CIBIS

CIBIS
Name	CIBIS
Type	Biomedical diagnostic system
Developed by	National Institutes of Health; Massachusetts Institute of Technology; Stanford University
Initial release	2018
Latest release	2023
Operating system	Cross-platform
License	Proprietary / Academic

CIBIS is an advanced clinical imaging and biomarker interpretation system developed as a translational research platform combining multimodal imaging, machine learning, and database integration. It integrates datasets and analytic modules from institutions such as Johns Hopkins University, Mayo Clinic, Harvard Medical School, University of California, San Francisco and collaborates with industry partners including Siemens Healthineers, GE Healthcare, and Philips. CIBIS aims to support diagnostic workflows in specialties like radiology, oncology, neurology, and cardiology by fusing imaging, molecular assays, and electronic health record inputs.

Overview

CIBIS is designed as a modular pipeline that ingests magnetic resonance imaging, computed tomography, positron emission tomography, histopathology whole-slide images, and genomic panels to produce standardized reports and decision support. The platform interoperates with systems from Epic Systems Corporation, Cerner Corporation, and Allscripts and uses ontologies from SNOMED International and LOINC to harmonize data. Core components include image preprocessing engines influenced by architectures from UCLA, feature extraction influenced by research at Carnegie Mellon University, and model repositories inspired by initiatives at OpenAI and DeepMind.

History

CIBIS originated from a joint grant funded by the National Institutes of Health and the Defense Advanced Research Projects Agency to bridge computational imaging and clinical translation. Early pilot studies were conducted at Massachusetts General Hospital and Brigham and Women's Hospital with contributions from the Broad Institute and the Whitehead Institute. Subsequent phases involved multicenter trials coordinated through the European Society for Radiology and the American College of Radiology. Major milestones include the 2019 prototype validation presented at the Radiological Society of North America annual meeting and the 2021 multicenter evaluation with datasets from Memorial Sloan Kettering Cancer Center and UCLA Medical Center.

Architecture and Design

The architecture couples a distributed compute fabric using platforms such as Kubernetes clusters provisioned on Amazon Web Services, Google Cloud Platform, and Microsoft Azure with containerized microservices. Data lakes employ storage models compatible with OMOP Common Data Model and imaging formats based on DICOM standards. Model components include convolutional neural networks derived from innovations at Google Brain and recurrent modules reflecting work from University of Toronto labs. Visualization layers integrate toolkits from MIT Media Lab and interactive viewers like those developed at Oxford University. Security and identity management align with frameworks from NIST and Health Level Seven International.

Clinical Applications

CIBIS has been applied to tasks such as automated tumor segmentation in glioblastoma cohorts, lesion detection in lung cancer screening programs, ischemic core estimation in acute ischemic stroke triage, and plaque characterization in coronary artery disease studies. It supports biomarker discovery for precision oncology programs at Dana-Farber Cancer Institute and pharmacodynamic monitoring in trials run by Pfizer and Roche. Use cases presented at conferences hosted by American Society of Clinical Oncology and European Association for the Study of the Liver demonstrated performance on curated datasets from The Cancer Genome Atlas and population registries like SEER Program.

Performance and Evaluation

Performance assessments used statistical frameworks advocated by CONSORT and reporting guidelines from STARD and TRIPOD. Evaluation datasets included external validation cohorts from Johns Hopkins Hospital and blinded reads benchmarked against consensus expert panels drawn from American Board of Radiology–certified specialists. Metrics reported included area under the receiver operating characteristic curve, sensitivity, specificity, and calibration curves compared with established scoring systems such as RANO for neuro-oncology and Lung-RADS for screening. Peer-reviewed results appeared in journals including The Lancet Oncology, Radiology, and Nature Medicine.

Deployment and Implementation

Clinical deployments followed phased rollouts with pilot programs at Cleveland Clinic and community health networks coordinated by Kaiser Permanente. Integration required interfacing with picture archiving and communication systems from Agfa HealthCare and validation per accreditation standards from Joint Commission. Training programs were developed with continuing medical education partners including American Medical Association and vendor workshops with Siemens Healthineers for technologists. Deployment scenarios ranged from on-premises hybrid models in tertiary centers to cloud-native instances for research consortia like Global Alliance for Genomics and Health.

Ethical, Regulatory, and Safety Considerations

CIBIS development engaged institutional review boards at centers including Columbia University Irving Medical Center and adherence to regulatory pathways overseen by U.S. Food and Drug Administration and European Medicines Agency for software as a medical device. Ethical review emphasized issues raised by scholars at Harvard Law School and Oxford Internet Institute concerning algorithmic bias and data governance. Safety frameworks aligned with guidance from World Health Organization and incident reporting protocols adopted from FDA MAUDE databases. Data privacy efforts referenced standards under Health Insurance Portability and Accountability Act and General Data Protection Regulation compliance programs.

Category:Medical software