Ahmed Abdulkadir

3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation

This paper introduces a network for volumetric segmentation that learns from sparsely annotated volumetric images. We outline two attractive use cases of this method: (1) In a semi-automated setup, the user annotates some slices in the volume to be segmented. The network learns from these sparse annotations and provides a dense 3D segmentation. (2) In a fully-automated setup, we assume that a representative, sparsely annotated training set exists. Trained on this data set, the network densely segments new volumetric images. The proposed network extends the previous u-net architecture from Ronneberger et al. by replacing all 2D operations with their 3D counterparts. The implementation performs on-the-fly elastic deformations for efficient data augmentation during training. It is trained end-to-end from scratch, i.e., no pre-trained network is required. We test the performance of the proposed method on a complex, highly variable 3D structure, the Xenopus kidney, and achieve good results for both use cases.

Diagnostic neuroimaging across diseases

Fully automated classification algorithms have been successfully applied to diagnose a wide range of neurological and psychiatric diseases. They are sufficiently robust to handle data from different scanners for many applications and in specific cases outperform radiologists. This article provides an overview of current applications taking structural imaging in Alzheimers Disease and schizophrenia as well as functional imaging to diagnose depression as examples. In this context, we also report studies aiming to predict the future course of the disease and the response to treatment for the individual. This has obvious clinical relevance but is also important for the design of treatment studies that may aim to include a cohort with a predicted fast disease progression to be more sensitive to detect treatment effects. In the second part, we present our own opinions on i) the role these classification methods can play in the clinical setting; ii) where their limitations are at the moment and iii) how those can be overcome. Specifically, we discuss strategies to deal with disease heterogeneity, diagnostic uncertainties, a probabilistic framework for classification and multi-class classification approaches.

Standardized evaluation of algorithms for computer-aided diagnosis of dementia based on structural MRI: The CADDementia challenge

Algorithms for computer-aided diagnosis of dementia based on structural MRI have demonstrated high performance in the literature, but are difficult to compare as different data sets and methodology were used for evaluation. In addition, it is unclear how the algorithms would perform on previously unseen data, and thus, how they would perform in clinical practice when there is no real opportunity to adapt the algorithm to the data at hand. To address these comparability, generalizability and clinical applicability issues, we organized a grand challenge that aimed to objectively compare algorithms based on a clinically representative multi-center data set. Using clinical practice as the starting point, the goal was to reproduce the clinical diagnosis. Therefore, we evaluated algorithms for multi-class classification of three diagnostic groups: patients with probable Alzheimers disease, patients with mild cognitive impairment and healthy controls. The diagnosis based on clinical criteria was used as reference standard, as it was the best available reference despite its known limitations. For evaluation, a previously unseen test set was used consisting of 354 T1-weighted MRI scans with the diagnoses blinded. Fifteen research teams participated with a total of 29 algorithms. The algorithms were trained on a small training set (n=30) and optionally on data from other sources (e.g., the Alzheimers Disease Neuroimaging Initiative, the Australian Imaging Biomarkers and Lifestyle flagship study of aging). The best performing algorithm yielded an accuracy of 63.0% and an area under the receiver-operating-characteristic curve (AUC) of 78.8%. In general, the best performances were achieved using feature extraction based on voxel-based morphometry or a combination of features that included volume, cortical thickness, shape and intensity. The challenge is open for new submissions via the web-based framework: http://caddementia.grand-challenge.org.

Gray matter atrophy pattern in elderly with subjective memory impairment.

Individuals with subjective memory impairment (SMI) report worsening of memory without impairment in cognitive tests. Despite normal cognitive performance, they may be at higher risk of cognitive decline compared with individuals without SMI.

Insomnia Does Not Appear to be Associated With Substantial Structural Brain Changes

STUDY OBJECTIVES Sleep has been demonstrated to significantly modulate brain plasticity and the manifestation of mental disorders. However, previous studies on the effect of disrupted sleep on brain structure have reported inconsistent results. The goal of the current study was to investigate brain morphometry in a well-characterized large sample of patients with primary insomnia (PI) in comparison with good sleeper controls. DESIGN Automated parcellation and pattern recognition approaches were supplemented by voxelwise analyses of gray and white matter volumes to analyze magnetic resonance images. All analyses included age, sex, and total intracranial volume as covariates. SETTING Department of Psychiatry and Psychotherapy of the University of Freiburg Medical Center. PARTICIPANTS There were 28 patients with PI (10 males; 18 females; age 43.7 ± 14.2 y) and 38 healthy, good sleepers (17 males; 21 females; age 39.6 ± 8.9 y). INTERVENTIONS N/A. RESULTS No significant between-group differences were observed in any of the investigated brain morphometry variables. CONCLUSIONS Altered brain function in insomnia does not appear to have a substantial effect on brain morphometry on a macroscopic level.

Effects of hardware heterogeneity on the performance of SVM Alzheimer's disease classifier

Fully automated machine learning methods based on structural magnetic resonance imaging (MRI) data can assist radiologists in the diagnosis of Alzheimers disease (AD). These algorithms require large data sets to learn the separation of subjects with and without AD. Training and test data may come from heterogeneous hardware settings, which can potentially affect the performance of disease classification. A total of 518 MRI sessions from 226 healthy controls and 191 individuals with probable AD from the multicenter Alzheimers Disease Neuroimaging Initiative (ADNI) were used to investigate whether grouping data by acquisition hardware (i.e. vendor, field strength, coil system) is beneficial for the performance of a support vector machine (SVM) classifier, compared to the case where data from different hardware is mixed. We compared the change of the SVM decision value resulting from (a) changes in hardware against the effect of disease and (b) changes resulting simply from rescanning the same subject on the same machine. Maximum accuracy of 87% was obtained with a training set of all 417 subjects. Classifiers trained with 95 subjects in each diagnostic group and acquired with heterogeneous scanner settings had an empirical detection accuracy of 84.2±2.4% when tested on an independent set of the same size. These results mirror the accuracy reported in recent studies. Encouragingly, classifiers trained on images acquired with homogenous and heterogeneous hardware settings had equivalent cross-validation performances. Two scans of the same subject acquired on the same machine had very similar decision values and were generally classified into the same group. Higher variation was introduced when two acquisitions of the same subject were performed on two scanners with different field strengths. The variation was unbiased and similar for both diagnostic groups. The findings of the study encourage the pooling of data from different sites to increase the number of training samples and thereby improving performance of disease classifiers. Although small, a change in hardware could lead to a change of the decision value and thus diagnostic grouping. The findings of this study provide estimators for diagnostic accuracy of an automated disease diagnosis method involving scans acquired with different sets of hardware. Furthermore, we show that the level of confidence in the performance estimation significantly depends on the size of the training sample, and hence should be taken into account in a clinical setting.

An evaluation of volume-based morphometry for prediction of mild cognitive impairment and Alzheimer's disease

Voxel-based morphometry from conventional T1-weighted images has proved effective to quantify Alzheimers disease (AD) related brain atrophy and to enable fairly accurate automated classification of AD patients, mild cognitive impaired patients (MCI) and elderly controls. Little is known, however, about the classification power of volume-based morphometry, where features of interest consist of a few brain structure volumes (e.g. hippocampi, lobes, ventricles) as opposed to hundreds of thousands of voxel-wise gray matter concentrations. In this work, we experimentally evaluate two distinct volume-based morphometry algorithms (FreeSurfer and an in-house algorithm called MorphoBox) for automatic disease classification on a standardized data set from the Alzheimers Disease Neuroimaging Initiative. Results indicate that both algorithms achieve classification accuracy comparable to the conventional whole-brain voxel-based morphometry pipeline using SPM for AD vs elderly controls and MCI vs controls, and higher accuracy for classification of AD vs MCI and early vs late AD converters, thereby demonstrating the potential of volume-based morphometry to assist diagnosis of mild cognitive impairment and Alzheimers disease.

A comparison of different automated methods for the detection of white matter lesions in MRI data

White matter hyperintensities (WMH) are the focus of intensive research and have been linked to cognitive impairment and depression in the elderly. Cumbersome manual outlining procedures make research on WMH labour intensive and prone to subjective bias. This study compares fully automated supervised detection methods that learn to identify WMH from manual examples against unsupervised approaches on the combination of FLAIR and T1 weighted images. Data were collected from ten subjects with mild cognitive impairment and another set of ten individuals who fulfilled diagnostic criteria for dementia. Data were split into balanced groups to create a training set used to optimize the different methods. Manual outlining served as gold standard to evaluate performance of the automated methods that identified each voxel either as intact or as part of a WMH. Otsus approach for multiple thresholds which is based only on voxel intensities of the FLAIR image produced a high number of false positives at grey matter boundaries. Performance on an independent test set was similarly disappointing when simply applying a threshold to the FLAIR that was found from training data. Among the supervised methods, precision-recall curves of support vector machines (SVM) indicated advantages over the performance achieved by K-nearest-neighbor classifiers (KNN). The curves indicated a clear benefit from optimizing the threshold of the SVM decision value and the voting rule of the KNN. Best performance was reached by selecting training voxels according to their distance to the lesion boundary and repeated training after replacing the feature vectors from those voxels that did not form support vectors of the SVM. The study demonstrates advantages of SVM for the problem of detecting WMH at least for studies that include only FLAIR and T1 weighted images. Various optimization strategies are discussed and compared against each other.

Interregional compensatory mechanisms of motor functioning in progressing preclinical neurodegeneration

Understanding brain reserve in preclinical stages of neurodegenerative disorders allows determination of which brain regions contribute to normal functioning despite accelerated neuronal loss. Besides the recruitment of additional regions, a reorganisation and shift of relevance between normally engaged regions are a suggested key mechanism. Thus, network analysis methods seem critical for investigation of changes in directed causal interactions between such candidate brain regions. To identify core compensatory regions, fifteen preclinical patients carrying the genetic mutation leading to Huntingtons disease and twelve controls underwent fMRI scanning. They accomplished an auditory paced finger sequence tapping task, which challenged cognitive as well as executive aspects of motor functioning by varying speed and complexity of movements. To investigate causal interactions among brain regions a single Dynamic Causal Model (DCM) was constructed and fitted to the data from each subject. The DCM parameters were analysed using statistical methods to assess group differences in connectivity, and the relationship between connectivity patterns and predicted years to clinical onset was assessed in gene carriers. In preclinical patients, we found indications for neural reserve mechanisms predominantly driven by bilateral dorsal premotor cortex, which increasingly activated superior parietal cortices the closer individuals were to estimated clinical onset. This compensatory mechanism was restricted to complex movements characterised by high cognitive demand. Additionally, we identified task-induced connectivity changes in both groups of subjects towards pre- and caudal supplementary motor areas, which were linked to either faster or more complex task conditions. Interestingly, coupling of dorsal premotor cortex and supplementary motor area was more negative in controls compared to gene mutation carriers. Furthermore, changes in the connectivity pattern of gene carriers allowed prediction of the years to estimated disease onset in individuals. Our study characterises the connectivity pattern of core cortical regions maintaining motor function in relation to varying task demand. We identified connections of bilateral dorsal premotor cortex as critical for compensation as well as task-dependent recruitment of pre- and caudal supplementary motor area. The latter finding nicely mirrors a previously published general linear model-based analysis of the same data. Such knowledge about disease specific inter-regional effective connectivity may help identify foci for interventions based on transcranial magnetic stimulation designed to stimulate functioning and also to predict their impact on other regions in motor-associated networks.

Applying Automated MR-Based Diagnostic Methods to the Memory Clinic: A Prospective Study.

Abstract Several studies have demonstrated that fully automated pattern recognition methods applied to structural magnetic resonance imaging (MRI) aid in the diagnosis of dementia, but these conclusions are based on highly preselected samples that significantly differ from that seen in a dementia clinic. At a single dementia clinic, we evaluated the ability of a linear support vector machine trained with completely unrelated data to differentiate between Alzheimer’s disease (AD), frontotemporal dementia (FTD), Lewy body dementia, and healthy aging based on 3D-T1 weighted MRI data sets. Furthermore, we predicted progression to AD in subjects with mild cognitive impairment (MCI) at baseline and automatically quantified white matter hyperintensities from FLAIR-images. Separating additionally recruited healthy elderly from those with dementia was accurate with an area under the curve (AUC) of 0.97 (according to Fig. 4). Multi-class separation of patients with either AD or FTD from other included groups was good on the training set (AUC >  0.9) but substantially less accurate (AUC = 0.76 for AD, AUC = 0.78 for FTD) on 134 cases from the local clinic. Longitudinal data from 28 cases with MCI at baseline and appropriate follow-up data were available. The computer tool discriminated progressive from stable MCI with AUC = 0.73, compared to AUC = 0.80 for the training set. A relatively low accuracy by clinicians (AUC = 0.81) illustrates the difficulties of predicting conversion in this heterogeneous cohort. This first application of a MRI-based pattern recognition method to a routine sample demonstrates feasibility, but also illustrates that automated multi-class differential diagnoses have to be the focus of future methodological developments and application studies