Deep Learning-based Convolutional Neural Network Research Articles

Genetically optimised SMOTE-based adversarial discriminative domain adaptation for rotor fault diagnosis at variable operating conditions

Abstract For safety, reliability, and uninterrupted output of gas turbines, aviation engines, power-generating equipment, pumps, gears, compressors etc., rotor mass imbalance must be detected and diagnosed to avoid catastrophic failure. Industry 4.0 relies on predictive digital maintenance and deep learning-based convolutional neural network (CNN), which predicts defects but fails if the operating conditions change. Research studies in various fields indicate that the domain shift issue occurs due to source and target samples being from different domains, which reduces prediction capability. Moreover, research studies are scarce in examining prediction capability under varying operating speeds for rotor mass imbalance. Hence, this research proposes the adversarial discriminative domain adaptation (ADDA) technique which predicts machine failures under various operational conditions. The efficacy of ADDA has been explored by introducing 1D-CNN as a source and a target encoder inside ADDA's architecture to take advantage of CNN's feature extraction capability. Further, this research effectively tackles CNN's inherent issues of overfitting and hyperparameters value selection. Furthermore, The real-world scenario has more healthy samples than fault condition samples, causing a multiclass imbalance in sample data, which affects the classification decision boundary and causes biased prediction. Hence, the proposed methodology first addresses the class imbalance through synthetic minority oversampling (SMOTE), then genetic algorithm optimizes 1D-CNN's hyperparameters, and the effective dropout layer positioning solves the overfitting. Finally, the deep learning-based SMOTE_ADDA_GO-1D-CNN decreases domain discrepancy with adversarial discriminative domain adaptation. The proposed methodology's efficacy has been explored through F1-Score, which is used as multiclass evaluation metrics, and it has been benchmarked against standard machine learning and deep learning algorithms. The test results of the proposed methodology surpassed all of them with maximum prediction accuracy. Thus, this study contributes to rotor mass imbalance detection and diagnosis for multiclass imbalanced data under varying operational conditions by successfully overcoming potential challenges during fault prediction.

A new ViT-Based augmentation framework for wafer map defect classification to enhance the resilience of semiconductor supply chains

Wafer map defect classification plays a crucial role in sustaining the semiconductor supply chain during industrial disruptions by ensuring continuity, resilience, and efficiency while aligning with sustainability principles. Up to 30 percent of production costs is lost to chip testing and yield losses for semiconductor manufacturing supply chain. In the semiconductor practice, wafer map defects recognition plays a linchpin role in the front-end-of-line stage. Defect pattern recognition can directly pinpoint the assignable causes and provide the domain experts with actionable insights. However, various wafer defect types occur differently from each other, which makes the collected wafer map dataset to be highly imbalanced in defect classes. In the face of data imbalance, the classification model usually cannot provide satisfactory classification performance. In this aspect, this paper intends to investigate the wafer defect map classification problem by using Vision Transformer (ViT) as an alternative data augmentation approach. The primary purpose is to alleviate the class-imbalance issue and then the performance of a deep learning-based convolutional neural network for wafer defect classification can be effectively improved. The experimental results demonstrate that the proposed data augmentation method by using ViT proves to be a potential generative model for improving wafer map defect classification, particularly robust on the individual minority class. In a word, the proposed augmentation framework for wafer map defect classification facilitates a more targeted and efficient approach to quality control, resource utilization, and production optimization in maintaining a sustainable semiconductor supply chain, particularly in times of industrial disruption.

Construction of deep learning-based convolutional neural network model for automatic detection of fluid hysteroscopic endometrial micropolyps in infertile women with chronic endometritis

Objective(s)Chronic endometritis (CE) is a localized mucosal inflammatory disorder associated with female infertility of unknown etiology, endometriosis, tubal factors, repeated implantation failure, and recurrent pregnancy loss, along with atypical uterine bleeding and iron deficiency anemia. Diagnosis of CE has traditionally relied on endometrial biopsy and detection of CD138(+) endometrial stromal plasmacytes. To develop a less invasive diagnostic system for CE, we aimed to construct a deep learning-based convolutional neural network (CNN) model for the automatic detection of endometrial micropolyps (EMiP), a fluid hysteroscopy (F-HSC) finding recognized as tiny protrusive lesions that are closely related to this disease. Study designThis is an in silico study using archival images of F-HSC performed at an infertility center in a private clinic. A total of 244 infertile women undergoing F-HSC on the days 6–12 of the menstrual cycle between April 2019 and December 2021 with histopathologically-confirmed CE with the aid of immunohistochemistry for CD138 were utilized. ResultsThe archival F-HSC images of 208 women (78 with EMiP and 130 without EMiP) who met the inclusion criteria were finally subjected to analysis. Following preprocessing of the images, half a set was input into a CNN architecture for training, whereas the remaining images were utilized as the test set to evaluate the performance of the model, which was compared with that of the experienced gynecologists. The sensitivity, specificity, accuracy, precision, and F1-score of the CNN model-aided diagnosis were 93.6 %, 92.3 %, 92.8 %, 88.0 %, and 0.907, respectively. The area under the receiver operating characteristic curves of the CNN model-aided diagnosis (0.930) was at a similar level (p > .05) to the value of conventional diagnosis by three experienced gynecologists (0.927, 0.948, and 0.906). ConclusionThese findings indicate that our deep learning-based CNN is capable of recognizing EMiP in F-HSC images and holds promise for further development of the computer-aided diagnostic system for CE.

Deep learning-based CNN for multiclassification of ocular diseases using transfer learning

ABSTRACT Effective and timely diagnosis and treatment of ocular diseases is essential for swift recovery of the patients. Among ocular diseases, cataract and glaucoma are the most prevalent globally and need adequate attention. The present paper aims to develop an optimised deep learning based convolutional neural network (CNN) for the multi-classification of ocular diseases (normal, glaucoma and cataract). Three pre-trained CNNs (SqueezeNet, Darknet-53, EfficientNet-b0) were optimised concerning batch size (6/8/10) & optimiser type (SGDM, RMSProp, Adam) for obtaining maximum possible accuracy in the detection of multiple ocular diseases (cataract & glaucoma). Darknet-53 (batch size-6, optimiser type-Adam) gave the highest accuracy of 99.4% for a test sample of 1000 images. The performance metrics of Darknet-53 have been computed using a confusion matrix. Confusion matrix is also applied to calculate accuracy, sensitivity, specificity, f1 score and receiver operating curve (ROC). Through comparative performance analysis of the three CNNs, SqueezeNet, Darknet-53 and EfficientNet-b0 achieved the highest accuracy of 95%, 99.4% and 90%, respectively. The results indicate the importance of batch size and optimiser type on the performance of CNN models.

Emotional Recognition Based on Faces through Deep Learning Algorithms

Facial expressions have long been a straightforward way for humans to determine emotions, but computer systems find it significantly more difficult to do the same. Emotion recognition from facial expressions, a subfield of social signal processing, is employed in many different circumstances, but is especially useful for human-computer interaction. Many studies have been conducted on automatic emotion recognition, with the majority utilizing machine learning techniques. However, the identification of basic emotions such as fear, sadness, surprise, anger, happiness, and contempt remains a challenging subject in computer vision. Recently, deep learning has gained more attention as potential solutions for a range of real-world problems, such as emotion recognition. In this work, we refined the convolutional neural network method to discern seven basic emotions and assessed several preprocessing approaches to illustrate their impact on CNN performance. The goal of this research is to enhance facial emotions and features by using emotional recognition. Computers may be able to forecast mental states more accurately and respond with more customised answers if they can identify or recognise the facial expressions that elicit human responses. Consequently, we investigate how a convolutional neural network-based deep learning technique may enhance the recognition of emotions from facial features (CNN). Consequently, we investigate how a convolutional neural network-based deep learning technique may enhance the recognition of emotions from facial features (CNN). Our dataset, which comprises of roughly 32,298 pictures for testing and training, includes multiple face expressions. After noise removal from the input image, the pretraining phase helps reveal face detection, including feature extraction. The preprocessing system helps with this.

Convolutional Neural Network-Based Deep Learning Engine for Mastoidectomy Instrument Recognition and Movement Tracking.

To develop a convolutional neural network-based computer vision model to recognize and track 2 mastoidectomy surgical instruments-the drill and the suction-irrigator-from intraoperative video recordings of mastoidectomies. Technological development and model validation. Academic center. Ten 1-minute videos of mastoidectomies done for cochlear implantation by varying levels of resident surgeons were collected. For each video, containing 900 frames, an open-access computer vision annotation tool was used to annotate the drill and suction-irrigator class images with bounding boxes. A mastoidectomy instrument tracking module, which extracts the center coordinates of bounding boxes, was developed using a feature pyramid network and layered with DETECTRON, an open-access faster-region-based convolutional neural network. Eight videos were used to train the model, and 2 videos were used for testing. Outcome measures included Intersection over Union (IoU) ratio, accuracy, and average precision. For an IoU of 0.5, the mean average precision for the drill was 99% and 86% for the suction-irrigator. The model proved capable of generating maps of drill and suction-irrigator stroke direction and distance for the entirety of each video. This computer vision model can identify and track the drill and suction-irrigator from videos of intraoperative mastoidectomies performed by residents with excellent precision. It can now be employed to retrospectively study objective mastoidectomy measures of expert and resident surgeons, such as drill and suction-irrigator stroke concentration, economy of motion, speed, and coordination, setting the stage for characterization of objective expectations for safe and efficient mastoidectomies.

Abstract 1043: Identification of the novel immune synapse-localized proteome for immuno-oncology using Microscoop-induced targeted photo-biotinylation

Abstract Although tens of immune synapse (IS) protein species are known, there remain many unknown IS-localized protein species. Understanding the proteome of the IS between a target cell and a lymphocyte is crucial for advancing immuno-oncology. The low abundance of ISs and the lack of a definitive enrichment marker have limited efficient proteomic profiling. Here, we used Microscoop™, an innovative system that integrates microscopy, machine learning, and targeted photo-biotinylation to enable precise and spatially specific enrichment of IS proteins, thereby facilitating proteomic discovery of the IS. We utilized Raji B cells as antigen-presenting cells (APCs) and induced IS formation with Jurkat T cells. The system first employed immunofluorescence imaging of CD3, a common IS marker in Jurkat T cells, and used a convolutional neural network-based deep learning algorithm to train the recognition of ISs from CMTPX-stained Raji B cells. Our automated system then successfully achieved spatially-targeted biotin-tagging of proteins at ISs through thousands of fields of view of imaging, deep learning-driven pattern generation, and photochemical labeling. Subsequent streptavidin pull-down and mass spectrometry analysis enabled the identification of IS-specific proteins. Remarkably, our spatial proteomic approach led to the isolation and identification of hundreds of different species at the IS interface, including proteins associated with key components of T-cell receptor (TCR) signaling pathways such as the TCR/CD3 complex, Src and Tec family tyrosine kinases, and pivotal NF-kB signaling proteins. Beyond these, we also identified a significant enrichment of proteins not previously associated with the IS. Our study not only illuminates previously unknown aspects of immune regulation at the IS interface, but also shedded light on proteins potentially important for immuno-oncology. These insights have significant implications for cancer research, particularly in understanding and manipulating immune responses for therapeutic purposes. Citation Format: Weng-Man Chong, Cecile Li, Chun-Kai Huang, Chantal Hoi Yin Cheung, Hsiao-Jen Chang, Chia-Wen Chung, Jung-Chi Liao. Identification of the novel immune synapse-localized proteome for immuno-oncology using Microscoop-induced targeted photo-biotinylation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1043.

Pattern Recognition in Embedded Systems for Event Occurrences

Abstract: PR (pattern recognition) typically includes interaction with humans and other complicated processes in the real world, embedded systems are ideal candidates. A typical PR application often considered the more perceptual branch of AI, responds to external events that the system detects through physical sensors or input devices by activating actuators or displaying relevant information. To explore the embedded recognition system and apply the deep learning algorithm to face detection, the deep learning-based Convolutional Neural Network (CNN)suggests two deep face detection methods. These are presented to use the deep learning algorithm.This was done to make it possible for us to use the deep learning algorithm for face detection. Because of this, to analyze the built-in face recognition system and applied the deep learning algorithm to the process of identifying faces. In addition, to do both things simultaneously. OMTCNN's training accuracy is 85.14%, higher than the unimproved algorithm. Accuracy of the recognition and calculation acceleration modules boosts embedded system face detection and identification performance. Embedded deep learning recognition is helpful.

A comparative analysis of CNN-based deep learning architectures for early diagnosis of bone cancer using CT images

Bone cancer is a rare in which cells in the bone grow out of control, resulting in destroying the normal bone tissue. A benign type of bone cancer is harmless and does not spread to other body parts, whereas a malignant type can spread to other body parts and might be harmful. According to Cancer Research UK (2021), the survival rate for patients with bone cancer is 40% and early detection can increase the chances of survival by providing treatment at the initial stages. Prior detection of these lumps or masses can reduce the risk of death and treat bone cancer early. The goal of this current study is to utilize image processing techniques and deep learning-based Convolution neural network (CNN) to classify normal and cancerous bone images. Medical image processing techniques, like pre-processing (e.g., median filter), K-means clustering segmentation, and, canny edge detection were used to detect the cancer region in Computer Tomography (CT) images for parosteal osteosarcoma, enchondroma and osteochondroma types of bone cancer. After segmentation, the normal and cancerous affected images were classified using various existing CNN-based models. The results revealed that AlexNet model showed a better performance with a training accuracy of 98%, validation accuracy of 98%, and testing accuracy of 100%.

Improving rainfall-runoff modeling in the Mekong river basin using bias-corrected satellite precipitation products by convolutional neural networks

Accurate rainfall-runoff (RR) modeling is crucial for effective Mekong River Basin (MRB) water resource management. Satellite precipitation products (SPPs) can offer valuable data for such modeling; however, these products often exhibit biases that may adversely affect hydrological simulations. This study aimed to improve RR modeling using bias-corrected SPPs and the Soil and Water Assessment Tool (SWAT) model for MRB. A convolutional neural network-based deep learning framework was employed to correct biases in four SPPs (TRMM, PERSIANN-CDR, CHIRPS, and CMORPH), resulting in four respective bias-corrected SPPs (ADJ_TRMM, ADJ_CDR, ADJ_CHIR, and ADJ_CMOR). The bias-corrected products were compared against a gauge-based dataset in terms of rainfall analysis, and their performance within the SWAT model was assessed over calibration (2004–2013) and validation (2014–2015). Bias-corrected products demonstrated superior performance in rainfall analysis, with ADJ_TRMM outperforming other products. The SWAT model calibration results showed satisfactory performance across all stations, with a Nash-Sutcliffe Efficiency (NSE) ranging from [0.76–0.87]. Integrating bias-corrected SPPs into the SWAT model significantly increased the RR simulations in the MRB, indicated by higher NSE values [0.72–0.85] compared to uncorrected SPPs [-0.37 to 0.85] at the Kratie station. Besides, the inconsistent performance of bias-corrected products between rainfall analysis and RR modeling was observed, with ADJ_CDR outperforming ADJ_TRMM in the SWAT model. These results highlight the significance of using bias-corrected SPPs in hydrological modeling applications, especially in areas with limited ground-based precipitation data, and highlight the need for further research to refine bias correction methods and address the limitations of the SWAT model.

Novel Antenna for Partial Discharge Detection and Classification: A Convolutional Neural Network-Based Deep Learning Approach

Novel Antenna for Partial Discharge Detection and Classification: A Convolutional Neural Network-Based Deep Learning Approach

DeepRetroMoCo: deep neural network-based retrospective motion correction algorithm for spinal cord functional MRI.

There are distinct challenges in the preprocessing of spinal cord fMRI data, particularly concerning the mitigation of voluntary or involuntary movement artifacts during image acquisition. Despite the notable progress in data processing techniques for movement detection and correction, applying motion correction algorithms developed for the brain cortex to the brainstem and spinal cord remains a challenging endeavor. In this study, we employed a deep learning-based convolutional neural network (CNN) named DeepRetroMoCo, trained using an unsupervised learning algorithm. Our goal was to detect and rectify motion artifacts in axial T2*-weighted spinal cord data. The training dataset consisted of spinal cord fMRI data from 27 participants, comprising 135 runs for training and 81 runs for testing. To evaluate the efficacy of DeepRetroMoCo, we compared its performance against the sct_fmri_moco method implemented in the spinal cord toolbox. We assessed the motion-corrected images using two metrics: the average temporal signal-to-noise ratio (tSNR) and Delta Variation Signal (DVARS) for both raw and motion-corrected data. Notably, the average tSNR in the cervical cord was significantly higher when DeepRetroMoCo was utilized for motion correction, compared to the sct_fmri_moco method. Additionally, the average DVARS values were lower in images corrected by DeepRetroMoCo, indicating a superior reduction in motion artifacts. Moreover, DeepRetroMoCo exhibited a significantly shorter processing time compared to sct_fmri_moco. Our findings strongly support the notion that DeepRetroMoCo represents a substantial improvement in motion correction procedures for fMRI data acquired from the cervical spinal cord. This novel deep learning-based approach showcases enhanced performance, offering a promising solution to address the challenges posed by motion artifacts in spinal cord fMRI data.

Network anomaly detection and performance evaluation of Convolutional Neural Networks on UNSW-NB15 dataset

The present study uses the benchmark UNSW-NB15 datasets to detect network anomalies using the proposed deep learning-based Convolutional Neural Network (CNN) model. Several studies have already worked on network anomaly detection with some limits. However, earlier research has used old datasets and obtained limited accuracy. In addition, previous studies could not detect the latest malicious activities. Therefore, in the present study, firstly, we implemented the machine learning-based random forest method to diminish the dimensionality of the dataset and pick the most notable features. In this case, the random forest method has set only fifteen features from the forty-one and reduced the complexity of the data. Subsequently, the CNN model was trained and tested on the UNSW-NB15 dataset with 99.00% testing accuracy. The performance of the offered CNN model was assessed using precision, recall and the f1-values. The present study’s outcome can be used in real-time malicious activity detection.

Efficient Diagnosis of Bacterial Leaf Spot in Tomato Plants using Deep Learning CNN Models

Latest field surveys show that the bacterial spot caused by Xanthomonas campestris pv. vesicatoria (Xcv) has devastated commercial tomato output all over the world. The purpose of this study is to more quickly and accurately diagnose bacterial leaf spot problems in tomato plants. To do this, a deep learning-based Convolutional Neural Network (CNN) named Xantho_Net has been proposed to classify the bacterial and non-bacterial tomato leafs. Various state-of-art pre-trained models i.e., VGG16, MobileNet, DenseNet_201, Inception-ResNet_v2 were selected for comparison purposes. Experiments proved that the proposed CNN model is more accurate than other models with 99% accuracy.

Detection of the separated endodontic instrument on periapical radiographs using a deep learning-based convolutional neural network algorithm.

The study evaluated the diagnostic performance of an artificial intelligence system to detect separated endodontic instruments on periapical radiograph radiographs. Three hundred seven periapical radiographs were collected and divided into 222 for training and 85 for testing to be fed to the Mask R-CNN model. Periapical radiographs were assigned to the training and test set and labelled on the DentiAssist labeling platform. Labelled polygonal objects had their bounding boxes automatically generated by the DentiAssist system. Fractured instruments were classified and segmented. As a result of the proposed method, the mean average precision (mAP) metric was 98.809%, the precision value was 95.238, while the recall reached 98.765 and the f1 score 96.969%. The threshold value of 80% was chosen for the bounding boxes working with the Intersection over Union (IoU) technique. The Mask R-CNN distinguished separated endodontic instruments on periapical radiographs.

Geographical discrimination of Asian red pepper powders using 1H NMR spectroscopy and deep learning-based convolution neural networks

This study investigated an innovative approach to discriminate the geographical origins of Asian red pepper powders by analyzing one-dimensional 1H NMR spectra through a deep learning-based convolution neural network (CNN). 1H NMR spectra were collected from 300 samples originating from China, Korea, and Vietnam and used as input data. Principal component analysis − linear discriminant analysis and support vector machine models were employed for comparison. Bayesian optimization was used for hyperparameter optimization, and cross-validation was performed to prevent overfitting. As a result, all three models discriminated the origins of the test samples with over 95 % accuracy. Specifically, the CNN models achieved a 100 % accuracy rate. Gradient-weighted class activation mapping analysis verified that the CNN models recognized the origins of the samples based on variations in metabolite distributions. This research demonstrated the potential of deep learning-based classification of 1H NMR spectra as an accurate and reliable approach for determining the geographical origins of various foods.

Dental bitewing radiographs segmentation using deep learning-based convolutional neural network algorithms.

Dental radiographs, particularly bitewing radiographs, are widely used in dental diagnosis and treatment Dentalimage segmentation is difficult for various reasons, such as intricate structures, low contrast, noise, roughness, andunclear borders, resulting in poor image quality. Recent developments in deep learning models have improvedperformance in analyzing dental images. In this research, our primary objective is to determine the most effectivesegmentation technique for bitewing radiographs based on different metrics: accuracy, training time, and thenumber of training parameters as a reflection of architectural cost. In this research, we employ several deep learning models, namely Resnet-18, Resnet-50, Xception, InceptionResnet v2, and Mobilenetv2, to segment bitewing radiographs. The process begins by importing the radiographsinto MATLAB®(MathWorks Inc), where the images are first improved, then segmented using the graph cut methodbased on regions to produce a binary mask that distinguishes the background from the original X-ray. The deep learning models were trained on 298 and 99 radiograph training and validation sets and were evaluatedusing 99 images from the testing set. We also compare the segmentation model using several criteria, includingaccuracy, speed, and size, to determine which network is superior. Furthermore, we compare our findings with priorresearch to provide a comprehensive understanding of the advancements made in dental image segmentation. Theaccurate segmentation achieved was 93.67% and 94.42% by the Resnet-18 and Resnet-50 models, respectively. This research advances dental image analysis and facilitates more accurate diagnoses and treatment planning bydetermining the best segmentation technique. The outcomes of this study can guide researchers and practitionersin selecting appropriate segmentation methods for practical dental image analysis.

3-D Tactile-Based Object Recognition for Robot Hands Using Force-Sensitive and Bend Sensor Arrays

Tactile sensing is a particularly important and challenging task for a modern robot to safely manipulate objects, interact with humans in a shared space, and provide various services. This paper presents a 3D tactile glove for robots with the combination of a piezoresistive-based force sensor array (412 sensors) covering the full hand and a resistive bend sensor array (5 sensors) on the back of five fingers. Deep learning-based CNN (Convolutional Neural Network) methods using the designed tactile glove are proposed for object recognition. In the experiment for recognizing 15 objects with a dexterous robot hand, an average classification accuracy of 93.67% has been achieved. Comparison experiments with three other typical classifiers (the Quadratic SVM, Weighted KNN, and Bagged Trees) and our CNN methods show an average recognition accuracy of 91.67% with the 3D tactile glove, revealing an accuracy improvement of 4.17% over only using the force sensor array. We further apply our 3D tactile glove and the multi-modal CNN to identify three other objects and demonstrate their generalization ability of tactile object recognition with an average success accuracy of 78.33%. The proposed 3D tactile glove can be further used in human-robot interactions, the design of prosthetics and humanoid robots, and for improving the intelligence level in brain-computer collaborative systems.

A Deep-Learning Algorithm to Predict Short-Term Progression to Geographic Atrophy on Spectral-Domain Optical Coherence Tomography

The identification of patients at risk of progressing from intermediate age-related macular degeneration (iAMD) to geographic atrophy (GA) is essential for clinical trials aimed at preventing disease progression. DeepGAze is a fully automated and accurate convolutional neural network-based deep learning algorithm for predicting progression from iAMD to GA within 1 year from spectral-domain optical coherence tomography (SD-OCT) scans. To develop a deep-learning algorithm based on volumetric SD-OCT scans to predict the progression from iAMD to GA during the year following the scan. This retrospective cohort study included participants with iAMD at baseline and who either progressed or did not progress to GA within the subsequent 13 months. Participants were included from centers in 4 US states. Data set 1 included patients from the Age-Related Eye Disease Study 2 AREDS2 (Ancillary Spectral-Domain Optical Coherence Tomography) A2A study (July 2008 to August 2015). Data sets 2 and 3 included patients with imaging taken in routine clinical care at a tertiary referral center and associated satellites between January 2013 and January 2023. The stored imaging data were retrieved for the purpose of this study from July 1, 2022, to February 1, 2023. Data were analyzed from May 2021 to July 2023. A position-aware convolutional neural network with proactive pseudointervention was trained and cross-validated on Bioptigen SD-OCT volumes (data set 1) and validated on 2 external data sets comprising Heidelberg Spectralis SD-OCT scans (data sets 2 and 3). Prediction of progression to GA within 13 months was evaluated with area under the receiver-operator characteristic curves (AUROC) as well as area under the precision-recall curve (AUPRC), sensitivity, specificity, positive predictive value, negative predictive value, and accuracy. The study included a total of 417 patients: 316 in data set 1 (mean [SD] age, 74 [8]; 185 [59%] female), 53 in data set 2, (mean [SD] age, 83 [8]; 32 [60%] female), and 48 in data set 3 (mean [SD] age, 81 [8]; 32 [67%] female). The AUROC for prediction of progression from iAMD to GA within 1 year was 0.94 (95% CI, 0.92-0.95; AUPRC, 0.90 [95% CI, 0.85-0.95]; sensitivity, 0.88 [95% CI, 0.84-0.92]; specificity, 0.90 [95% CI, 0.87-0.92]) for data set 1. The addition of expert-annotated SD-OCT features to the model resulted in no improvement compared to the fully autonomous model (AUROC, 0.95; 95% CI, 0.92-0.95; P = .19). On an independent validation data set (data set 2), the model predicted progression to GA with an AUROC of 0.94 (95% CI, 0.91-0.96; AUPRC, 0.92 [0.89-0.94]; sensitivity, 0.91 [95% CI, 0.74-0.98]; specificity, 0.80 [95% CI, 0.63-0.91]). At a high-specificity operating point, simulated clinical trial recruitment was enriched for patients progressing to GA within 1 year by 8.3- to 20.7-fold (data sets 2 and 3). The fully automated, position-aware deep-learning algorithm assessed in this study successfully predicted progression from iAMD to GA over a clinically meaningful time frame. The ability to predict imminent GA progression could facilitate clinical trials aimed at preventing the condition and could guide clinical decision-making regarding screening frequency or treatment initiation.

Structural and functional connectome relationships in early childhood

There is strong evidence that the functional connectome is highly related to the white matter connectome in older children and adults, though little is known about structure-function relationships in early childhood. We investigated the development of cortical structure-function coupling in children longitudinally scanned at 1, 2, 4, and 6 years of age (N = 360) and in a comparison sample of adults (N = 89). We also applied a novel graph convolutional neural network-based deep learning model with a new loss function to better capture inter-subject heterogeneity and predict an individual’s functional connectivity from the corresponding structural connectivity. We found regional patterns of structure-function coupling in early childhood that were consistent with adult patterns. In addition, our deep learning model improved the prediction of individual functional connectivity from its structural counterpart compared to existing models.