Mammography-based deep learning model for coronary artery calcification.

AIMS: Mammography, commonly used for breast cancer screening in women, can also predict cardiovascular disease. We developed mammography-based deep learning models for predicting coronary artery calcium (CAC) scores, an established predictor of coronary events. METHODS AND RESULTS: We evaluated a subset of Korean adults who underwent image mammography and CAC computed tomography and randomly selected approximately 80% of the participants as the training dataset, used to develop a convolutional neural network (CNN) to predict detectable CAC. The sensitivity, specificity, area under the receiver operating characteristic curve (AUROC), and overall accuracy of the model's performance were evaluated. The training and validation datasets included 5235 and 1208 women, respectively [mean age, 52.6 (±10.2) years], including non-zero cases (46.8%). The CNN-based deep learning prediction model based on the Resnet18 model showed the best performance. The model was further improved using contrastive learning strategies based on positive and negative samples: sensitivity, 0.764 (95% CI, 0.667-0.830); specificity, 0.652 (95% CI, 0.614-0.710); AUROC, 0.761 (95% CI, 0.742-0.780); and accuracy, 70.8% (95% CI, 68.8-72.4). Moreover, including age and menopausal status in the model further improved its performance (AUROC, 0.776; 95% CI, 0.762-0.790). The Framingham risk score yielded an AUROC of 0.736 (95% CI, 0.712-0.761). CONCLUSION: Mammography-based deep learning models showed promising results for predicting CAC, performing comparably to conventional risk models. This indicates mammography's potential for dual-risk assessment in breast cancer and cardiovascular disease. Further research is necessary to validate these findings in diverse populations, with a particular focus on representation from national breast screening programmes.

Improving the Accuracy of Continuous Blood Glucose Measurement Using Personalized Calibration and Machine Learning.

Despite tremendous developments in continuous blood glucose measurement (CBGM) sensors, they are still not accurate for all patients with diabetes. As glucose concentration in the blood is <1% of the total blood volume, it is challenging to accurately measure glucose levels in the interstitial fluid using CBGM sensors due to within-patient and between-patient variations. To address this issue, we developed a novel data-driven approach to accurately predict CBGM values using personalized calibration and machine learning. First, we scientifically divided measured blood glucose into smaller groups, namely, hypoglycemia (<80 mg/dL), nondiabetic (81-115 mg/dL), prediabetes (116-150 mg/dL), diabetes (151-181 mg/dL), severe diabetes (181-250 mg/dL), and critical diabetes (>250 mg/dL). Second, we separately trained each group using different machine learning models based on patients' personalized parameters, such as physical activity, posture, heart rate, breath rate, skin temperature, and food intake. Lastly, we used multilayer perceptron (MLP) for the D1NAMO dataset (training to test ratio: 70:30) and grid search for hyperparameter optimization to predict accurate blood glucose concentrations. We successfully applied our proposed approach in nine patients with type 1 diabetes and observed that the mean absolute relative difference (MARD) decreased from 17.8% to 8.3%.

Neurologic Dysfunction Assessment in Parkinson Disease Based on Fundus Photographs Using Deep Learning.

Importance: Until now, other than complex neurologic tests, there have been no readily accessible and reliable indicators of neurologic dysfunction among patients with Parkinson disease (PD). This study was conducted to determine the role of fundus photography as a noninvasive and readily available tool for assessing neurologic dysfunction among patients with PD using deep learning methods. Objective: To develop an algorithm that can predict Hoehn and Yahr (H-Y) scale and Unified Parkinson's Disease Rating Scale part III (UPDRS-III) score using fundus photography among patients with PD. Design, Settings, and Participants: This was a prospective decision analytical model conducted at a single tertiary-care hospital. The fundus photographs of participants with PD and participants with non-PD atypical motor abnormalities who visited the neurology department of Kangbuk Samsung Hospital from October 7, 2020, to April 30, 2021, were analyzed in this study. A convolutional neural network was developed to predict both the H-Y scale and UPDRS-III score based on fundus photography findings and participants' demographic characteristics. Main Outcomes and Measures: The area under the receiver operating characteristic curve (AUROC) was calculated for sensitivity and specificity analyses for both the internal and external validation data sets. Results: A total of 615 participants were included in the study: 266 had PD (43.3%; mean [SD] age, 70.8 [8.3] years; 134 male individuals [50.4%]), and 349 had non-PD atypical motor abnormalities (56.7%; mean [SD] age, 70.7 [7.9] years; 236 female individuals [67.6%]). For the internal validation data set, the sensitivity was 83.23% (95% CI, 82.07%-84.38%) and 82.61% (95% CI, 81.38%-83.83%) for the H-Y scale and UPDRS-III score, respectively. The specificity was 66.81% (95% CI, 64.97%-68.65%) and 65.75% (95% CI, 62.56%-68.94%) for the H-Y scale and UPDRS-III score, respectively. For the external validation data set, the sensitivity and specificity were 70.73% (95% CI, 66.30%-75.16%) and 66.66% (95% CI, 50.76%-82.25%), respectively. Lastly, the calculated AUROC and accuracy were 0.67 (95% CI, 0.55-0.79) and 70.45% (95% CI, 66.85%-74.04%), respectively. Conclusions and Relevance: This decision analytical model reveals amalgamative insights into the neurologic dysfunction among PD patients by providing information on how to apply a deep learning method to evaluate the association between the retina and brain. Study data may help clarify recent research findings regarding dopamine pathologic cascades between the retina and brain among patients with PD; however, further research is needed to expand the clinical implication of this algorithm.

Generating future fundus images for early age-related macular degeneration based on generative adversarial networks.

BACKGROUND AND OBJECTIVE: Age-related macular degeneration (AMD) is one of the most common diseases that can lead to blindness worldwide. Recently, various fundus image analyzing studies are done using deep learning methods to classify fundus images to aid diagnosis and monitor AMD disease progression. But until now, to the best of our knowledge, no attempt was made to generate future synthesized fundus images that can predict AMD progression. In this paper, we developed a deep learning model using fundus images for AMD patients with different time elapses to generate synthetic future fundus images. METHOD: We exploit generative adversarial networks (GANs) with additional drusen masks to maintain the pathological information. The dataset included 8196 fundus images from 1263 AMD patients. A proposed GAN-based model, called Multi-Modal GAN (MuMo-GAN), was trained to generate synthetic predicted-future fundus images. RESULTS: The proposed deep learning model indicates that the additional drusen masks can help to learn the AMD progression. Our model can generate future fundus images with appropriate pathological features. The drusen development over time is depicted well. Both qualitative and quantitative experiments show that our model is more efficient to monitor the AMD disease as compared to other studies. CONCLUSION: This study could help individualized risk prediction for AMD patients. Compared to existing methods, the experimental results show a significant improvement in terms of tracking the AMD stage in both image-level and pixel-level.

Multi-Site Infant Brain Segmentation Algorithms: The iSeg-2019 Challenge.

To better understand early brain development in health and disorder, it is critical to accurately segment infant brain magnetic resonance (MR) images into white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF). Deep learning-based methods have achieved state-of-the-art performance; h owever, one of the major limitations is that the learning-based methods may suffer from the multi-site issue, that is, the models trained on a dataset from one site may not be applicable to the datasets acquired from other sites with different imaging protocols/scanners. To promote methodological development in the community, the iSeg-2019 challenge (http://iseg2019.web.unc.edu) provides a set of 6-month infant subjects from multiple sites with different protocols/scanners for the participating methods. T raining/validation subjects are from UNC (MAP) and testing subjects are from UNC/UMN (BCP), Stanford University, and Emory University. By the time of writing, there are 30 automatic segmentation methods participated in the iSeg-2019. In this article, 8 top-ranked methods were reviewed by detailing their pipelines/implementations, presenting experimental results, and evaluating performance across different sites in terms of whole brain, regions of interest, and gyral landmark curves. We further pointed out their limitations and possible directions for addressing the multi-site issue. We find that multi-site consistency is still an open issue. We hope that the multi-site dataset in the iSeg-2019 and this review article will attract more researchers to address the challenging and critical multi-site issue in practice.

Incorporated region detection and classification using deep convolutional networks for bone age assessment.

Bone age assessment plays an important role in the endocrinology and genetic investigation of patients. In this paper, we proposed a deep learning-based approach for bone age assessment by integration of the Tanner-Whitehouse (TW3) methods and deep convolution networks based on extracted regions of interest (ROI)-detection and classification using Faster-RCNN and Inception-v4 networks, respectively. The proposed method allows exploration of expert knowledge from TW3 and features engineering from deep convolution networks to enhance the accuracy of bone age assessment. The experimental results showed a mean absolute error of about 0.59 years between expert radiologists and the proposed method, which is the best performance among state-of-the-art methods.

Benchmark on Automatic 6-month-old Infant Brain Segmentation Algorithms: The iSeg-2017 Challenge.

Accurate segmentation of infant brain magnetic resonance (MR) images into white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) is an indispensable foundation for early studying of brain growth patterns and morphological changes in neurodevelopmental disorders. Nevertheless, in the isointense phase (approximately 6-9 months of age), due to inherent myelination and maturation process, WM and GM exhibit similar levels of intensity in both T1-weighted (T1w) and T2-weighted (T2w) MR images, making tissue segmentation very challenging. Despite many efforts were devoted to brain segmentation, only few studies have focused on the segmentation of 6-month infant brain images. With the idea of boosting methodological development in the community, iSeg-2017 challenge (http://iseg2017.web.unc.edu) provides a set of 6-month infant subjects with manual labels for training and testing the participating methods. Among the 21 automatic segmentation methods participating in iSeg-2017, we review the 8 top-ranked teams, in terms of Dice ratio, modified Hausdorff distance and average surface distance, and introduce their pipelines, implementations, as well as source codes. We further discuss limitations and possible future directions. We hope the dataset in iSeg-2017 and this review article could provide insights into methodological development for the community.

Distributed Systematic Network Coding for Reliable Content Uploading in Wireless Multimedia Sensor Networks.

Recently, the wireless sensor network paradigm is shifting toward research aimed at enabling the robust delivery of multimedia content. A challenge is to deliver multimedia content with predefined levels of Quality of Service (QoS) under resource constraints such as bandwidth, energy, and delay. In this paper, we propose a distributed systematic network coding (DSNC) scheme for reliable multimedia content uploading over wireless multimedia sensor networks, in which a large number of multimedia sensor nodes upload their own content to a sink through a cluster head node. The design objective is to increase the reliability and bandwidth-efficient utilization in uploading with low decoding complexity. The proposed scheme consists of two phases: in the first phase, each sensor node distributedly encodes the content into systematic network coding packets and transmits them to the cluster head; then in the second phase, the cluster head encodes all successfully decoded incoming packets from multiple sensor nodes into innovative systematic network coding packets and transmits them to the sink. A bandwidth-efficient and channel-aware error control algorithm is proposed to enhance the bandwidth-efficient utilization by dynamically determining the optimal number of innovative coded packets. For performance analysis and evaluation, we firstly derive the closed-form equations of decoding probability to validate the effectiveness of the proposed uploading scheme. Furthermore, we perform various simulations along with a discussion in terms of three performance metrics: decoding probability, redundancy, and image quality measurement. The analytical and experimental results demonstrate that the performance of our proposed DSNC outperforms the existing uploading schemes.

Fully automated, level set-based segmentation for knee MRIs using an adaptive force function and template: data from the osteoarthritis initiative.

BACKGROUND: This study focuses on osteoarthritis (OA), which affects millions of adults and occurs in knee cartilage. Diagnosis of OA requires accurate segmentation of cartilage structures. Existing approaches to cartilage segmentation of knee imaging suffer from either lack of fully automatic algorithm, sub-par segmentation accuracy, or failure to consider all three cartilage tissues. METHODS: We propose a novel segmentation algorithm for knee cartilages with level set-based segmentation method and novel template data. We used 20 normal subjects from osteoarthritis initiative database to construct new template data. We adopt spatial fuzzy C-mean clustering for automatic initialization of contours. Force function of our algorithm is modified to improve segmentation performance. RESULTS: The proposed algorithm resulted in dice similarity coefficients (DSCs) of 87.1, 84.8 and 81.7 % for the femoral, patellar, and tibial cartilage, respectively from 10 subjects. The DSC results showed improvements of 8.8, 4.3 and 3.5 % for the femoral, patellar, and tibial cartilage respectively compared to existing approaches. Our algorithm could be applied to all three cartilage structures unlike existing approaches that considered only two cartilage tissues. CONCLUSIONS: Our study proposes a novel fully automated segmentation algorithm adapted for three types of knee cartilage tissues. We leverage state-of-the-art level set approach with newly constructed knee template. The experimental results show that the proposed method improves the performance by an average of 5 % over existing methods.