Overcoming barriers to data sharing with medical image generation: a comprehensive evaluation.
Ontology highlight
ABSTRACT:
SUBMITTER: DuMont Schutte A
PROVIDER: S-EPMC8463544 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Analyzing medical data to find abnormalities is a time-consuming and costly task, particularly for rare abnormalities, requiring tremendous efforts from medical experts. Therefore, artificial intelligence has become a popular tool for the automatic processing of medical data, acting as a supportive tool for doctors. However, the machine learning models used to build these tools are highly dependent on the data used to train them. Large amounts of data can be difficult to obtain in medicine due to privacy reasons, expensive and time-consuming annotations, and a general lack of data samples for infrequent lesions. In this study, we present a novel synthetic data generation pipeline, called SinGAN-Seg, to produce synthetic medical images with corresponding masks using a single training image. Our method is different from the traditional generative adversarial networks (GANs) because our model needs only a single image and the corresponding ground truth to train. We also show that the synthetic data generation pipeline can be used to produce alternative artificial segmentation datasets with corresponding ground truth masks when real datasets are not allowed to share. The pipeline is evaluated using qualitative and quantitative comparisons between real data and synthetic data to show that the style transfer technique used in our pipeline significantly improves the quality of the generated data and our method is better than other state-of-the-art GANs to prepare synthetic images when the size of training datasets are limited. By training UNet++ using both real data and the synthetic data generated from the SinGAN-Seg pipeline, we show that the models trained on synthetic data have very close performances to those trained on real data when both datasets have a considerable amount of training data. In contrast, we show that synthetic data generated from the SinGAN-Seg pipeline improves the performance of segmentation models when training datasets do not have a considerable amount of data. All experiments were performed using an open dataset and the code is publicly available on GitHub.
| S-EPMC9060378 | biostudies-literature
Project description:ImportanceThe benefits of responsible sharing of individual-participant data (IPD) from clinical studies are well recognized, but stakeholders often disagree on how to align those benefits with privacy risks, costs, and incentives for clinical trialists and sponsors. The International Committee of Medical Journal Editors (ICMJE) required a data sharing statement (DSS) from submissions reporting clinical trials effective July 1, 2018. The required DSSs provide a window into current data sharing rates, practices, and norms among trialists and sponsors.ObjectiveTo evaluate the implementation of the ICMJE DSS requirement in 3 leading medical journals: JAMA, Lancet, and New England Journal of Medicine (NEJM).Design, setting, and participantsThis is a cross-sectional study of clinical trial reports published as articles in JAMA, Lancet, and NEJM between July 1, 2018, and April 4, 2020. Articles not eligible for DSS, including observational studies and letters or correspondence, were excluded. A MEDLINE/PubMed search identified 487 eligible clinical trials in JAMA (112 trials), Lancet (147 trials), and NEJM (228 trials). Two reviewers evaluated each of the 487 articles independently.ExposurePublication of clinical trial reports in an ICMJE medical journal requiring a DSS.Main outcomes and measuresThe primary outcomes of the study were declared data availability and actual data availability in repositories. Other captured outcomes were data type, access, and conditions and reasons for data availability or unavailability. Associations with funding sources were examined.ResultsA total of 334 of 487 articles (68.6%; 95% CI, 64%-73%) declared data sharing, with nonindustry NIH-funded trials exhibiting the highest rates of declared data sharing (89%; 95% CI, 80%-98%) and industry-funded trials the lowest (61%; 95% CI, 54%-68%). However, only 2 IPD sets (0.6%; 95% CI, 0.0%-1.5%) were actually deidentified and publicly available as of April 10, 2020. The remaining were supposedly accessible via request to authors (143 of 334 articles [42.8%]), repository (89 of 334 articles [26.6%]), and company (78 of 334 articles [23.4%]). Among the 89 articles declaring that IPD would be stored in repositories, only 17 (19.1%) deposited data, mostly because of embargo and regulatory approval. Embargo was set in 47.3% of data-sharing articles (158 of 334), and in half of them the period exceeded 1 year or was unspecified.Conclusions and relevanceMost trials published in JAMA, Lancet, and NEJM after the implementation of the ICMJE policy declared their intent to make clinical data available. However, a wide gap between declared and actual data sharing exists. To improve transparency and data reuse, journals should promote the use of unique pointers to data set location and standardized choices for embargo periods and access requirements.
| S-EPMC7844597 | biostudies-literature
Project description:Computer vision (CV) has the potential to change medicine fundamentally. Expert knowledge provided by CV can enhance diagnosis. Unfortunately, existing algorithms often remain below expectations, as databases used for training are usually too small, incomplete, and heterogeneous in quality. Moreover, data protection is a serious obstacle to the exchange of data. To overcome this limitation, we propose to use generative models (GMs) to produce high-resolution synthetic radiographs that do not contain any personal identification information. Blinded analyses by CV and radiology experts confirmed the high similarity of synthesized and real radiographs. The combination of pooled GM improves the performance of CV algorithms trained on smaller datasets, and the integration of synthesized data into patient data repositories can compensate for underrepresented disease entities. By integrating federated learning strategies, even hospitals with few datasets can contribute to and benefit from GM training.
| S-EPMC7821879 | biostudies-literature
Project description:BACKGROUND:In the current information age, the use of data has become essential for decision making in public health at the local, national, and global level. Despite a global commitment to the use and sharing of public health data, this can be challenging in reality. No systematic framework or global operational guidelines have been created for data sharing in public health. Barriers at different levels have limited data sharing but have only been anecdotally discussed or in the context of specific case studies. Incomplete systematic evidence on the scope and variety of these barriers has limited opportunities to maximize the value and use of public health data for science and policy. METHODS:We conducted a systematic literature review of potential barriers to public health data sharing. Documents that described barriers to sharing of routinely collected public health data were eligible for inclusion and reviewed independently by a team of experts. We grouped identified barriers in a taxonomy for a focused international dialogue on solutions. RESULTS:Twenty potential barriers were identified and classified in six categories: technical, motivational, economic, political, legal and ethical. The first three categories are deeply rooted in well-known challenges of health information systems for which structural solutions have yet to be found; the last three have solutions that lie in an international dialogue aimed at generating consensus on policies and instruments for data sharing. CONCLUSIONS:The simultaneous effect of multiple interacting barriers ranging from technical to intangible issues has greatly complicated advances in public health data sharing. A systematic framework of barriers to data sharing in public health will be essential to accelerate the use of valuable information for the global good.
| S-EPMC4239377 | biostudies-literature
Project description:Recent advances in computer vision have shown promising results in image generation. Diffusion probabilistic models have generated realistic images from textual input, as demonstrated by DALL-E 2, Imagen, and Stable Diffusion. However, their use in medicine, where imaging data typically comprises three-dimensional volumes, has not been systematically evaluated. Synthetic images may play a crucial role in privacy-preserving artificial intelligence and can also be used to augment small datasets. We show that diffusion probabilistic models can synthesize high-quality medical data for magnetic resonance imaging (MRI) and computed tomography (CT). For quantitative evaluation, two radiologists rated the quality of the synthesized images regarding "realistic image appearance", "anatomical correctness", and "consistency between slices". Furthermore, we demonstrate that synthetic images can be used in self-supervised pre-training and improve the performance of breast segmentation models when data is scarce (Dice scores, 0.91 [without synthetic data], 0.95 [with synthetic data]).
| S-EPMC10163245 | biostudies-literature
Project description:DNA sequences that form secondary structures or bind protein complexes are known barriers to replication and potential inducers of genome instability. In order to determine which helicases facilitate DNA replication across these barriers, we analyzed fork progression through them in wild-type and mutant yeast cells, using 2-dimensional gel-electrophoretic analysis of the replication intermediates. We show that the Srs2 protein facilitates replication of hairpin-forming CGG/CCG repeats and prevents chromosome fragility at the repeat, whereas it does not affect replication of G-quadruplex forming sequences or a protein-bound repeat. Srs2 helicase activity is required for hairpin unwinding and fork progression. Also, the PCNA binding domain of Srs2 is required for its in vivo role of replication through hairpins. In contrast, the absence of Sgs1 or Pif1 helicases did not inhibit replication through structural barriers, though Pif1 did facilitate replication of a telomeric protein barrier. Interestingly, replication through a protein barrier but not a DNA structure barrier was modulated by nucleotide pool levels, illuminating a different mechanism by which cells can regulate fork progression through protein-mediated stall sites. Our analyses reveal fundamental differences in the replication of DNA structural versus protein barriers, with Srs2 helicase activity exclusively required for fork progression through hairpin structures.
| S-EPMC3273818 | biostudies-other
Project description:BackgroundUnited States data protection laws vary depending on the data type and its context. Data projects involving social determinants of health often concern different data protection laws, making them difficult to navigate.ObjectiveWe systematically aggregated and assessed useful online resources to help navigate the data-sharing landscape.MethodsWe included publicly available resources that discussed legal data-sharing issues with some health relevance and published between 2010 and 2019. We conducted an iterative search with a common string pattern using a general-purpose search engine that targeted 24 different sectors identified by Data Across Sectors for Health. We scored each online resource for its depth of legal and data-sharing discussions and value for addressing legal barriers.ResultsOut of 3710 total search hits, 2721 unique URLs were reviewed for scope, 322 received full-text review, and 154 were selected for final coding. Legal agreements, consent, and agency guidance were the most widely covered legal topics, with HIPAA (The Health Insurance Portability and Accountability Act), Family Educational Rights and Privacy Act (FERPA), Title 42 of the Code of Federal Regulations Part 2 being the top 3 federal laws discussed. Clinical health care was the most prominent sector with a mention in 73 resources.ConclusionsThis is the first systematic study of publicly available resources on legal data-sharing issues. We found existing gaps where resources covering certain laws or applications may be needed. The volume of resources we found is an indicator that real and perceived legal issues are a substantial barrier to efforts in leveraging data from different sectors to promote health.
| S-EPMC9490527 | biostudies-literature
Project description:High-throughput phenotyping (HTP) platforms are capable of monitoring the phenotypic variation of plants through multiple types of sensors, such as red green and blue (RGB) cameras, hyperspectral sensors, and computed tomography, which can be associated with environmental and genotypic data. Because of the wide range of information provided, HTP datasets represent a valuable asset to characterize crop phenotypes. As HTP becomes widely employed with more tools and data being released, it is important that researchers are aware of these resources and how they can be applied to accelerate crop improvement. Researchers may exploit these datasets either for phenotype comparison or employ them as a benchmark to assess tool performance and to support the development of tools that are better at generalizing between different crops and environments. In this review, we describe the use of image-based HTP for yield prediction, root phenotyping, development of climate-resilient crops, detecting pathogen and pest infestation, and quantitative trait measurement. We emphasize the need for researchers to share phenotypic data, and offer a comprehensive list of available datasets to assist crop breeders and tool developers to leverage these resources in order to accelerate crop breeding.
| S-EPMC8561249 | biostudies-literature
Project description:Modern microscopes produce vast amounts of image data, and computational methods are needed to analyze and interpret these data. Furthermore, a single image analysis project may require tens or hundreds of analysis steps starting from data import and pre-processing to segmentation and statistical analysis; and ending with visualization and reporting. To manage such large-scale image data analysis projects, we present here a modular workflow system called Anima. Anima is designed for comprehensive and efficient image data analysis development, and it contains several features that are crucial in high-throughput image data analysis: programing language independence, batch processing, easily customized data processing, interoperability with other software via application programing interfaces, and advanced multivariate statistical analysis. The utility of Anima is shown with two case studies focusing on testing different algorithms developed in different imaging platforms and an automated prediction of alive/dead C. elegans worms by integrating several analysis environments. Anima is a fully open source and available with documentation at www.anduril.org/anima.
| S-EPMC4115631 | biostudies-literature
Project description:Background Data security has been a critical topic of research and discussion since the onset of data sharing in e-health systems. Although digitalization of data has increased efficiency and speed, it has also made data vulnerable to cyber attacks. Medical records in particular seem to be the regular victims of hackers. Several data breach incidents throughout history have warranted the invention of security measures against these threats. Although various security procedures like firewalls, virtual private networks, encryption, etc are present, a mix of these approaches are required for maximum security in medical image and data sharing. Methods Relatively new, blockchain has become an effective tool for safeguarding sensitive information. However, to ensure overall protection of medical data (images), security measures have to be taken at each step, from the beginning, during and even after transmission of medical images which is ensured by zero trust security model. In this research, a number of studies that deal with these two concepts were studied and a decentralized and trustless framework was proposed by combining these two concepts for secured medical data and image transfer and storage. Results Research output suggested blockchain technology ensures data integrity by maintaining an audit trail of every transaction while zero trust principles make sure the medical data is encrypted and only authenticated users and devices interact with the network. Thus the proposed model solves a lot of vulnerabilities related to data security. Conclusions A system to combat medical/health data vulnerabilities has been proposed. The system makes use of the immutability of blockchain, the additional security of zero trust principles, and the scalability of off chain data storage using Inter Planetary File Systems (IPFS). The adoption of this system suggests to enhance the security of medical or health data transmission.
| S-EPMC7542122 | biostudies-literature