Rapid artificial intelligence solutions in a pandemic-The COVID-19-20 Lung CT Lesion Segmentation Challenge.

Artificial intelligence (AI) methods for the automatic detection and quantification of COVID-19 lesions in chest computed tomography (CT) might play an important role in the monitoring and management of the disease. We organized an international challenge and competition for the development and comparison of AI algorithms for this task, which we supported with public data and state-of-the-art benchmark methods. Board Certified Radiologists annotated 295 public images from two sources (A and B) for algorithms training (n=199, source A), validation (n=50, source A) and testing (n=23, source A; n=23, source B). There were 1,096 registered teams of which 225 and 98 completed the validation and testing phases, respectively. The challenge showed that AI models could be rapidly designed by diverse teams with the potential to measure disease or facilitate timely and patient-specific interventions. This paper provides an overview and the major outcomes of the COVID-19 Lung CT Lesion Segmentation Challenge - 2020.

Development and evaluation of a machine learning-based point-of-care screening tool for genetic syndromes in children: a multinational retrospective study.

BACKGROUND: Delays in the diagnosis of genetic syndromes are common, particularly in low and middle-income countries with limited access to genetic screening services. We, therefore, aimed to develop and evaluate a machine learning-based screening technology using facial photographs to evaluate a child's risk of presenting with a genetic syndrome for use at the point of care. METHODS: In this retrospective study, we developed a facial deep phenotyping technology based on deep neural networks and facial statistical shape models to screen children for genetic syndromes. We trained the machine learning models on facial photographs from children (aged <21 years) with a clinical or molecular diagnosis of a genetic syndrome and controls without a genetic syndrome matched for age, sex, and race or ethnicity. Images were obtained from three publicly available databases (the Atlas of Human Malformations in Diverse Populations of the National Human Genome Research Institute, Face2Gene, and the dataset available from Ferry and colleagues) and the archives of the Children's National Hospital (Washington, DC, USA), in addition to photographs taken on a standard smartphone at the Children's National Hospital. We designed a deep learning architecture structured into three neural networks, which performed image standardisation (Network A), facial morphology detection (Network B), and genetic syndrome risk estimation, accounting for phenotypic variations due to age, sex, and race or ethnicity (Network C). Data were divided randomly into 40 groups for cross validation, and the performance of the model was evaluated in terms of accuracy, sensitivity, and specificity in both the total population and stratified by race or ethnicity, age, and sex. FINDINGS: Our dataset included 2800 facial photographs of children (1318 [47%] female and 1482 [53%] male; 1576 [56%] White, 432 [15%] African, 430 [15%] Hispanic, and 362 [13%] Asian). 1400 children with 128 genetic conditions were included (the most prevalent being Williams-Beuren syndrome [19%], Cornelia de Lange syndrome [17%], Down syndrome [16%], 22q11.2 deletion [13%], and Noonan syndrome [12%] syndrome) in addition to 1400 photographs of matched controls. In the total population, our deep learning-based model had an accuracy of 88% (95% CI 87-89) for the detection of a genetic syndrome, with 90% sensitivity (95% CI 88-92) and 86% specificity (95% CI 84-88). Accuracy was greater in White (90%, 89-91) and Hispanic populations (91%, 88-94) than in African (84%, 81-87) and Asian populations (82%, 78-86). Accuracy was also similar in male (89%, 87-91) and female children (87%, 85-89), and similar in children younger than 2 years (86%, 84-88) and children aged 2 years or older (eg, 89% [87-91] for those aged 2 years to <5 years). INTERPRETATION: This genetic screening technology could support early risk stratification at the point of care in global populations, which has the potential accelerate diagnosis and reduce mortality and morbidity through preventive care. FUNDING: Children's National Hospital and Government of Abu Dhabi.

Facial analysis technology for the detection of Down syndrome in the Democratic Republic of the Congo.

Down syndrome is one of the most common chromosomal anomalies affecting the world's population, with an estimated frequency of 1 in 700 live births. Despite its relatively high prevalence, diagnostic rates based on clinical features have remained under 70% for most of the developed world and even lower in countries with limited resources. While genetic and cytogenetic confirmation greatly increases the diagnostic rate, such resources are often non-existent in many low- and middle-income countries, particularly in Sub-Saharan Africa. To address the needs of countries with limited resources, the implementation of mobile, user-friendly and affordable technologies that aid in diagnosis would greatly increase the odds of success for a child born with a genetic condition. Given that the Democratic Republic of the Congo is estimated to have one of the highest rates of birth defects in the world, our team sought to determine if smartphone-based facial analysis technology could accurately detect Down syndrome in individuals of Congolese descent. Prior to technology training, we confirmed the presence of trisomy 21 using low-cost genomic applications that do not need advanced expertise to utilize and are available in many low-resourced countries. Our software technology trained on 132 Congolese subjects had a significantly improved performance (91.67% accuracy, 95.45% sensitivity, 87.88% specificity) when compared to previous technology trained on individuals who are not of Congolese origin (p < 5%). In addition, we provide the list of most discriminative facial features of Down syndrome and their ranges in the Congolese population. Collectively, our technology provides low-cost and accurate diagnosis of Down syndrome in the local population.

SegSRGAN: Super-resolution and segmentation using generative adversarial networks - Application to neonatal brain MRI.

BACKGROUND AND OBJECTIVE: One of the main issues in the analysis of clinical neonatal brain MRI is the low anisotropic resolution of the data. In most MRI analysis pipelines, data are first re-sampled using interpolation or single image super-resolution techniques and then segmented using (semi-)automated approaches. In other words, image reconstruction and segmentation are then performed separately. In this article, we propose a methodology and a software solution for carrying out simultaneously high-resolution reconstruction and segmentation of brain MRI data. METHODS: Our strategy mainly relies on generative adversarial networks. The network architecture is described in detail. We provide information about its implementation, focusing on the most crucial technical points (whereas complementary details are given in a dedicated GitHub repository). We illustrate the behavior of the proposed method for cortex analysis from neonatal MR images. RESULTS: The results of the method, evaluated quantitatively (Dice, peak signal-to-noise ratio, structural similarity, number of connected components) and qualitatively on a research dataset (dHCP) and a clinical one (Epirmex), emphasize the relevance of the approach, and its ability to take advantage of data-augmentation strategies. CONCLUSIONS: Results emphasize the potential of our proposed method/software with respect to practical medical applications. The method is provided as a freely available software tool, which allows one to carry out his/her own experiments, and involve the method for the super-resolution reconstruction and segmentation of arbitrary cerebral structures from any MR image dataset.

Unsupervised MRI Homogenization: Application to Pediatric Anterior Visual Pathway Segmentation.

Deep learning strategies have become ubiquitous optimization tools for medical image analysis. With the appropriate amount of data, these approaches outperform classic methodologies in a variety of image processing tasks. However, rare diseases and pediatric imaging often lack extensive data. Specially, MRI are uncommon because they require sedation in young children. Moreover, the lack of standardization in MRI protocols introduces a strong variability between different datasets. In this paper, we present a general deep learning architecture for MRI homogenization that also provides the segmentation map of an anatomical region of interest. Homogenization is achieved using an unsupervised architecture based on variational autoencoder with cycle generative adversarial networks, which learns a common space (i.e. a representation of the optimal imaging protocol) using an unpaired image-to-image translation network. The segmentation is simultaneously generated by a supervised learning strategy. We evaluated our method segmenting the challenging anterior visual pathway using three brain T1-weighted MRI datasets (variable protocols and vendors). Our method significantly outperformed a non-homogenized multi-protocol U-Net.

Multiscale brain MRI super-resolution using deep 3D convolutional networks.

The purpose of super-resolution approaches is to overcome the hardware limitations and the clinical requirements of imaging procedures by reconstructing high-resolution images from low-resolution acquisitions using post-processing methods. Super-resolution techniques could have strong impacts on structural magnetic resonance imaging when focusing on cortical surface or fine-scale structure analysis for instance. In this paper, we study deep three-dimensional convolutional neural networks for the super-resolution of brain magnetic resonance imaging data. First, our work delves into the relevance of several factors in the performance of the purely convolutional neural network-based techniques for the monomodal super-resolution: optimization methods, weight initialization, network depth, residual learning, filter size in convolution layers, number of the filters, training patch size and number of training subjects. Second, our study also highlights that one single network can efficiently handle multiple arbitrary scaling factors based on a multiscale training approach. Third, we further extend our super-resolution networks to the multimodal super-resolution using intermodality priors. Fourth, we investigate the impact of transfer learning skills onto super-resolution performance in terms of generalization among different datasets. Lastly, the learnt models are used to enhance real clinical low-resolution images. Results tend to demonstrate the potential of deep neural networks with respect to practical medical image applications.

An iterative multi-atlas patch-based approach for cortex segmentation from neonatal MRI.

Brain structure analysis in the newborn is a major health issue. This is especially the case for preterm neonates, in order to obtain predictive information related to the child development. In particular, the cortex is a structure of interest, that can be observed in magnetic resonance imaging (MRI). However, neonatal MRI data present specific properties that make them challenging to process. In this context, multi-atlas approaches constitute an efficient strategy, taking advantage of images processed beforehand. The method proposed in this article relies on such a multi-atlas strategy. More precisely, it uses two paradigms: first, a non-local model based on patches; second, an iterative optimization scheme. Coupling both concepts allows us to consider patches related not only to the image information, but also to the current segmentation. This strategy is compared to other multi-atlas methods proposed in the literature. Experiments on dHCP datasets show that the proposed approach provides robust cortex segmentation results.