MIDAS

Overview

In collaboration with ICMR, IISc, and ARTPARK, Triveous designed and developed MIDAS (Medical Imaging Datasets for India)—a cloud-native, AI-ready medical imaging platform built as a Digital Public Good. MIDAS addresses the urgent need for standardized and representative medical imaging datasets to enable meaningful AI innovation in India’s healthcare ecosystem.

The platform was showcased at the Indo-France AI Summit 2025, affirming its international relevance and excellence.

The Problem

India’s healthcare system faces systemic challenges:

• Inconsistent, fragmented, and non-standardized imaging datasets
• Limited access for AI researchers and institutions to ethically governed, diverse datasets
• Lack of scalable infrastructure for medical image storage, curation, and AI training
• Insufficient tools and workflows for collaborative data collection and annotation

These barriers restrict the use of AI in diagnosis, screening, and treatment—especially in underserved and remote regions.

Our Solution

Triveous engineered MIDAS as a robust, scalable, and modular platform to address these challenges and advance healthcare AI in India.

Key Solution Pillars:

• Secure and scalable cloud infrastructure to ingest, store, and manage large volumes of imaging data
• Intuitive web interfaces for data upload, annotation, and curation across institutions
• Modular deployments enabling each hospital or research institution to run its own MIDAS instance with full data control
• Standards-compliant architecture, aligned with national digital health infrastructure
• Built in collaboration with clinicians, data scientists, and technologists to ensure real-world relevance

Building a Pioneering AI Digital Public Good

MIDAS is now recognized as a pioneering AI-driven digital public good, democratizing access to high-quality medical imaging datasets.

It enables AI researchers to:

• Develop and validate diagnostic models
• Conduct early disease detection studies
• Explore treatment planning and population health analytics

By addressing fundamental data and infrastructure gaps, MIDAS fosters significant advancements in medical AI research, especially for resource-constrained and diverse populations.

Key Features

• Gold-Standard Datasets: Expert-curated and validated datasets with rigorous quality control, designed for high-impact healthcare use cases such as oncology and neurology.
• Streamlined Annotation Workflows: AI-assisted tools and structured labeling processes reduce annotation time by up to 40%, improving efficiency and consistency.
• Hub-and-Spoke Deployment Model: Implemented across a distributed network of clinical and research institutions, enabling decentralized data ownership and region-wise scalability.
• Modular System Architecture: Each participating hub operates an independent, secure MIDAS instance—ensuring full data control while supporting interoperability across the network.
• Privacy-First & Compliant: Built with strong anonymization, audit logging, and adherence to national ethical and data privacy standards to ensure responsible data use.
• Built for India’s Digital Public Infrastructure (DPI): Designed as an open, interoperable Digital Public Good aligned with national health data standards and prepared for federated AI research.

Real-World Deployment

MIDAS is deployed across multiple medical hubs in India through a hub-and-spoke model. Each participating institution runs a standalone MIDAS instance, maintaining full control over its data while contributing to a national research effort.

Current deployments include:

• AIIMS Delhi (Oral Cancer and Neurosurgery Departments)
• Sharad Pawar Dental College, Maharashtra
• Sri Ramachandra Institute, Tamil Nadu
• KLE Dental College, Karnataka

New deployments are underway to expand the platform’s reach and dataset diversity.

Measurable Impact

• 10,000+ medical images curated for AI training and validation
• 3 AI diagnostic models developed using MIDAS datasets
• 20+ institutions onboarded, including  PGIMER, NIMHANS, SGPGI, and more.
• 100 GB monthly data ingestion, with 99.9% platform uptime
• 40% increase in annotation efficiency via intelligent workflow design

Global Recognition

MIDAS was showcased at the Indo-France AI Summit 2025 in Paris, recognized for setting new benchmarks in healthcare data infrastructure. It is now seen as a blueprint for ethically governed, open-source platforms that accelerate medical AI globally.

The Triveous Contribution

• Co-created the product with clinical and data science stakeholders
• Built a cloud-native, secure, and modular architecture for scale
• Enabled real-time feedback, iteration, and rapid prototyping
• Designed intuitive user interfaces for data curation and access

Conclusion

MIDAS is more than a product—it's a national and global enabler.
It demonstrates how AI-driven digital public goods, when thoughtfully designed, can transform access, equity, and innovation in healthcare.