🟠AWS Architecture Blog·January 29, 2026

AWS Architecture for AI-Powered Cancer Diagnostics at Artera

This article details Artera's scalable AWS architecture for their AI-powered prostate cancer diagnostic platform, which processes large biopsy images and delivers personalized treatment recommendations. It highlights the use of AWS services for secure data ingestion, complex AI model orchestration, and regulatory compliance, addressing challenges of high-resolution image processing and sensitive patient data management.

AI & ML Infrastructure Cloud & Infrastructure Distributed Systems

Read original on AWS Architecture Blog

Artera's AI-powered prostate cancer diagnostic platform is a prime example of building a scalable and compliant system for processing sensitive medical data with AI/ML workloads. The core challenge involved managing extremely large biopsy image files (up to 8 GB each), breaking them down into smaller patches for model processing, and serving millions of these patches to foundation models for training and inference, all while adhering to strict healthcare regulations like HIPAA.

Architectural Overview

The system leverages a comprehensive suite of AWS services, emphasizing a modern, scalable design. The architecture is designed to handle high-volume data ingestion, complex workflow orchestration, and globally distributed deployments while ensuring data locality and stringent security.

ℹ️

Key System Design Challenges

<ul><li>Large Data Volume: Handling biopsy images up to 8GB, requiring chunking and high-volume serving to ML models.</li><li>Complex Workflows: Orchestrating multiple AI models and preprocessing steps.</li><li>Regulatory Compliance: Adhering to HIPAA and other data residency requirements for sensitive patient data.</li><li>Performance: Accelerating diagnosis time to results for critical medical decisions.</li></ul>

Data Ingestion and Frontend

AWS Global Accelerator: Improves availability and performance for global users by routing traffic through the AWS network to the nearest region.
Application Load Balancer: Distributes incoming traffic across web portal instances.
Amazon CloudFront: Serves static assets for the portal, providing low-latency content delivery worldwide.
Amazon ECS: Hosts the containerized web portal, providing the user interface for medical professionals to upload images and retrieve results.

AI/ML Processing and Storage

Amazon EKS: Manages Kubernetes clusters for AI/ML inference workloads, enabling the analysis of biopsy images using computer vision models. It also facilitates the training of massive foundation models.
Amazon EFS: Provides shared file storage accessible by both Amazon ECS and Amazon EKS, crucial for storing and processing large biopsy images efficiently during the preprocessing and model execution phases.
Amazon S3: Serves as durable and secure storage for raw biopsy images and analysis results, forming the backbone of their data lake.
Amazon RDS: A managed relational database for patient records, diagnostic results, and application metadata, ensuring high availability and reliability.
Amazon ElastiCache: Utilized for in-memory caching to enhance application performance and reduce latency for frequently accessed data.

Security and Monitoring

AWS IAM: Manages granular access controls and permissions across all AWS resources.
AWS KMS: Handles encryption keys for sensitive patient data, critical for HIPAA compliance.
Amazon CloudWatch: Provides comprehensive monitoring of the entire infrastructure for performance, health, and operational insights.

This architecture demonstrates a robust pattern for building compliant, high-performance AI/ML platforms on the cloud, particularly for scenarios involving large, sensitive datasets and complex computational workflows. The emphasis on containerization (ECS/EKS), managed services (RDS, EFS, ElastiCache), global network optimization (Global Accelerator, CloudFront), and strong security controls (IAM, KMS) highlights best practices in modern cloud-native system design.

AWSAI/MLHealthcareData ProcessingScalabilityComplianceContainerizationMicroservices

Comments

Loading comments...

Architecture Design

Design this yourself

Design a highly scalable, secure, and compliant AI-powered diagnostic platform on AWS for processing large medical images (e.g., biopsy slides) and delivering real-time inferences. The system must support global users, handle multi-gigabyte input files, orchestrate complex machine learning workflows, and ensure HIPAA compliance. Key architectural considerations include efficient data ingestion, distributed model training and inference with containerization, secure data storage, and low-latency access to results.

Other design angles

· Design only the distributed ML inference pipeline for a medical imaging AI, focusing on throughput, latency, and model versioning.· Architect the data ingestion and preprocessing pipeline for multi-gigabyte medical images, emphasizing parallel processing, error handling, and secure storage in a multi-regional setup.· Design a secure, multi-tenant portal for medical professionals to upload and retrieve AI-diagnostic results, including authentication, authorization, and audit trails.