With our novel reImagine technology, we are able to offer Medical Image Analsysis services to all our clinical and industrial partners. Our state-of-the art solution, is composed of four layers: Deploy, Engine, Compute and Interact plus all the Apps which are used for domain specific solutions.
The Deploy layer is responsible for configuring a customized version of the Kubernetes container orchestrator on a cloud service for further processing. It can work on bare metal for an on premises solution or on major cloud providers such as AWS and Google Compute.
The Engine layer configures several key pieces of our infrastructure such as security (RBAC, Secrets and Certificates), storage (NFS and S3-Minio buckets), Databases (Mongo and Orthanc for DICOM data), Lamba functions for serverless operations (Kubeless), Workflow managers (Argo, Kubeflow) and loggers (Prometeo).
The Compute layer is the core of the platform and hosts all the algorithms that are used by the apps and pools the modes used for Deep Learning, Finite Element Simulations and Agent Based Modeling. Every researcher has the opportunity to upload her algorithm in the compute repository and it will become available through all the system.
Finally the Rocket layer is responsible of the interaction between the apps and holds all the APIs that expose the previous functionalities to the rest of the worls. It performs vital roles such as user authentications, quota management, project and data authorization, data consistency and jobs submission. It also includes a the JS codebase with medical image and mesh viewer functionality.
So far we released six apps that are in use by our partners: EView for simulating liver electroporation, US for analysis of cadiac ultrasound imaging, INSPIRE for lung segmentation and analysis of cancer nodules and Cronic Obstructive Pulmonary Disease, FIRST for segmentation of fetal organs and planning of fetal surgeries.
We are in the process of releasing our solution open source: you can find the last code here.