e-BrainScience

Background and challenges

With increased life expectancy in modern society, the number of individuals who will potentially become demented is growing proportionally. Current estimates count world-wide over 48 million people suffering from dementia bringing the social cost of care to 1% of world’s gross domestic product – GDP. These numbers led the World Health Organisation to classify neurocognitive disorders as a global public health priority. Nowadays neuroimaging tools are essential in the diagnosis and treatment of individuals suffering from brain injuries and disorders. Compared to visual assessment, automated diagnostic methods based on brain imaging are more reproducible and have demonstrated a high accuracy in separating AD from healthy aging, but also the clinically more challenging separations between different types of neurocognitive disorders. Similarly, although ApoE genotypes carrying higher risk for AD are easily obtainable, this information is rarely integrated in machine learning-based diagnostics for AD. Although encouraging, implementations into clinical routine have been challenging.

Scenarios

The goal is to makes data on populations of patients broadly available for research use, by providing software-as-service to clinicians, neuroscientists, epidemiologists and pharma both for diagnosis and research in clinics and for collaborative neuroscience research using clinical data.
The first scenario foreseen consists in providing standardized workflow for the pre-processing of neuroimaging data, the users will be able to select and configure neuroimaging workflow from data conversion, to images segmentation and brain features extraction.The second scenario foreseen consists in providing an innovative data analysis system that wide users (clinicians, neuroscientists, epidemiologists) can access and use to analyse clinical and research data without moving the data from the hospital or private cloud servers where they reside and without infringing on patient privacy. The strategy is to use cloud computing and machine learning approaches and create a meeting place for neuroscience and IT for collaborative brain disease research as well as benefitting clinicians on a daily basis.

MORPHEMIC added-value

The primary added-value would be to boost the neuroimaging field by using cloud computing. To achieve this goal, the algorithms have to be benchmarked and validated against established results and codes, understanding the particularities of the clinical data and tuning the algorithms for optimal results. Second, they will be used to perform as mentioned before more advanced and dedicated studies of a wide variety of clinical data. A robust set of ML algorithms capable of performing similar tasks as codes like SPM would be greatly used and appreciated by the neuroimaging community. Most users are running current solution on their laptop or desktop and Those with large patient cohorts are on HPC. However, the increased complexity and data available will make necessary the use of GPUs to achieve the desired results in a reasonable amount of time. MORPHEMIC and Proactive open new solutions and provides cloud-ready federated components and software-as-a-service deployed in community (public cloud) and local private (private cloud) execution environment with:

Both scenarios:

  • Centralised access to software and data deployed in private execution environments
  • Private environments were algorithms from data pre-processing, brain feature extraction to data mining are
    executed locally, and where the de-identified data are stored
  • Community execution environment to orchestrate the execution of statistical and machine-learning
    algorithms for advanced multi-datasets, cross-centre descriptive, and predictive analytics and federates the
    results
  • Master orchestrator components that are running in community execution environment, connected to the
    distributed private execution environments via web services, fetch the aggregate results of the algorithms
    executed in the private execution environments and aggregate them in a cross-centre data analysis result.

KPIs

  • Users: number of clinical researchers using the workflow- Target 100
  • Data: amount of data and patients records- Target 10 000 records
  • Scientific workflow: number of pre-processing and data analyses workflow integrated: image conversion, image
    normalization, brain maps creation, brain atlas creation, neuromorphometrics computation