Dirk Vandermeulen

KU Leuven | ku leuven · Department of Electrical Engineering (ESAT)

The cranial vault in humans is highly variable, clinically relevant, and heritable, yet its genetic architecture remains poorly understood. Here, we conduct a joint multi-ancestry and admixed multivariate genome-wide association study on 3D cranial vault shape extracted from magnetic resonance images of 6772 children from the ABCD study cohort yiel...

Vertebral fractures (VFs) are the hallmark of osteoporosis, being one of the most frequent types of fragility fracture and an early sign of the disease. They are associated with significant morbidity and mortality. VFs are incidentally found in one out of five imaging studies, however, more than half of the VFs are not identified nor reported in pa...

The cranial vault - the portion of the skull surrounding the brain and cerebellum - is highly variable, clinically relevant, and heritable, yet its genetic architecture remains poorly understood. Here, we conducted a joint multi-ancestry and admixed multivariate GWAS on 3D cranial vault shape extracted from magnetic resonance images of 6,772 childr...

In this article, we look into some essential aspects of convolutional neural networks (CNNs) with the focus on medical image segmentation. First, we discuss the CNN architecture, thereby highlighting the spatial origin of the data, voxel-wise classification and the receptive field. Second, we discuss the sampling of input-output pairs, thereby high...

We know that both the CNN mapping function and the sampling scheme are of paramount importance for CNN-based image analysis. It is clear that both functions operate in the same space, with an image axis $\mathcal{I}$ and a feature axis $\mathcal{F}$. Remarkably, we found that no frameworks existed that unified the two and kept track of the spatial...

This article focuses on the control center of each human body: the brain. We will point out the pivotal role of the cerebral vasculature and how its complex mechanisms may vary between subjects. We then emphasize a specific acute pathological state, i.e., acute ischemic stroke, and show how medical imaging and its analysis can be used to define the...

The clinical interest is often to measure the volume of a structure, which is typically derived from a segmentation. In order to evaluate and compare segmentation methods, the similarity between a segmentation and a predefined ground truth is measured using popular discrete metrics, such as the Dice score. Recent segmentation methods use a differen...

Albeit the Dice loss is one of the dominant loss functions in medical image segmentation, most research omits a closer look at its derivative, i.e. the real motor of the optimization when using gradient descent. In this paper, we highlight the peculiar action of the Dice loss in the presence of missing or empty labels. First, we formulate a theoret...

A novel algorithm for generating artificial training samples from triangulated three-dimensional (3D) surface models within the context of dental implant recognition is proposed. The proposed algorithm is based on the calculation of two-dimensional (2D) projections (from a number of different angles) of 3D volumetric representations of computer-aid...

Albeit the Dice loss is one of the dominant loss functions in medical image segmentation, most research omits a closer look at its derivative, i.e. the real motor of the optimization when using gradient descent. In this paper, we highlight the peculiar action of the Dice loss in the presence of missing or empty labels. First, we formulate a theoret...

Bloodstain impact pattern Area of Origin (AO) estimation is an important but time-consuming process in criminal investigations. HemoVision is a software package that automates and accelerates this process. To date, however, no study has been published that evaluates HemoVision’s accuracy. Moreover, HemoVision relies on an automated variant of the t...

Convolutional neural networks (CNNs) for brain tumor segmentation are generally developed using complete sets of magnetic resonance imaging (MRI) sequences for both training and inference. As such, these algorithms are not trained for realistic, clinical scenarios where parts of the MRI sequences which were used for training, are missing during inf...

Machine learning driven medical image segmentation has become standard in medical image analysis. However, deep learning models are prone to overconfident predictions. This has led to a renewed focus on calibrated predictions in the medical imaging and broader machine learning communities. Calibrated predictions are estimates of the probability of...

Registration is an essential tool in image analysis. Deep learning based alternatives have recently become popular, achieving competitive performance at a faster speed. However, many contemporary techniques are limited to volumetric representations, despite increased popularity of 3D surface and shape data in medical image analysis. We propose a on...

Registration is an essential tool in image analysis. Deep learning based alternatives have recently become popular, achieving competitive performance at a faster speed. However, many contemporary techniques are limited to volumetric representations, despite increased popularity of 3D surface and shape data in medical image analysis. We propose a on...

Machine learning driven medical image segmentation has become standard in medical image analysis. However, deep learning models are prone to overconfident predictions. This has lead to a renewed focus on calibrated predictions in the medical imaging and broader machine learning communities. Calibrated predictions are estimates of the probability of...

Objectives To develop and evaluate a geometric deep-learning network to automatically place seven palatal landmarks on digitized maxillary dental casts. Settings and Sample population The sample comprised individuals with permanent dentition of various ethnicities. The network was trained from manual landmark annotations on 732 dental casts and ev...

Purpose: Three-dimensional reconstructions of the human anatomy have been available for surgery planning or diagnostic purposes for a few years now. The different image modalities usually rely on several consecutive two-dimensional acquisitions in order to reconstruct the three-dimensional volume. Hence, such acquisitions are expensive, time-deman...

Our motivating application is a real-world problem: COVID-19 classification from CT imaging, for which we present an explainable Deep Learning approach based on a semi-supervised classification pipeline that employs variational autoencoders to extract efficient feature embedding. We have optimized the architecture of two different networks for CT i...

In many medical imaging and classical computer vision tasks, the Dice score and Jaccard index are used to evaluate the segmentation performance. Despite the existence and great empirical success of metric-sensitive losses, i.e. relaxations of these metrics such as soft Dice, soft Jaccard and Lovasz-Softmax, many researchers still use per-pixel loss...

The clinical interest is often to measure the volume of a structure, which is typically derived from a segmentation. In order to evaluate and compare segmentation methods, the similarity between a seg-mentation and a predefined ground truth is measured using popular discrete metrics, such as the Dice score. Recent segmentation methods use a differe...

Staging third molar development is commonly used for age assessment in sub-adults. Current staging techniques are, at most, semi-automated and rely on manual interactions prone to operator variability. The aim of this study was to fully automate the staging process by employing the full potential of deep learning, using convolutional neural network...

Recent research on COVID-19 suggests that CT imaging provides useful information to assess disease progression and assist diagnosis, in addition to help understanding the disease. There is an increasing number of studies that propose to use deep learning to provide fast and accurate quantification of COVID-19 using chest CT scans. The main tasks of...

In many medical imaging and classical computer vision tasks, the Dice score and Jaccard index are used to evaluate the segmentation performance. Despite the existence and great empirical success of metric-sensitive losses, i.e. relaxations of these metrics such as soft Dice, soft Jaccard and Lovász-Softmax, many researchers still use per-pixel loss...

Given sufficient training samples, statistical shape models can provide detailed population representations for use in anthropological and computational genetic studies, injury biomechanics, musculoskeletal disease models or implant design optimization. While the technique has become extremely popular for the description of isolated anatomical stru...

Considering the high demand for the identification of unknown remains in South Africa, a need exists to establish reliable facial approximation techniques that will take into account sex and age and, most importantly, be useful within the South African context. This study aimed to provide accurate statistical models for predicting nasal soft-tissue...

Segmentation is a fundamental task in medical image analysis. The clinical interest is often to measure the volume of a structure. To evaluate and compare segmentation methods, the similarity between a segmentation and a predefined ground truth is measured using metrics such as the Dice score. Recent segmentation methods based on convolutional neur...

The scientific community in the field of craniofacial reconstruction recognised that manual facial reconstruction methods require a high degree of anatomical and sculptural expertise and remain difficult and subjective in practice. The main critiques are the inherent subjectivity in manual methods, the references used, the non-consideration of popu...

The scientific community in the field of craniofacial reconstruction recognised that manual facial reconstruction methods require a high degree of anatomical and sculptural expertise and remain difficult and subjective in practice. The main critiques are the inherent subjectivity in manual methods, the references used, the non-consideration of popu...

Osteoporosis induced fractures occur worldwide about every 3 s. Vertebral compression fractures are early signs of the disease and considered risk predictors for secondary osteoporotic fractures. We present a detection method to opportunistically screen spine-containing CT images for the presence of these vertebral fractures. Inspired by radiology...

Manual landmarking is used in several manual and semi-automated prediction guidelines for approximation of the nose. The manual placement of landmarks may, however, render the analysis less repeatable due to observer subjectivity and, consequently, have an impact on the accuracy of the human facial approximation. In order to address this subjectivi...

CONCLUSION Our method achieves an AUC of 0.95±0.02 outperforming Valentinitsch et al. We also illustrate that our method achieves higher recall (0.905) on the operating point reported by Bar et al. The results of our 5-fold cross-validation experiment demonstrate that our 3D data-driven method compares favourably to state-of-the-art using 2.5D lear...

The Dice score and Jaccard index are commonly used metrics for the evaluation of segmentation tasks in medical imaging. Convolutional neural networks trained for image segmentation tasks are usually optimized for (weighted) cross-entropy. This introduces an adverse discrepancy between the learning optimization objective (the loss) and the end targe...

Purpose: Statistical shape modeling provides a powerful tool for describing and analyzing human anatomy. By linearly combining the variance of the shape of a population of a given anatomical entity, statistical shape models (SSMs) identify its main modes of variation and may approximate the total variance of that population to a selected threshold,...

Osteoporosis induced fractures occur worldwide about every 3 seconds. Vertebral compression fractures are early signs of the disease and considered risk predictors for secondary osteoporotic fractures. We present a detection method to opportunistically screen spine-containing CT images for the presence of these vertebral fractures. Inspired by radi...

Osteoporosis induced fractures occur worldwide about every 3 seconds. Vertebral compression fractures are early signs of the disease and considered risk predictors for secondary osteoporotic fractures. We present a detection method to opportunistically screen spine-containing CT images for the presence of these vertebral fractures. Inspired by radi...

The Dice score and Jaccard index are commonly used metrics for the evaluation of segmentation tasks in medical imaging. Convolutional neural networks trained for image segmentation tasks are usually optimized for (weighted) cross-entropy. This introduces an adverse discrepancy between the learning optimization objective (the loss) and the end targe...

Segmentation is a fundamental task in medical image analysis. The clinical interest is often to measure the volume of a structure. To evaluate and compare segmentation methods, the similarity between a segmentation and a predefined ground truth is measured using metrics such as the Dice score. Recent segmentation methods based on convolutional neur...

Staging third molar development is commonly used for age estimation in subadults. Automated developmental stage allocation to the mandibular left third molar in panoramic radiographs has been examined in a pilot study. This method used an AlexNet Deep Convolutional Neural Network (CNN) approach to stage lower left third molars, which had been selec...

Facial recognition from DNA refers to the identification or verification of unidentified biological material against facial images with known identity. One approach to establish the identity of unidentified biological material is to predict the face from DNA, and subsequently to match against facial images. However, DNA phenotyping of the human fac...

In light of the great demand for the identification of unknown remains in South Africa, a need exists to establish reliable facial approximation techniques that will not only take into account sex and age, but most importantly be specific for the South African population. The aim of this study was to provide accurate statistical models for predicti...

Image segmentation has become an important tool in orthopedic and biomechanical research. However, it greatly remains a time-consuming and laborious task. In this manuscript, we propose a fully automatic model-based segmentation pipeline for the full lower limb in computed tomography (CT) images. The method relies on prior shape model fitting, foll...

Medical image computing aims at developing computational strategies for robust, automated, quantitative analysis of relevant information from medical imaging data in order to support diagnosis, therapy planning and follow-up, and biomedical research. Medical image analysis is complicated by the complexity of the data itself—involving 3D tomographic...

Tissue intensity distributions in medical images can have varying degrees of statistical dispersion, which is referred to as heteroscedasticity. This can influence image contrast and gradients, but can also negatively affect the performance of general-purpose distance metrics. Numerous methods to preprocess heteroscedastic images have already been...

The profile of the nose is an important feature for facial approximations. Although several manual and semi-automated prediction guidelines exist for estimating the shape of the nose, the reliability and applicability of these methods to South Africans groups are unknown. The aim of this study was to predict the displacements of capulometric landma...

Genome-wide association scans of complex multipartite traits like the human face typically use preselected phenotypic measures. Here we report a data-driven approach to phenotyping facial shape at multiple levels of organization, allowing for an open-ended description of facial variation while preserving statistical power. In a sample of 2,329 pers...

Manual landmarking is used in several manual and semi-automated prediction guidelines for the approximation of the human nose. The manual placement of landmarks may render the analysis less repeatable due to observer subjectivity and may have an impact on the accuracy of the human facial approximation. In order to achieve the best accuracy for huma...

Background: Automated methods to evaluate growth of hand and wrist bones on radiographs and magnetic resonance imaging have been developed. They can be applied to estimate age in children and subadults. Automated methods require the software to (1) recognise the region of interest in the image(s), (2) evaluate the degree of development and (3) cor...

Forensic scientists in South Africa follow soft tissue thickness recommendations for facial approximations based on international guidelines. From the literature and in practice, ancestral variations may significantly affect the accuracy of facial approximations. In this study, variability in the nasal shape of white and black South Africans is eva...

Sitting is part of our daily work and leisure activities and can be performed in different configurations. To date, the impact of different sitting configurations on hip joint loading has not been studied. We therefore evaluated the hip joint reaction force (HJRF) and hip flexion angle in a virtual representative male Caucasian population by means...

Forensic scientists in South Africa follow soft tissue thickness recommendations for facial approximations based on international guidelines. From the literature and in practice, ancestral variations may significantly affect the accuracy of facial approximations. The purpose of this study was to approximate the shape of the nose in South Africans b...

Forensic scientists in South Africa follow soft tissue thickness recommendations for facial approximations based on international guidelines. From the literature and in practice, ancestral variations may significantly affect the accuracy of facial approximations. In this study, variability in the nasal shape of white and black South Africans is eva...

The purpose of this study was to create a computer-assisted method for approximation of the nose from CBCT scans. Data on the nose are relevant to South African forensic artists and forensic anthropologists. Currently, forensic artists in South Africa follow specific guidelines of soft tissue thicknesses based on North American cadaver studies. Stu...

Although several manual and semi-automated prediction guidelines for the approximation of the nose exist, the reliability of these methods is unknown. This study examined a published method to predict the dimensions of the external nose in South African groups. The method described by Lee and collaborators involves the use of CBCT scans from a Kore...

The purpose of this study was to create a computer-assisted method for approximation of the nose from CBCT scans. Data on the nose are relevant to South African forensic artists and forensic anthropology. Currently, forensic artists in South Africa follow specific guidelines of soft tissue thicknesses based on North American cadaver studies. As sho...

Skeletal dimensions as predictors of the external morphology of the nose using cone-beam computed tomography (CBCT) in a South Africans sample. Although several manual and semi-automated prediction guidelines for the approximation of the nose exist, the reliability of these methods is unknown. This study examined a published method to predict the...

The purpose of this study was to create an automatic and reproducible computer-assisted method for approximation of the nose with specific relevance to the South African population from retrospective Cone Beam Computed Tomography (CBCT) scans for forensic anthropological implications. The anatomy of the nose is complex and partly composed of bone a...

Comparing ear photographs is considered to be an important aspect of disaster victim identification and other forensic and security applications. An interesting approach concerns the construction of 3D ear models by fitting the parameters of a ‘standard’ ear shape, in order to transform it into an optimal approximation of a 3D ear image. A feature...

This paper presents a local-to-global statistical approach for modeling the major components of left ventricular (LV) shape using its 3-D landmark representation. The rationale for dividing the LV into local areas is bi-fold: 1) to better identify abnormalities that lead to local shape remodeling and, 2) to decrease the number of shape variables by...

INTRODUCTION: A growing body of evidence supports the regeneration potential of dental tissues after regenerative endodontic treatment (RET). Nevertheless, a standard method for the evaluation of RET outcome is lacking. The aim of this study was to develop a standardized quantitative method for RET outcome analysis based on cone-beam computed tomog...

The human external ears, or pinnae, have an intriguing shape and, like most parts of the human external body, bilateral symmetry is observed between left and right. It is a well-known part of our auditory sensory system and mediates the spatial localization of incoming sounds in 3D from monaural cues due to its shape-specific filtering as well as b...

Objectives Traditional Bloodstain Pattern Analysis (BPA) estimates impact angles for bloodstain spatters by employing ellipses and an inverse sine. This approach is based upon a simplification of the stain formation process, and ignores the physical properties of blood, and its intricate interactions with the target surface. This research presents...

Comparing ear photographs is considered to be an important aspect of victim identification. In this paper we study how automated ear comparison can be improved with soft computing techniques. More specifically we describe and illustrate how bipolar data modelling tech-niques can be used for handling data imperfections more adequately. In order to m...

Ear biometric authentication is considered to be an important aspect of human identification and is, among other techniques, used in victim identification for practical reasons . State-of-the-art techniques transform 2D ear photos to 3D ear models to adequately cope with geometrical and photometric normalisation issues. From each 3D ear model a fea...

Objectives Conventional blood pattern analysis is a tedious and time- consuming process due to the many actions that must be performed by the pattern analyst. Digital photographs can now be employed to aid in the process of scene analysis, by letting computer programs perform some of the required steps. However, current computerized methods are all...

Objective To estimate body weight using the data of a post mortem total body CT scan and establish its efficacy in a forensic context. Material and method PMCT data were used to compute fat, soft tissue and bone volume using in house developed software based on density and HU. The digitally calculated body weight was then compared with the measure...