Artificial neural networks have capacity to learn and store information about process faults via associative memory, and thus have an associative diagnostic ability with respect to faults that occur in a process. Knowledge of the faults to be learned by the network evolves from sets of data, namely values of steady-state process variables collected under normal operating condition and those collected under faulty conditions, together with information about the degree of the faults and their causes.Here, we describe how to apply artificial neural networks to fault diagnosis. A suitable two-stage multilayer neural network is proposed as the network to be used for diagnosis. The first stage of the network discriminates between the causes of faults when fed the noisy process measurements. Once the fault is identified, the second stage of the network estimates the degree of the fault. Thus, the diagnosis of incipient faults becomes possible.

No abstract

White matter bundle segmentation using diffusion MRI fiber tractography has become the method of choice to identify white matter fiber pathways in vivo in human brains. However, like other analyses of complex data, there is considerable variability in segmentation protocols and techniques. This can result in different reconstructions of the same intended white matter pathways, which directly affects tractography results, quantification, and interpretation. In this study, we aim to evaluate and quantify the variability that arises from different protocols for bundle segmentation. Through an open call to users of fiber tractography, including anatomists, clinicians, and algorithm developers, 42 independent teams were given processed sets of human wholebrain streamlines and asked to segment 14 white matter fascicles on six subjects. In total, we received 57 different bundle segmentation protocols, which enabled detailed volume-based and streamline-based analyses of agreement and disagreement among protocols for each fiber pathway. Results show that even when given the exact same sets of underlying streamlines, the variability across protocols for bundle segmentation is greater than all other sources of variability in the virtual dissection process, including variability within protocols and variability across subjects. In order to foster the use of tractography bundle dissection in routine clinical settings, and as a fundamental analytical tool, future endeavors must aim to resolve and reduce this heterogeneity. Although external validation is needed to verify the anatomical accuracy of bundle dissections, reducing heterogeneity is a step towards reproducible research and may be achieved through the use of standard nomenclature and definitions of white matter bundles and well-chosen constraints and decisions in the dissection process.

Purpose The reproducibility of neurite orientation dispersion and density imaging (NODDI) metrics in the human brain has not been explored across different magnetic resonance (MR) scanners from different vendors. This study aimed to evaluate the scanrescan and inter-vendor reproducibility of NODDI metrics in white and gray matter of healthy subjects using two 3-T MR scanners from two vendors. Methods Ten healthy subjects (7 males; mean age 30 ± 7 years, range 23-37 years) were included in the study. Whole-brain diffusion-weighted imaging was performed with b-values of 1000 and 2000 s/mm 2 using two 3-T MR scanners from two different vendors. Automatic extraction of the region of interest was performed to obtain NODDI metrics for whole and localized areas of white and gray matter. The coefficient of variation (CoV) and intraclass correlation coefficient (ICC) were calculated to assess the scan-rescan and inter-vendor reproducibilities of NODDI metrics. Results The scan-rescan and inter-vendor reproducibility of NODDI metrics (intracellular volume fraction and orientation dispersion index) were comparable with those of diffusion tensor imaging (DTI) metrics. However, the inter-vendor reproducibilities of NODDI (CoV = 2.3-14%) were lower than the scan-rescan reproducibility (CoV: scanner A = 0.8-3.8%; scanner B = 0.8-2.6%). Compared with the finding of DTI metrics, the reproducibility of NODDI metrics was lower in white matter and higher in gray matter. Conclusion The lower inter-vendor reproducibility of NODDI in some brain regions indicates that data acquired from different MRI scanners should be carefully interpreted.

No abstract