Encoding Stability into Laser Powder Bed Fusion Monitoring Using Temporal Features and Pore Density Modelling.

In laser powder bed fusion (LPBF), melt pool instability can lead to the development of pores in printed parts, reducing the part's structural strength. While camera-based monitoring systems have been introduced to improve melt pool stability, these systems only measure melt pool stability in limited, indirect ways. We propose that melt pool stability can be improved by explicitly encoding stability into LPBF monitoring systems through the use of temporal features and pore density modelling. We introduce the temporal features, in the form of temporal variances of common LPBF monitoring features (e.g., melt pool area, intensity), to explicitly quantify printing stability. Furthermore, we introduce a neural network model trained to link these video features directly to pore densities estimated from the CT scans of previously printed parts. This model aims to reduce the number of online printer interventions to only those that are required to avoid porosity. These contributions are then implemented in a full LPBF monitoring system and tested on prints using 316L stainless steel. Results showed that our explicit stability quantification improved the correlation between our predicted pore densities and true pore densities by up to 42%.

PAPPI: Personalized analysis of plantar pressure images using statistical modelling and parametric mapping.

Quantitative analyses of plantar pressure images typically occur at the group level and under the assumption that individuals within each group display homogeneous pressure patterns. When this assumption does not hold, a personalized analysis technique is required. Yet, existing personalized plantar pressure analysis techniques work at the image level, leading to results that can be unintuitive and difficult to interpret. To address these limitations, we introduce PAPPI: the Personalized Analysis of Plantar Pressure Images. PAPPI is built around the statistical modelling of the relationship between plantar pressures in healthy controls and their demographic characteristics. This statistical model then serves as the healthy baseline to which an individual's real plantar pressures are compared using statistical parametric mapping. As a proof-of-concept, we evaluated PAPPI on a cohort of 50 hallux valgus patients. PAPPI showed that plantar pressures from hallux valgus patients did not have a single, homogeneous pattern, but instead, 5 abnormal pressure patterns were observed in sections of this population. When comparing these patterns to foot pain scores (i.e. Foot Function Index, Manchester-Oxford Foot Questionnaire) and radiographic hallux angle measurements, we observed that patients with increased pressure under metatarsal 1 reported less foot pain than other patients in the cohort, while patients with abnormal pressures in the heel showed more severe hallux valgus angles and more foot pain. Also, incidences of pes planus were higher in our hallux valgus cohort compared to the modelled healthy controls. PAPPI helped to clarify recent discrepancies in group-level plantar pressure studies and showed its unique ability to produce quantitative, interpretable, and personalized analyses for plantar pressure images.

A Machine Learning Approach to Growth Direction Finding for Automated Planting of Bulbous Plants.

In agricultural robotics, a unique challenge exists in the automated planting of bulbous plants: the estimation of the bulb's growth direction. To date, no existing work addresses this challenge. Therefore, we propose the first robotic vision framework for the estimation of a plant bulb's growth direction. The framework takes as input three x-ray images of the bulb and extracts shape, edge, and texture features from each image. These features are then fed into a machine learning regression algorithm in order to predict the 2D projection of the bulb's growth direction. Using the x-ray system's geometry, these 2D estimates are then mapped to the 3D world coordinate space, where a filtering on the estimate's variance is used to determine whether the estimate is reliable. We applied our algorithm on 27,200 x-ray simulations from T. Apeldoorn bulbs on a standard desktop workstation. Results indicate that our machine learning framework is fast enough to meet industry standards (<0.1 seconds per bulb) while providing acceptable accuracy (e.g. error < 30° in 98.40% of cases using an artificial 3-layer neural network). The high success rates of the proposed framework indicate that it is worthwhile to proceed with the development and testing of a physical prototype of a robotic bulb planting system.

An assessment of the information lost when applying data reduction techniques to dynamic plantar pressure measurements.

Data reduction techniques are commonly applied to dynamic plantar pressure measurements, often prior to the measurement's analysis. In performing these data reductions, information is discarded from the measurement before it can be evaluated, leading to unkonwn consequences. In this study, we aim to provide the first assessment of what impact data reduction techniques have on plantar pressure measurements. Specifically, we quantify the extent to which information of any kind is discarded when performing common data reductions. Plantar pressure measurements were collected from 33 healthy controls, 8 Hallux Valgus patients, and 10 Metatarsalgia patients. Eleven common data reductions were then applied to the measurements, and the resulting datasets were compared to the original measurement in three ways. First, information theory was used to estimate the information content present in the original and reduced datasets. Second, principal component analysis was used to estimate the number of intrinsic dimensions present. Finally, a permutational multivariate ANOVA was performed to evaluate the significance of group differences between the healthy controls, Hallux Valgus, and Metatarsalgia groups. The evaluated data reductions showed a minimum of 99.1% loss in information content and losses of dimensionality between 20.8% and 83.3%. Significant group differences were also lost after each of the 11 data reductions (α=0.05), but these results may differ for other patient groups (especially those with highly-deformed footprints) or other region of interest definitions. Nevertheless, the existence of these results suggest that the diagnostic content of dynamic plantar pressure measurements is yet to be fully exploited.

Predictive connectome subnetwork extraction with anatomical and connectivity priors.

We present a new method to identify anatomical subnetworks of the human connectome that are optimally predictive of targeted clinical variables, developmental outcomes or disease states. Given a training set of structural or functional brain networks, derived from diffusion MRI (dMRI) or functional MRI (fMRI) scans respectively, our sparse linear regression model extracts a weighted subnetwork. By enforcing novel backbone network and connectivity based priors along with a non-negativity constraint, the discovered subnetworks are simultaneously anatomically plausible, well connected, positively weighted and reasonably sparse. We apply our method to (1) predicting the cognitive and neuromotor developmental outcomes of a dataset of 168 structural connectomes of preterm neonates, and (2) predicting the autism spectrum category of a dataset of 1013 resting-state functional connectomes from the Autism Brain Imaging Data Exchange (ABIDE) database. We find that the addition of each of our novel priors improves prediction accuracy and together outperform other state-of-the-art prediction techniques. We then examine the structure of the learned subnetworks in terms of topological features and with respect to established function and physiology of different regions of the brain.

STAPP: Spatiotemporal analysis of plantar pressure measurements using statistical parametric mapping.

BACKGROUND: Pedobarography produces large sets of plantar pressure samples that are routinely subsampled (e.g. using regions of interest) or aggregated (e.g. center of pressure trajectories, peak pressure images) in order to simplify statistical analysis and provide intuitive clinical measures. RESEARCH QUESTION: We hypothesize that these data reductions discard gait information that can be used to differentiate between groups or conditions. METHODS: To test the hypothesis of null information loss, we created an implementation of statistical parametric mapping (SPM) for dynamic plantar pressure datasets (i.e. plantar pressure videos). Our SPM software framework brings all plantar pressure videos into anatomical and temporal correspondence, then performs statistical tests at each sampling location in space and time. Novelly, we introduce non-linear temporal registration into the framework in order to normalize for timing differences within the stance phase. We refer to our software framework as STAPP: spatiotemporal analysis of plantar pressure measurements. Using STAPP, we tested our hypothesis on plantar pressure videos from 33 healthy subjects walking at different speeds. RESULTS: As walking speed increased, STAPP was able to identify significant decreases in plantar pressure at mid-stance from the heel through the lateral forefoot. The extent of these plantar pressure decreases has not previously been observed using existing plantar pressure analysis techniques. SIGNIFICANCE: We therefore conclude that the subsampling of plantar pressure videos - a task which led to the discarding of gait information in our study - can be avoided using STAPP.

Three-dimensional quantitative analysis of healthy foot shape: a proof of concept study.

BACKGROUND: Foot morphology has received increasing attention from both biomechanics researches and footwear manufacturers. Usually, the morphology of the foot is quantified by 2D footprints. However, footprint quantification ignores the foot's vertical dimension and hence, does not allow accurate quantification of complex 3D foot shape. METHODS: The shape variation of healthy 3D feet in a population of 31 adult women and 31 adult men who live in Belgium was studied using geometric morphometric methods. The effect of different factors such as sex, age, shoe size, frequency of sport activity, Body Mass Index (BMI), foot asymmetry, and foot loading on foot shape was investigated. Correlation between these factors and foot shape was examined using multivariate linear regression. RESULTS: The complex nature of a foot's 3D shape leads to high variability in healthy populations. After normalizing for scale, the major axes of variation in foot morphology are (in order of decreasing variance): arch height, combined ball width and inter-toe distance, global foot width, hallux bone orientation (valgus-varus), foot type (e.g. Egyptian, Greek), and midfoot width. These first six modes of variation capture 92.59% of the total shape variation. Higher BMI results in increased ankle width, Achilles tendon width, heel width and a thicker forefoot along the dorsoplantar axis. Age was found to be associated with heel width, Achilles tendon width, toe height and hallux orientation. A bigger shoe size was found to be associated with a narrow Achilles tendon, a hallux varus, a narrow heel, heel expansion along the posterior direction, and a lower arch compared to smaller shoe size. Sex was found to be associated with differences in ankle width, Achilles tendon width, and heel width. Frequency of sport activity was associated with Achilles tendon width and toe height. CONCLUSION: A detailed analysis of the 3D foot shape, allowed by geometric morphometrics, provides insights in foot variations in three dimensions that can not be obtained from 2D footprints. These insights could be applied in various scientific disciplines, including orthotics and shoe design.

BrainNetCNN: Convolutional neural networks for brain networks; towards predicting neurodevelopment.

We propose BrainNetCNN, a convolutional neural network (CNN) framework to predict clinical neurodevelopmental outcomes from brain networks. In contrast to the spatially local convolutions done in traditional image-based CNNs, our BrainNetCNN is composed of novel edge-to-edge, edge-to-node and node-to-graph convolutional filters that leverage the topological locality of structural brain networks. We apply the BrainNetCNN framework to predict cognitive and motor developmental outcome scores from structural brain networks of infants born preterm. Diffusion tensor images (DTI) of preterm infants, acquired between 27 and 46 weeks gestational age, were used to construct a dataset of structural brain connectivity networks. We first demonstrate the predictive capabilities of BrainNetCNN on synthetic phantom networks with simulated injury patterns and added noise. BrainNetCNN outperforms a fully connected neural-network with the same number of model parameters on both phantoms with focal and diffuse injury patterns. We then apply our method to the task of joint prediction of Bayley-III cognitive and motor scores, assessed at 18 months of age, adjusted for prematurity. We show that our BrainNetCNN framework outperforms a variety of other methods on the same data. Furthermore, BrainNetCNN is able to identify an infant's postmenstrual age to within about 2 weeks. Finally, we explore the high-level features learned by BrainNetCNN by visualizing the importance of each connection in the brain with respect to predicting the outcome scores. These findings are then discussed in the context of the anatomy and function of the developing preterm infant brain.

STEAM - Statistical Template Estimation for Abnormality Mapping: A personalized DTI analysis technique with applications to the screening of preterm infants.

We introduce the STEAM DTI analysis engine: a whole brain voxel-based analysis technique for the examination of diffusion tensor images (DTIs). Our STEAM analysis technique consists of two parts. First, we introduce a collection of statistical templates that represent the distribution of DTIs for a normative population. These templates include various diffusion measures from the full tensor, to fractional anisotropy, to 12 other tensor features. Second, we propose a voxel-based analysis (VBA) pipeline that is reliable enough to identify areas in individual DTI scans that differ significantly from the normative group represented in the STEAM statistical templates. We identify and justify choices in the VBA pipeline relating to multiple comparison correction, image smoothing, and dealing with non-normally distributed data. Finally, we provide a proof of concept for the utility of STEAM on a cohort of 134 very preterm infants. We generated templates from scans of 55 very preterm infants whose T1 MRI scans show no abnormalities and who have normal neurodevelopmental outcome. The remaining 79 infants were then compared to the templates using our VBA technique. We show: (a) that our statistical templates display the white matter development expected over the modeled time period, and (b) that our VBA results detect abnormalities in the diffusion measurements that relate significantly with both the presence of white matter lesions and with neurodevelopmental outcomes at 18months. Most notably, we show that STEAM produces personalized results while also being able to highlight abnormalities across the whole brain and at the scale of individual voxels. While we show the value of STEAM on DTI scans from a preterm infant cohort, STEAM can be equally applied to other cohorts as well. To facilitate this whole-brain personalized DTI analysis, we made STEAM publicly available at http://www.sfu.ca/bgb2/steam.

Structural network analysis of brain development in young preterm neonates.

Preterm infants develop differently than those born at term and are at higher risk of brain pathology. Thus, an understanding of their development is of particular importance. Diffusion tensor imaging (DTI) of preterm infants offers a window into brain development at a very early age, an age at which that development is not yet fully understood. Recent works have used DTI to analyze structural connectome of the brain scans using network analysis. These studies have shown that, even from infancy, the brain exhibits small-world properties. Here we examine a cohort of 47 normal preterm neonates (i.e., without brain injury and with normal neurodevelopment at 18 months of age) scanned between 27 and 45 weeks post-menstrual age to further the understanding of how the structural connectome develops. We use full-brain tractography to find white matter tracts between the 90 cortical and sub-cortical regions defined in the University of North Carolina Chapel Hill neonatal atlas. We then analyze the resulting connectomes and explore the differences between weighting edges by tract count versus fractional anisotropy. We observe that the brain networks in preterm infants, much like infants born at term, show high efficiency and clustering measures across a range of network scales. Further, the development of many individual region-pair connections, particularly in the frontal and occipital lobes, is significantly correlated with age. Finally, we observe that the preterm infant connectome remains highly efficient yet becomes more clustered across this age range, leading to a significant increase in its small-world structure.

A cross-sectional piecewise constant model for segmenting highly curved fiber tracts in diffusion MR images.

We propose a cross-sectional piecewise constant model for the segmentation of highly curved fiber tracts in diffusion MRI scans. An "anchor curve", obtained via tractography, provides the overall shape of the tract and allows us to examine the tract's microstructure at the level of cross-sectional planes normal to the curve. Each cross-section is modeled as a piecewise constant image, allowing us to address changes in measured diffusion due to the curving of the tract while still capturing overall tract structure. Results on both synthetic and real data show improved segmentation quality compared to state-of-the-art methods, particularly in areas of crossing fibers.

Detecting structure in diffusion tensor MR images.

We derive herein first and second-order differential operators for detecting structure in diffusion tensor MRI (DTI). Unlike existing methods, we are able to generate full first and second-order differentials without dimensionality reduction and while respecting the underlying manifold of the data. Further, we extend corner and curvature feature detectors to DTI using our differential operators. Results using the feature detectors on diffusion tensor MR images show the ability to highlight structure within the image that existing methods cannot.