Artificial Intelligence Pipeline Research Articles

Deep Learning Models for Magnetic Cardiography Edge Sensors Implementing Noise Processing and Diagnostics

Remote health monitoring has become a necessity due to reduced healthcare access resulting from pandemic lockdowns and the increasing aging population. Electrocardiography (ECG) is the standard for cardiac monitoring and arrhythmia identification, but it is inconvenient for long-time remote monitoring. Recently, Magnetocardiography (MCG) sensors that operate at room temperature became available based on spintronic sensors. However, MCG analysis is affected by the low-frequency noise present at the sensors. In this paper, we present an artificial intelligence (AI)-aided multi-model pipeline combining two AI architectures, defined as model-M1 and model-M2, targeted for ultra-edge Internet of Things (IoT) sensors to simulate arrhythmia detection. Model-M1 is a denoising preprocessor based on a sliding-window assisted deep-learning (DL) model. We investigate various methods to achieve high accuracy with lightweight computation. Model-M2 is a lightweight DL model that analyzes denoised ECG output from model-M1 to identify arrhythmia. We use multiple publicly available clinically annotated datasets to evaluate our proposal. We find that denoising by model-M1 retains the features, which assist the model-M2 in achieving high classification accuracy, compared to using a conventional moving average filter. This AI pipeline architecture is promising for privacy-preserving ultra-edge medical sensing devices.

A narrative review of artificial intelligence-assisted histopathologic diagnosis and decision-making for non-small cell lung cancer: achievements and limitations.

ObjectiveTo summarize the current evidence regarding the applications, workflow, and limitations of artificial intelligence (AI) in the management of patients pathologically-diagnosed with lung cancer.BackgroundLung cancer is one of the most common cancers and the leading cause of cancer-related deaths worldwide. AI technologies have been applied to daily medical workflow and have achieved an excellent performance in predicting histopathologic subtypes, analyzing gene mutation profiles, and assisting in clinical decision-making for lung cancer treatment. More advanced deep learning for classifying pathologic images with minimal human interactions has been developed in addition to the conventional machine learning scheme.MethodsStudies were identified by searching databases, including PubMed, EMBASE, Web of Science, and Cochrane Library, up to February 2021 without language restrictions.ConclusionsA number of studies have evaluated AI pipelines and confirmed that AI is robust and efficacious in lung cancer diagnosis and decision-making, demonstrating that AI models are a useful tool for assisting oncologists in health management. Although several limitations that pose an obstacle for the widespread use of AI schemes persist, the unceasing refinement of AI techniques is poised to overcome such problems. Thus, AI technology is a promising tool for use in diagnosing and managing lung cancer.

Toward Generalizability in the Deployment of Artificial Intelligence in Radiology: Role of Computation Stress Testing to Overcome Underspecification.

The clinical deployment of artificial intelligence (AI) applications in medical imaging is perhaps the greatest challenge facing radiology in the next decade. One of the main obstacles to the incorporation of automated AI-based decision-making tools in medicine is the failure of models to generalize when deployed across institutions with heterogeneous populations and imaging protocols. The most well-understood pitfall in developing these AI models is overfitting, which has, in part, been overcome by optimizing training protocols. However, overfitting is not the only obstacle to the success and generalizability of AI. Underspecification is also a serious impediment that requires conceptual understanding and correction. It is well known that a single AI pipeline, with prescribed training and testing sets, can produce several models with various levels of generalizability. Underspecification defines the inability of the pipeline to identify whether these models have embedded the structure of the underlying system by using a test set independent of, but distributed identically, to the training set. An underspecified pipeline is unable to assess the degree to which the models will be generalizable. Stress testing is a known tool in AI that can limit underspecification and, importantly, assure broad generalizability of AI models. However, the application of stress tests is new in radiologic applications. This report describes the concept of underspecification from a radiologist perspective, discusses stress testing as a specific strategy to overcome underspecification, and explains how stress tests could be designed in radiology-by modifying medical images or stratifying testing datasets. In the upcoming years, stress tests should become in radiology the standard that crash tests have become in the automotive industry. Keywords: Computer Applications-General, Informatics, Computer-aided Diagnosis © RSNA, 2021.

Artificial Intelligence for Automatic Measurement of Left Ventricular Strain in Echocardiography

ObjectivesThis study sought to examine if fully automated measurements of global longitudinal strain (GLS) using a novel motion estimation technology based on deep learning and artificial intelligence (AI) are feasible and comparable with a conventional speckle-tracking application. BackgroundGLS is an important parameter when evaluating left ventricular function. However, analyses of GLS are time consuming and demand expertise, and thus are underused in clinical practice. MethodsIn this study, 200 patients with a wide range of left ventricle (LV) function were included. Three standard apical cine-loops were analyzed using the AI pipeline. The AI method measured GLS and was compared with a commercially available semiautomatic speckle-tracking software (EchoPAC v202, GE Healthcare. ResultsThe AI method succeeded to both correctly classify all 3 standard apical views and perform timing of cardiac events in 89% of patients. Furthermore, the method successfully performed automatic segmentation, motion estimates, and measurements of GLS in all examinations, across different cardiac pathologies and throughout the spectrum of LV function. GLS was −12.0 ± 4.1% for the AI method and −13.5 ± 5.3% for the reference method. Bias was −1.4 ± 0.3% (95% limits of agreement: 2.3 to −5.1), which is comparable with intervendor studies. The AI method eliminated measurement variability and a complete GLS analysis was processed within 15 s. ConclusionsThrough the range of LV function this novel AI method succeeds, without any operator input, to automatically identify the 3 standard apical views, perform timing of cardiac events, trace the myocardium, perform motion estimation, and measure GLS. Fully automated measurements based on AI could facilitate the clinical implementation of GLS.

Artificial Intelligence for Unstructured Healthcare Data: Application to Coding of Patient Reporting of Adverse Drug Reactions.

Adverse drug reaction (ADR) reporting is a major component of drug safety monitoring; its input will, however, only be optimized if systems can manage to deal with its tremendous flow of information, based primarily on unstructured text fields. The aim of this study was to develop an automated system allowing to code ADRs from patient reports. Our system was based on a knowledge base about drugs, enriched by supervised machine learning (ML) models trained on patients reporting data. To train our models, we selected all cases of ADRs reported by patients to a French Pharmacovigilance Centre through a national web‐portal between March 2017 and March 2019 (n = 2,058 reports). We tested both conventional ML models and deep‐learning models. We performed an external validation using a dataset constituted of a random sample of ADRs reported to the Marseille Pharmacovigilance Centre over the same period (n = 187). Here, we show that regarding area under the curve (AUC) and F‐measure, the best model to identify ADRs was gradient boosting trees (LGBM), with an AUC of 0.93 (0.92–0.94) and F‐measure of 0.72 (0.68–0.75). This model was run for external validation showing an AUC of 0.91 and a F‐measure of 0.58. We evaluated an artificial intelligence pipeline that was found able to learn how to identify correctly ADRs from unstructured data. This result allowed us to start a new study using more data to further improve our performance and offer a tool that is useful in practice to efficiently manage drug safety information.

Managing and Deploying Distributed and Deep Neural Models Through Kafka-ML in the Cloud-to-Things Continuum

The Internet of Things (IoT) is constantly growing, generating an uninterrupted data stream pipeline to monitor physical world information. Hence, Artificial Intelligence (AI) continuously evolves, improving life quality and business and academic activities. Kafka-ML is an open-source framework that focuses on managing Machine Learning (ML) and AI pipelines through data streams in production scenarios. Consequently, it facilitates Deep Neural Network (DNN) deployments in real-world applications. However, this framework does not consider the distribution of DNN models on the Cloud-to-Things Continuum. Distributed DNN significantly reduces latency, allocating the computational and network load between different infrastructures. In this work, we have extended our Kafka-ML framework to support the management and deployment of Distributed DNN throughout the Cloud-to-Things Continuum. Moreover, we have considered the possibility of including early exits in the Cloud-to-Things layers to provide immediate responses upon predictions. We have evaluated these new features by adapting and deploying the DNN model AlexNet in three different Cloud-to-Things scenarios. Experiments demonstrate that Kafka-ML can significantly improve response time and throughput by distributing DNN models throughout the Cloud-to-Things Continuum, compared to a Cloud-only deployment.

Browser-based Data Annotation, Active Learning, and Real-Time Distribution of Artificial Intelligence Models: From Tumor Tissue Microarrays to COVID-19 Radiology

Background: Artificial intelligence (AI) is fast becoming the tool of choice for scalable and reliable analysis of medical images. However, constraints in sharing medical data outside the institutional or geographical space, as well as difficulties in getting AI models and modeling platforms to work across different environments, have led to a “reproducibility crisis” in digital medicine. Methods: This study details the implementation of a web platform that can be used to mitigate these challenges by orchestrating a digital pathology AI pipeline, from raw data to model inference, entirely on the local machine. We discuss how this federated platform provides governed access to data by consuming the Application Program Interfaces exposed by cloud storage services, allows the addition of user-defined annotations, facilitates active learning for training models iteratively, and provides model inference computed directly in the web browser at practically zero cost. The latter is of particular relevance to clinical workflows because the code, including the AI model, travels to the user's data, which stays private to the governance domain where it was acquired. Results: We demonstrate that the web browser can be a means of democratizing AI and advancing data socialization in medical imaging backed by consumer-facing cloud infrastructure such as Box.com. As a case study, we test the accompanying platform end-to-end on a large dataset of digital breast cancer tissue microarray core images. We also showcase how it can be applied in contexts separate from digital pathology by applying it to a radiology dataset containing COVID-19 computed tomography images. Conclusions: The platform described in this report resolves the challenges to the findable, accessible, interoperable, reusable stewardship of data and AI models by integrating with cloud storage to maintain user-centric governance over the data. It also enables distributed, federated computation for AI inference over those data and proves the viability of client-side AI in medical imaging.Availability: The open-source application is publicly available at https://episphere.github.io/path, with a short video demonstration at https://youtu.be/z59jToy2TxE.

The potential of video imagery from worldwide cabled observatory networks to provide information supporting fish-stock and biodiversity assessment

AbstractSeafloor multiparametric fibre-optic-cabled video observatories are emerging tools for standardized monitoring programmes, dedicated to the production of real-time fishery-independent stock assessment data. Here, we propose that a network of cabled cameras can be set up and optimized to ensure representative long-term monitoring of target commercial species and their surrounding habitats. We highlight the importance of adding the spatial dimension to fixed-point-cabled monitoring networks, and the need for close integration with Artificial Intelligence pipelines, that are necessary for fast and reliable biological data processing. We then describe two pilot studies, exemplary of using video imagery and environmental monitoring to derive robust data as a foundation for future ecosystem-based fish-stock and biodiversity management. The first example is from the NE Pacific Ocean where the deep-water sablefish (Anoplopoma fimbria) has been monitored since 2010 by the NEPTUNE cabled observatory operated by Ocean Networks Canada. The second example is from the NE Atlantic Ocean where the Norway lobster (Nephrops norvegicus) is being monitored using the SmartBay observatory developed for the European Multidisciplinary Seafloor and water column Observatories. Drawing from these two examples, we provide insights into the technological challenges and future steps required to develop full-scale fishery-independent stock assessments.

Abstract 854: An artificial intelligence pipeline for imaging mass cytometry data analysis and its application in ovarian cancer prognostic biomarker discovery

Abstract Imaging mass cytometry (IMC) is a powerful platform which enables high-dimensional, single-cell analysis of cell type and state. However, reliable methods used to analyze the IMC data remain to be developed. We seek to build an artificial intelligence (AI)-based analytics pipeline for imaging mass cytometry (IMC) to increase the accuracy of cell segmentation and spatial information extraction and apply the AI pipeline to analyze the IMC data derived from high-grade serous ovarian cancer samples for patient overall survival prediction. Multiplexed spatial analysis of the tumor microenvironment by IMC was performed on 41 formalin fixed paraffin embedded (FFPE) tissue samples obtained from treatment naïve high- grade serous ovarian cancer patients using a panel of 24 metal-tagged antibodies that are specific to tumor, immune and stromal cell markers. IMC data was collected using a a Helios CyTOF instrument equipped with Hyperion Imaging System (Fluidigm). Mask Region-Convolution Neural Network (Mask R-CNN) model was used for cell segmentation. Cell subtypes were derived by iterative phenograph clustering with different marker combinations and different subsets of cells in each iteration. Tumor area was calculated as a thresholded Gaussian blurred image of the density map of the center-of-mass of Keratin positive tumor cells. All cells with center-of-mass located within the tumor region were classified as intratumoral cells. The average cell composition in the nearest neighborhood of each cell type (distance between the center of mass of two cells < 20 μm) in the tumor cell compartment was computed. The results showed that cell segmentation by Mask R-CNN has a higher accuracy than traditional watershed segmentation. Significantly more granzymeB+ CD8+ T cells and CD11b+ Vista+ cells were found in long-term survivors comparing to short-term survivors in both tumor and stromal cell compartments of the tumor microenvironment. Additionally, CD196+, CD45RO+ and CD73+ cell densities in the tumor cell compartment were significantly lower in long-term than short-term survivors. Our results also showed that the mean numbers of CD73+ cells adjacent to Vista- CD4+ T cells, macrophages and B cells were significantly lower in long-term than short-term survivors. The mean number of granzymeB+ CD8+ T cells adjacent to Vista- CD4+ T cells was significantly higher in long-term than short-term survivors. These data demonstrated that the deep learning-based cell segmentation method achieved higher accuracy than the conventional watershed segmentation method. Furthermore, our AI pipeline can automatically extract cell count and cell neighborhood information in tumor. The cell count and neighborhood information could be further employed as features for machine learning to generate predictive biomarkers for ovarian cancer patient survival. Citation Format: Ying Zhu, Jianting Sheng, Sammy Ferri-Borgogno, Tsz-Lun Yeung, Jared K. Burks, Samuel C. Mok, Stephen T. Wong. An artificial intelligence pipeline for imaging mass cytometry data analysis and its application in ovarian cancer prognostic biomarker discovery [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 854.

Bonseyes AI Pipeline—Bringing AI to You

Next generation of embedded Information and Communication Technology (ICT) systems are interconnected and collaborative systems able to perform autonomous tasks. The remarkable expansion of the embedded ICT market, together with the rise and breakthroughs of Artificial Intelligence (AI), have put the focus on the Edge as it stands as one of the keys for the next technological revolution: the seamless integration of AI in our daily life. However, training and deployment of custom AI solutions on embedded devices require a fine-grained integration of data, algorithms, and tools to achieve high accuracy and overcome functional and non-functional requirements. Such integration requires a high level of expertise that becomes a real bottleneck for small and medium enterprises wanting to deploy AI solutions on the Edge , which, ultimately, slows down the adoption of AI on applications in our daily life. In this work, we present a modular AI pipeline as an integrating framework to bring data, algorithms, and deployment tools together. By removing the integration barriers and lowering the required expertise, we can interconnect the different stages of particular tools and provide a modular end-to-end development of AI products for embedded devices. Our AI pipeline consists of four modular main steps: (i) data ingestion, (ii) model training, (iii) deployment optimization, and (iv) the IoT hub integration. To show the effectiveness of our pipeline, we provide examples of different AI applications during each of the steps. Besides, we integrate our deployment framework, Low-Power Deep Neural Network (LPDNN), into the AI pipeline and present its lightweight architecture and deployment capabilities for embedded devices. Finally, we demonstrate the results of the AI pipeline by showing the deployment of several AI applications such as keyword spotting, image classification, and object detection on a set of well-known embedded platforms, where LPDNN consistently outperforms all other popular deployment frameworks.

Artificial Intelligence Pipeline to Bridge the Gap between Bench Researchers and Clinical Researchers in Precision Medicine.

Precision medicine informatics is a field of research that incorporates learning systems that generate new knowledge to improve individualized treatments using integrated data sets and models. Given the ever-increasing volumes of data that are relevant to patient care, artificial intelligence (AI) pipelines need to be a central component of such research to speed discovery. Applying AI methodology to complex multidisciplinary information retrieval can support efforts to discover bridging concepts within collaborating communities. This dovetails with precision medicine research, given the information rich multi-omic data that are used in precision medicine analysis pipelines. In this perspective article we define a prototype AI pipeline to facilitate discovering research connections between bioinformatics and clinical researchers. We propose building knowledge representations that are iteratively improved through AI and human-informed learning feedback loops supported through crowdsourcing. To illustrate this, we will explore the specific use case of nonalcoholic fatty liver disease, a growing health care problem. We will examine AI pipeline construction and utilization in relation to bench-to-bedside bridging concepts with interconnecting knowledge representations applicable to bioinformatics researchers and clinicians.

P1926Artificial intelligence in echocardiography for standard clinical metrics

Abstract Introduction A full transthoracic echocardiogram (TTE) study usually takes 40 - 60 mins to perform and report. Our aim was to validate an artificial intelligence (AI) which automatically calculates measurements with manual standard clinical metrics. Methods 41 patients with heart failure (HF) and 19 controls were enrolled retrospectively. A shortened 5-minute TTE exam was performed. Studies were exported from the hospital database in a DICOM format and fed to an AI pipeline to classify, segment and analyse each image. A convolutional neural network (CNN) was used to label each view into one of 23 classes. Views of interest (Apical 2-, 4-Chamber and Parasternal Short/Long Axis) were individually segmented using a segmentation CNN. View classification was trained on 4,000 labelled studies, segmentation models were trained for each view with 72 manually segmented images for PSAX, 128 for PLAX, 168 for A4C and 198 for A2C. The area-length formula was used to calculate left-ventricular volumes (LVEDV/LVESV), ejection fraction (AI-LVEF). Indexed LA volume (LAVOLI) LV mass (LVMI) were also compared. LVEDV, LVESV, LVEF and LVMI were averaged over multiple videos. Results Mean manual LVEF (M-LVEF) in HF patients was 39±10% vs 57±5% in controls. Compute time using was between 4 to 7 mins for classification, segmentation and analysis using a single Graphics Processing Unit (GPU). 11 (18%) non-physiological AI-ESV and associated AI-LVEF were excluded vs 2 (3%) M-LVEF (×2 7 95% CI 3 to 27%, p=0.008). AI generated measurements correlated well with manual measures LVEDV r=0.77, LVESV r=0.8, LVEF r=0.71, LAVOLI r=0.71, LVMI r=0.6, p<0.005. Mean absolute error of M-LVEF vs AI-LVEF was 7.4±6.6%. AI-LVEF, M-LVEF and other HF biomarkers had a similar discrimination for HF (AUC M-LVEF 0.93 vs AI-LVEF 0.88, 95% CI-0.03 to 0.15, p=0.19). AI vs Manual, Correlation Matrix and ROC Conclusion AI with minimal human input is approaching the accuracy required for clinical utility. AI has the ability to distinguish LV systolic dysfunction, and chamber volumes which could be applied to handheld ultrasound in real-time. Acknowledgement/Funding Health Research Council of New Zealand, Auckland Bioengineering Institute

Artificial Intelligence in Echocardiography for Standard Clinical Metrics

Background: A transthoracic echocardiogram (TTE) usually takes 40–60 minutes to perform and report. This study validated an artificial intelligence (AI) that automatically calculates measurements with manual standard clinical metrics. Method: Forty-one patients with heart failure (HF) and 19 controls were retrospectively enrolled. A shortened 5-minute TTE exam was performed. Studies were fed to an AI pipeline to classify, segment, and analyse. A convolutional neural network (CNN) was used to label each view. Views of interest (Apical 2-, 4-Chamber and PLAX or PSAX) were segmented using a segmentation CNN. The area-length formula was used to calculate left ventricular volumes (LVEDV and LVESV) and ejection fraction (AI-LVEF). Indexed left atrial volume (LAVOLI) and LV mass (LVMI) were also compared. The LVEDV, LVESV, LVEF, and LVMI were averaged over multiple videos. Results: Mean manual LVEF (M-LVEF) in HF patients was 39 ± 10% vs 57 ± 5% in controls. Eleven (18%) non-physiological AI-ESV and associated AI-LVEF were excluded vs two (3%) M-LVEF (χ2, 7; 95% CI, 3%-27%; p = 0.008). The AI measurements correlated well with manual measures (LVEDV r = 0.77; LVESV r = 0.8; LVEF r = 0.71; LAVOLI r = 0.71; LVMI r = 0.6; p < 0.005). The AI-LVEF, M-LVEF, and other HF biomarkers had a similar discrimination for HF (AUC M-LVEF 0.93 vs AI-LVEF 0.88; 95% CI, 0.03–0.15; p = 0.19). Conclusion: Artificial intelligence with minimal human input is approaching the accuracy required for clinical utility. It has the ability to distinguish LV systolic dysfunction, and chamber volumes, which could be applied to handheld ultrasound in real-time.

Artificial intelligence pipeline for meniscus segmentation and lesion detection

Artificial intelligence pipeline for meniscus segmentation and lesion detection

Using an artificial intelligence pipeline for digital pathology

Using an artificial intelligence pipeline for digital pathology