ConCeptCNN: A novel multi-filter convolutional neural network for the prediction of neurodevelopmental disorders using brain connectome

Individualized prediction models in ADHD: a systematic review and meta-regression

There have been increasing efforts to develop prediction models supporting personalised detection, prediction, or treatment of ADHD. We overviewed the current status of prediction science in ADHD by: (1) systematically reviewing and appraising available prediction models; (2) quantitatively assessing factors impacting the performance of published models. We did a PRISMA/CHARMS/TRIPOD-compliant systematic review (PROSPERO: CRD42023387502), searching, until 20/12/2023, studies reporting internally and/or externally validated diagnostic/prognostic/treatment-response prediction models in ADHD. Using meta-regressions, we explored the impact of factors affecting the area under the curve (AUC) of the models. We assessed the study risk of bias with the Prediction Model Risk of Bias Assessment Tool (PROBAST). From 7764 identified records, 100 prediction models were included (88% diagnostic, 5% prognostic, and 7% treatment-response). Of these, 96% and 7% were internally and externally validated, respectively. None was implemented in clinical practice. Only 8% of the models were deemed at low risk of bias; 67% were considered at high risk of bias. Clinical, neuroimaging, and cognitive predictors were used in 35%, 31%, and 27% of the studies, respectively. The performance of ADHD prediction models was increased in those models including, compared to those models not including, clinical predictors (β = 6.54, p = 0.007). Type of validation, age range, type of model, number of predictors, study quality, and other type of predictors did not alter the AUC. Several prediction models have been developed to support the diagnosis of ADHD. However, efforts to predict outcomes or treatment response have been limited, and none of the available models is ready for implementation into clinical practice. The use of clinical predictors, which may be combined with other type of predictors, seems to improve the performance of the models. A new generation of research should address these gaps by conducting high quality, replicable, and externally validated models, followed by implementation research.

Supervised contrastive learning enhances graph convolutional networks for predicting neurodevelopmental deficits in very preterm infants using brain structural connectome

Very preterm (VPT) infants (born at less than 32 weeks gestational age) are at high risk for various adverse neurodevelopmental deficits. Unfortunately, most of these deficits cannot be accurately diagnosed until the age of 2-5 years old. Given the benefits of early interventions, accurate diagnosis and prediction soon after birth are urgently needed for VPT infants. Previous studies have applied deep learning models to learn the brain structural connectome (SC) to predict neurodevelopmental deficits in the preterm population. However, none of these models are specifically designed for graph-structured data, and thus may potentially miss certain topological information conveyed in the brain SC. In this study, we aim to develop deep learning models to learn the SC acquired at term-equivalent age for early prediction of neurodevelopmental deficits at 2 years corrected age in VPT infants. We directly treated the brain SC as a graph, and applied graph convolutional network (GCN) models to capture complex topological information of the SC. In addition, we applied the supervised contrastive learning (SCL) technique to mitigate the effects of the data scarcity problem, and enable robust training of GCN models. We hypothesize that SCL will enhance GCN models for early prediction of neurodevelopmental deficits in VPT infants using the SC. We used a regional prospective cohort of ∼280 VPT infants who underwent MRI examinations at term-equivalent age from the Cincinnati Infant Neurodevelopment Early Prediction Study (CINEPS). These VPT infants completed neurodevelopmental assessment at 2 years corrected age to evaluate cognition, language, and motor skills. Using the SCL technique, the GCN model achieved mean areas under the receiver operating characteristic curve (AUCs) in the range of 0.72∼0.75 for predicting three neurodevelopmental deficits, outperforming several competing models. Our results support our hypothesis that the SCL technique is able to enhance the GCN model in our prediction tasks.

Machine Learning and MRI-based Diagnostic Models for ADHD: Are We There Yet?

OBJECTIVE

Machine learning (ML) has been applied to develop magnetic resonance imaging (MRI)-based diagnostic classifiers for attention-deficit/hyperactivity disorder (ADHD). This systematic review examines this literature to clarify its clinical significance and to assess the implications of the various analytic methods applied.

METHODS

A comprehensive literature search on MRI-based diagnostic classifiers for ADHD was performed and data regarding the utilized models and samples were gathered.

RESULTS

We found that, although most studies reported the classification accuracies, they varied in choice of MRI modalities, ML models, cross-validation and testing methods, and sample sizes. We found that the accuracies of cross-validation methods inflated the performance estimation compared with those of a held-out test, compromising the model generalizability. Test accuracies have increased with publication year but were not associated with training sample sizes. Improved test accuracy over time was likely due to the use of better ML methods along with strategies to deal with data imbalances.

CONCLUSION

Ultimately, large multi-modal imaging datasets, and potentially the combination with other types of data, like cognitive data and/or genetics, will be essential to achieve the goal of developing clinically useful imaging classification tools for ADHD in the future.

ClusterNet: a clustering distributed prior embedded detection network for early-stage esophageal squamous cell carcinoma diagnosis

BACKGROUND

Early and accurate diagnosis of esophageal squamous cell carcinoma (ESCC) is important for reducing mortality. Analyzing intrapapillary capillary loops' (IPCLs) patterns on magnification endoscopy with narrow band imaging (ME-NBI) has been demonstrated effective in the diagnosis of early-stage ESCC. However, even experienced endoscopists may face difficulty in finding and classifying countless IPCLs on ME-NBI.

PURPOSE

We propose a novel clustering prior embedded detection network: ClusterNet. ClusterNet is capable of analyzing the distribution of IPCLs on ME-NBI automatically and enables endoscopists to overview multiple types of visualization. With ClusterNet assisting, endoscopists may observe ME-NBI images more efficiently, thus they may also predict the pathology and make medical decisions more easily.

METHODS

We propose the first large-scale ME-NBI dataset with fine-grained annotations by consensus of expert endoscopists. The dataset is splitted into a training set and an independent testing set based on patients. With two strategies for embedding, ClusterNet can automatically take the clustering effect into consideration. Prior to this work, none of the existing approaches take the clustering effect, which is rather important in classifying the IPCLs, into account.

RESULTS

ClusterNet achieves an average precision of 81.2% and an average recall of 90.0% for the detection of IPCLs patterns on each patient of the independent testing set. We also compare ClusterNet with other state-of-the-art detection approaches. The performance of ClusterNet with embedding strategies is consistently superior to that of other approaches in terms of average precision, recall and F2-Score.

CONCLUSIONS

Experiments demonstrate that our proposed method is able to detect almost all the IPCLs patterns on ME-NBI and classify them according to the Japanese Endoscopic Society (JES) classification accurately.

Graph-Based Disease Prediction in Neuroimaging: Investigating the Impact of Feature Selection

In biomedical machine learning, data often appear in the form of graphs. Biological systems such as protein interactions and ecological or brain networks are instances of applications that benefit from graph representations. Geometric deep learning is an arising field of techniques that has extended deep neural networks to non-Euclidean domains such as graphs. In particular, graph convolutional neural networks have achieved advanced performance in semi-supervised learning in those domains. Over the last years, these methods have gained traction in neuroscience as they could be the key to a deeper understanding in clinical diagnosis at the systems or network level (for an individual brain but also for across a cohort of subjects). As a proof-of-principle, we study and validate a previous implementation of graph-based semi-supervised classification using a ridge classifier and graph convolutional neural networks. The models are trained on population graphs that integrate imaging and phenotypic information. Our analysis employs neuroimaging data of structural and functional connectivity for prediction of neurodevelopmental and neurodegenerative disorders. Here, we particularly study the effect of different strategies to reduce the dimensionality of the neuroimaging features on the graph nodes on the classification performance.