cortical thickness and subcortical gray matter volumes were estimated from structural magnetic resonance imaging (MRI) in 165 DLB patients. Cortical and subcortical infarcts were recorded and white matter hyperintensities (WMHs) were assessed. Subcortical only infarcts were more frequent (13.3%) than cortical only infarcts (3.1%) or both subcortical and cortical infarcts (2.4%). Infarcts, irrespective of type, were associated with WMHs. A higher WMH volume was associated with thinner orbitofrontal, retrosplenial, and posterior cingulate cortices, smaller thalamus and pallidum, and larger caudate volume. A higher WMH volume was associated with the presence of visual hallucinations and lower global cognitive performance, and tended to be associated with the absence of probable rapid eye movement sleep behavior disorder. Presence of infarcts was associated with the absence of parkinsonism. We conclude that cerebrovascular disease is associated with gray matter neurodegeneration in patients with probable DLB, which may have implications for the multifactorial treatment of probable DLB.

Background. Dementia with Lewy bodies (DLB) is a neurodegenerative disorder with a wide heterogeneity of symptoms, which suggests the existence of different subtypes. Objectives. We used data-driven analysis of MRI data to investigate DLB subtypes. Methods. We included 165 probable DLB from the Mayo Clinic and 3 centers from the European DLB consortium and performed a hierarchical cluster analysis to identify subtypes based on gray matter (GM) volumes. To characterize the subtypes, we used demographic and clinical data, as well as β-amyloid and cerebrovascular biomarkers at baseline, and cognitive decline over three years. Results. We identified 3 subtypes: an older subtype with reduced cortical GM volumes, worse cognition, and faster cognitive decline (n=49, 30%); a subtype with low GM volumes in fronto-occipital regions (n=76, 46%); and a subtype of younger patients with the highest cortical GM volumes, proportionally lower GM volumes in basal ganglia and the highest frequency of cognitive fluctuations (n=40, 24%). Conclusions. This study shows the existence of MRI subtypes in DLB, which may have implications for clinical workout, research, and therapeutic decisions.

Importance Identifying nutritional compounds which can reduce cognitive decline in older people is a hugely important topic. Objective To study the safety and effect of anthocyanins in maintaining cognitive functioning in people at increased risk for dementia. Design, setting, and participants Participants (206 individuals, aged 60–80 years) diagnosed with either mild cognitive impairment (MCI) or two or more cardiometabolic disorders (i.e., diabetes, hypertension, obesity) were enrolled at three different centres in Norway. Intervention Participants were randomly assigned to four capsules with a total of 320 mg/d of naturally purified anthocyanins or placebo 1:1 for 24 weeks. Main outcomes and measures The primary outcome was the Quality of Episodic Memory composite measure (0–100) from an online cognitive test battery CogTrack, which was administered at baseline and monthly for the next 24 weeks. Secondary outcomes included other cognitive scores from the CogTrack battery. We applied mixed effects models with a baseline test score, group, time and their interaction as fixed effects, as well as other predefined baseline covariates. The primary comparison was the group difference at week 24 based on a modified intention-to-treat principle. Results : The primary analysis did not show a significant group difference at 24 weeks (78.2 versus 76.8; adjusted mean difference 1.4 (95% confidence interval -0.9–3.7); effect size 0.15; p = 0.23). However, there was a significant difference in slopes during weeks 8–24 (p = 0.007); the anthocyanin group improved while the placebo group worsened. No differences were found for the secondary cognitive outcomes. Anthocyanin capsules were well-tolerated and safe to use. Conclusion Anthocyanin supplementation for 24 weeks was safe and well tolerated in people with MCI or cardiometabolic disorders. We found no significant group difference in episodic memory at the end of the study but statistically significant differences in slopes. Further studies are warranted to explore whether anthocyanins supplementation can reduce cognitive decline in people at increased risk of dementia. Trial registration ClinicalTrials.gov, (Identifier NCT03419039). http://www.clinicaltrials.gov/, NCT03419039.

ObjectiveIn a multicenter cohort of probable dementia with Lewy bodies (DLB), we tested the hypothesis that β-amyloid and tau biomarker positivity increases with age, which is modified by APOE genotype and sex, and that there are isolated and synergistic associations with the clinical phenotype.MethodsWe included 417 patients with DLB (age 45–93 years, 31% women). Positivity on β-amyloid (A+) and tau (T+) biomarkers was determined by CSF β-amyloid1-42 and phosphorylated tau in the European cohort and by Pittsburgh compound B and AV-1451 PET in the Mayo Clinic cohort. Patients were stratified into 4 groups: A−T−, A+T−, A−T+, and A+T+.ResultsA−T− was the largest group (39%), followed by A+T− (32%), A+T+ (15%), and A−T+ (13%). The percentage of A−T− decreased with age, and A+ and T+ increased with age in both women and men. A+ increased more in APOE ε4 carriers with age than in noncarriers. A+ was the main predictor of lower cognitive performance when considered together with T+. T+ was associated with a lower frequency of parkinsonism and probable REM sleep behavior disorder. There were no significant interactions between A+ and T+ in relation to the clinical phenotype.ConclusionsAlzheimer disease pathologic changes are common in DLB and are associated with the clinical phenotype. β-Amyloid is associated with cognitive impairment, and tau pathology is associated with lower frequency of clinical features of DLB. These findings have important implications for diagnosis, prognosis, and disease monitoring, as well as for clinical trials targeting disease-specific proteins in DLB.Classification of evidenceThis study provides Class II evidence that in patients with probable DLB, β-amyloid is associated with lower cognitive performance and tau pathology is associated with lower frequency of clinical features of DLB.

IMPORTANCE The ENIGMA clinical high risk (CHR) for psychosis initiative, the largest pooled neuroimaging sample of individuals at CHR to date, aims to discover robust neurobiological markers of psychosis risk.OBJECTIVE To investigate baseline structural neuroimaging differences between individuals at CHR and healthy controls as well as between participants at CHR who later developed a psychotic disorder (CHR-PS+) and those who did not (CHR-PS-).DESIGN, SETTING, AND PARTICIPANTS In this case-control study, baseline T1-weighted magnetic resonance imaging (MRI) data were pooled from 31 international sites participating in the ENIGMA Clinical High Risk for Psychosis Working Group. CHR status was assessed using the Comprehensive Assessment of At-Risk Mental States or Structured Interview for Prodromal Syndromes. MRI scans were processed using harmonized protocols and analyzed within a mega-analysis and meta-analysis framework from January to October 2020.MAIN OUTCOMES AND MEASURES Measures of regional cortical thickness (CT), surface area, and subcortical volumes were extracted from T1-weighted MRI scans. Independent variables were group (CHR group vs control group) and conversion status (CHR-PS+ group vs CHR-PS- group vs control group).RESULTS Of the 3169 included participants, 1428 (45.1%) were female, and the mean (SD; range) age was 21.1 (4.9; 9.5-39.9) years. This study included 1792 individuals at CHR and 1377 healthy controls. Using longitudinal clinical information, 253 in the CHR-PS+ group, 1234 in the CHR-PS- group, and 305 at CHR without follow-up data were identified. Compared with healthy controls, individuals at CHR exhibited widespread lower CT measures (mean [range] Cohen d = -0.13 [-0.17 to -0.09]), but not surface area or subcortical volume. Lower CT measures in the fusiform, superior temporal, and paracentral regions were associated with psychosis conversion (mean Cohen d = -0.22; 95% CI, -0.35 to 0.10). Among healthy controls, compared with those in the CHR-PS+ group, age showed a stronger negative association with left fusiform CT measures (F = 9.8; P < .001; q < .001) and left paracentral CT measures (F = 5.9; P = .005; q = .02). Effect sizes representing lower CT associated with psychosis conversion resembled patterns of CT differences observed in ENIGMA studies of schizophrenia (rho = 0.35; 95% CI, 0.12 to 0.55; P = .004) and individuals with 22q11.2 microdeletion syndrome and a psychotic disorder diagnosis (rho = 0.43; 95% CI, 0.20 to 0.61; P = .001).CONCLUSIONS AND RELEVANCE This study provides evidence for widespread subtle, lower CT measures in individuals at CHR. The pattern of CT measure differences in those in the CHR-PS+ group was similar to those reported in other large-scale investigations of psychosis. Additionally, a subset of these regions displayed abnormal age associations. Widespread disruptions in CT coupled with abnormal age associations in those at CHR may point to disruptions in postnatal brain developmental processes.Question How are brain morphometric features associated with later psychosis conversion in individuals at clinical high risk (CHR) for developing psychosis?Findings In this case-control study including 3169 participants, lower cortical thickness, but not cortical surface area or subcortical volume, was more pronounced in individuals at CHR in a manner highly consistent with thinner cortex in individuals with established psychosis. Regions that displayed lower cortical thickness in individuals at CHR who later developed a psychotic disorder additionally displayed abnormal associations with age.Meaning In this study, CHR status and later transition to psychosis was robustly associated with lower cortical thickness; abnormal age associations and specificity to cortical thickness may point to aberrant postnatal brain development in individuals at CHR, including pruning and myelination.This case-control study investigates baseline structural magnetic resonance imaging (MRI) differences between individuals at clinical high risk and healthy controls as well as between participants at clinical high risk who later developed a psychotic disorder and those who did not.

Introduction: We explored regional brain atrophy patterns and their clinical correlates in dementia with Lewy bodies (DLB). Methods: In this multicentre study, we included a total of 333 patients with DLB, 352 patients with Alzheimer's disease (AD), and 233 normal controls and used medial temporal lobe atrophy, posterior atrophy, and frontal atrophy (GCA-F) visual rating scales. Patients were classified according to four atrophy patterns. Results: Patients with DLB had higher scores on all the three atrophy scales than normal controls but had less medial temporal lobe atrophy than those with AD (all P values

Malnutrition is highly prevalent in older persons with dementia. Therefore, strong predictors of malnutrition in this population are crucial to initiating early interventions. This study evaluates the association between the probability of having malnutrition with the muscle volume and intramuscular fat (iMAT) of the masseter and the tongue in magnetic resonance imaging (MRI) of community-dwelling older persons diagnosed with mild dementia followed up for 5 years. This is a longitudinal study conducted in the western part of Norway. Muscle volume and iMAT of the tongue and masseter were computed from structural head MRI obtained from 65 participants of the Dementia Study of Western Norway using Slice-O-Matic software for segmentation. Malnutrition was assessed using the Global Leadership Initiative on Malnutrition Index. Linear mixed models were conducted. Having malnutrition at baseline was associated with lower muscle volume (odds ratio [OR] 0.60, standard error [SE] 0.20; p = .010) and higher iMAT (OR 3.31, SE 0.46; p = .010) in the tongue. At 5 years follow-up, those with lower muscle volume (OR 0.55, SE 0.20; p = .002) and higher iMAT (OR 2.52, SE 0.40; p = .022) in the tongue had a higher probability of presenting malnutrition. The masseter iMAT and volume were not associated with malnutrition in any of the adjusted models. In people diagnosed with mild dementia, tongue muscle volume and iMAT were associated with baseline malnutrition and the probability of developing malnutrition in a 5-year trajectory. In the masseter, there were no significant associations after adjustments.

ImportanceThe ENIGMA clinical high risk for psychosis (CHR) initiative, the largest pooled CHR-neuroimaging sample to date, aims to discover robust neurobiological markers of psychosis risk in a sample with known heterogeneous outcomes.ObjectiveWe investigated baseline structural neuroimaging differences between CHR subjects and healthy controls (HC), and between CHR participants who later developed a psychotic disorder (CHR-PS+) and those who did not (CHR-PS-). We assessed associations with age by group and conversion status, and similarities between the patterns of effect size maps for psychosis conversion and those found in other large-scale psychosis studies.Design, Setting, and ParticipantsBaseline T1-weighted MRI data were pooled from 31 international sites participating in the ENIGMA CHR Working Group. MRI scans were processed using harmonized protocols and analyzed within a mega- and meta-analysis framework from January-October 2020.Main Outcome(s) and Measure(s)Measures of regional cortical thickness (CT), surface area (SA), and subcortical volumes were extracted from T1-weighted MRI scans. Independent variables were group (CHR, HC) and conversion status (CHR-PS+, CHR-PS-, HC).ResultsThe final dataset consisted of 3,169 participants (CHR=1,792, HC=1,377, age range: 9.5 to 39.8 years, 45% female). Using longitudinal clinical information, we identified CHR-PS+ (N=253) and CHR-PS- (N=1,234). CHR exhibited widespread thinner cortex compared to HC (average d=-0.125, range: -0.09 to -0.17), but not SA or subcortical volume. Thinner cortex in the fusiform, superior temporal, and paracentral regions was associated with psychosis conversion (average d=-0.22). Age showed a stronger negative association with left fusiform and left paracentral CT in HC, compared to CHR-PS+. Regional CT psychosis conversion effect sizes resembled patterns of CT alterations observed in other ENIGMA studies of psychosis.Conclusions and RelevanceWe provide evidence for widespread subtle CT reductions in CHR. The pattern of regions displaying greater CT alterations in CHR-PS+ were similar to those reported in other large-scale investigations of psychosis. Additionally, a subset of these regions displayed abnormal age associations. Widespread CT disruptions coupled with abnormal age associations in CHR may point to disruptions in postnatal brain developmental processes.Key PointsQuestionHow do baseline brain morphometric features relate to later psychosis conversion in individuals at clinical high risk (CHR)?FindingsIn the largest coordinated international analysis to date, reduced baseline cortical thickness, but not cortical surface area or subcortical volume, was more pronounced in CHR, in a manner highly consistent with thinner cortex in established psychosis. Regions that displayed greater cortical thinning in future psychosis converters additionally displayed abnormal associations with age.MeaningCHR status and later transition to psychosis is robustly associated with reduced cortical thickness. Abnormal age associations and specificity to cortical thickness may point to aberrant postnatal brain development in CHR, including pruning and myelination.

Background: Hippocampal atrophy is presented in Alzheimer’s disease (AD) and dementia with Lewy bodies (DLB). Cognition, dual-tasks, muscular function, goal-related behaviors and neuropsychiatric symptoms are linked to hippocampal volumes and may lead to functional decline in activities of daily living. We examined the association between baseline hippocampal subfield volumes (HSv) in mild AD and DLB, and functional decline. Materials & methods: 12 HSv were computed from structural magnetic resonance images using Freesurfer 6.0 segmentation. Functional decline was assessed using the rapid disability rating scale score. Linear regressions were conducted. Results: In AD, HSv were smaller bilaterally. However, HSv were not associated with functional decline. Conclusion: Functional decline does not depend on HSv in mild AD and DLB.

BACKGROUND/OBJECTIVES Functional status is one of the most important markers of well-being in older adults, but the drivers of functional decline in dementia are not well known. The aim of our work was to study the association of neuropsychiatric symptoms (NPSs) with functional decline over 5 years in newly diagnosed people with Alzheimer´s disease (AD) and Lewy body dementia (LBD). DESIGN Secondary analysis of the Dementia Study of Western Norway longitudinal cohort study. SETTING Multicenter study conducted in memory clinics in western Norway. PARTICIPANTS We included a total of 196 patients newly diagnosed with AD (n = 111) and LBD (n = 85), followed up annually for 5 years. MAIN OUTCOMES AND MEASURES The outcome was the rapid disability rating scale (items 1-13). Linear mixed-effects models were used for analysis with the total score of the Norwegian Neuropsychiatric Inventory (NPI) as a predictor measured either at baseline or longitudinally, adjusted for potential confounders, including cognition. Effect modification was checked by introducing interactions with NPI score and stratifying by diagnosis. RESULTS The total NPI score longitudinal course was associated with functional decline in both AD and LBD. At baseline, the total NPI score predicted functional decline in AD. CONCLUSION NPSs were associated with the rate of functional decline in people with AD and LBD, independent of cognitive impairment. These results highlight the relevance of early detection and intervention of NPSs, which may also reduce functional decline. J Am Geriatr Soc 68:2257-2263, 2020.

Highlights • CSF Amyloid- β is related to medial temporal and posterior brain atrophy in Lewy body dementia. • CSF p-Tau is associated with posterior brain atrophy in Lewy body dementia. • CSF AD-related biomarkers are not related to atrophy in the frontal lobes., Background Alzheimer’s disease (AD)-related pathology is frequently found in patients with dementia with Lewy bodies (DLB). However, it is unknown how amyloid-β and tau-related pathologies influence neurodegeneration in DLB. Understanding the mechanisms underlying brain atrophy in DLB can improve our knowledge about disease progression, differential diagnosis, drug development and testing of anti-amyloid and anti-tau therapies in DLB. Objectives We aimed at investigating the combined effect of CSF amyloid-β42, phosphorylated tau and total tau on regional brain atrophy in DLB in the European DLB (E-DLB) cohort. Methods 86 probable DLB patients from the E-DLB cohort with CSF and MRI data were included. Random forest was used to analyze the association of CSF biomarkers (predictors) with visual rating scales for medial temporal lobe atrophy (MTA), posterior atrophy (PA) and global cortical atrophy scale-frontal subscale (GCA-F) (outcomes), including age, sex, education and disease duration as extra predictors. Results DLB patients with abnormal MTA scores had abnormal CSF Aβ42, shorter disease duration and older age. DLB patients with abnormal PA scores had abnormal levels of CSF Aβ42 and p-tau, older age, lower education and shorter disease duration. Abnormal GCA-F scores were associated with lower education, male sex, and older age, but not with any AD-related CSF biomarker. Conclusions This study shows preliminary data on the potential combined effect of amyloid-β and tau-related pathologies on the integrity of posterior brain cortices in DLB patients, whereas only amyloid-β seems to be related to MTA. Future availability of α-synuclein biomarkers will help us to understand the effect of α-synuclein and AD-related pathologies on brain integrity in DLB.

One of the most well-known and complex applications of artificial intelligence (AI) is Alzheimer’s detection, which lies in the field of medical imaging. The complexity in this task lies in the three-dimensional structure of the MRI scan images. In this paper, we propose to use 3D Convolutional Neural Networks (3D-CNN) for Alzheimer’s detection. 3D-CNNs have been a popular choice for this task. The novelty in our paper lies in the fact that we use a deeper 3D-CNN consisting of 10 layers. Also, with effectively training our model consisting of Batch Normalization layers that provide a regularizing effect, we don’t have to use any transfer learning. We also use the simple data augmentation technique of flipping. Our model is trained for binary classification that distinguishes between Alzheimer’s and normal, as well as multiclass classification consisting of Alzheimer’s, Mild Cognitive Impairment, and normal classes. We tested our model on the ADNI dataset and achieved 94.17% and 89.14% accuracy for binary classification and multiclass classification, respectively.

Dementia is an evolving challenge in society and early intervention is important. The ability to distinguish between different dementia and non-dementia early in course may be essential for successful patient care. Magnetic resonance (MR) imaging may aid as a noninvasive method to increase prediction accuracy. In this work we explored the use of two different 3D local binary pattern (LBP) texture features extracted from T1 MR images of the brain combined with a random forest classifier in an attempt to discern patients with Alzheimer's disease (AD), Lewy body dementia (LBD), and normal controls (NC). Analysis were conducted in areas with white matter lesions (WML) and normal appearing white matter (NAWM). We also calculated correlations between texture features and cognition measured by mini mental state examination (MMSE) controlling for age. Additionally, two different methods for handling the imbalanced data problem were tested, namely cost-sensitive classification and resampling of the data using the synthetic minority oversampling technique (SMOTE). Four different classification tasks were extensively tested, a three-class problem: AD vs. LBD vs. NC, a two-class problem: NC vs. AD, a two-class problem NC vs. LBD, and a two-class problem: AD vs. LBD. Results from 10 folds nested cross validation are reported as mean accuracy, precision, and recall with standard deviation in brackets. The two-class problems NC vs. AD and NC vs. LBD, show encouraging results with total accuracy of 0.97 (0.07) and 0.97 (0.06) respectively. The three-class problem and the two-class problem AD vs. LBD are not equally encouraging but shows higher accuracy than clinical diagnosis with a total accuracy of 0.79 (0.07) and 0.79 (0.15) respectively. Possible explanations may be that the AD- and LBD group are too similar concerning LBP texture analysis and that the LBD group is too small. Most of the texture features calculated for the AD subjects in the NAWM region were significantly correlated with cognition. Together with the positive classification results from the NAWM region this may suggest that the NAWM region is an important area for studying AD. Both cost-sensitive classification and resampling using SMOTE proved useful and improved the results considerably in many of the tests.

[No disponible]

Prostate Cancer (PCa) is one of the most common health threats in the developed world and the most diagnosed cancer among men in Norway. Detecting a lesion is a challenging task, and early diagnosis and treatment can be crucial for the patient. Computer-assisted systems using Deep Learning (DL) to analyze Magnetic Reso- nance Imaging (MRI) data could help improve diagnostic accuracy if used together with current practices. However, the DL systems require large amounts of annotated data which are rarely available. In addition, the presence of factors such as bias in the form of a higher prevalence of one class in classification tasks can harm the final results of the developed system. This thesis explores methods for performing anomaly detection on MRI image data of the prostate, where lesions are the anomalies. We tackle the lack of annotated data by using unsupervised learning and a type of neural network called Autoencoders (AEs). Specifically, Convolutional Autoencoders (CAEs), Variational Autoencoders (VAEs), and multi-view CAEs are compared. We evaluate the methods by means of Area Under the Curve (AUC) to measure the ability to separate the images with a lesion from those that do not. The results of this thesis show that using a multi-view CAE offers advantages over the single-view based CAE and VAE in terms of ROC-AUC and that a fully unsupervised approach holds potential for lesion detection in prostate MRI.

Prostate cancer is the second most occurring cancer and the sixth leading cause of cancer among men worldwide. The number of cases are expected to increase due to population growth and increase in expected lifetime. The screening test for prostate cancer is not without risk and there is chance of over-diagnosis with the traditional screening. The magnetic resonance imaging (MRI) examination is an essential tool and a comfortable method for diagnosing cancer. If some of the screening methods can be replaced by accurate MRI examinations it will be more comfortable for the patients. This thesis will explore how masked autoencoders with a generative self-supervised approach can improve diagnosis of prostate cancer from MRI images. Computer vision is used in many ways in the field of medical imaging. Hopefully computer vision based on deep learning and transformers can improve the field of medical imaging. However, the state of the art deep learning models require pre-training on huge amounts of unlabeled data and fine-tuning on large amounts of data. This thesis investigates if a masked vision transformer can learn to predict prostate cancer lesions on a limited amount of data. The final results suggest that a simple model of small size is not sufficient to for prediction prostate cancer.