KAYRA: A Microservice Architecture for AI-Assisted Karyotyping with Cloud and On-Premise Deployment

2026-04-29 · Source: Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computer Vision & Pattern Recognition, Cloud Computing & IT Infrastructure · Depth: Advanced, quick

Summary

KAYRA is an end-to-end karyotyping system designed as a containerized microservice pipeline for clinical cytogenetic laboratories. Its machine learning stack integrates an EfficientNet-B5 + U-Net semantic segmenter, a Mask R-CNN (ResNet-50 + FPN) instance detector, and a ResNet-18 classifier, employing a cascaded ROI-narrowing strategy. The system supports both cloud and on-premise deployments using the same container images, accommodating diverse clinical data egress policies. A pilot clinical evaluation on 459 chromosomes from 10 metaphase spreads demonstrated a segmentation accuracy of 98.91%, classification accuracy of 89.1%, and rotation accuracy of 89.76%. KAYRA significantly outperforms older density-thresholding systems and shows superior segmentation compared to modern AI-supported references, reaching TRL 6 maturity and integrating human-in-the-loop expert review.

Key takeaway

For AI Scientists developing diagnostic tools, KAYRA demonstrates that a multi-model cytogenetic AI service can be effectively packaged as a microservice architecture. This approach supports flexible cloud or on-premise deployment, crucial for clinical environments with strict data governance, while delivering strong empirical performance and integrating necessary human-in-the-loop expert review workflows. Consider this architecture for your next clinical AI system.

Key insights

KAYRA offers a microservice AI architecture for karyotyping with flexible deployment and strong clinical performance.

Principles

• Microservices enable flexible deployment.
• Cascaded ROI-narrowing improves model focus.
• Human-in-the-loop is crucial for diagnostics.

Method

KAYRA orchestrates an EfficientNet-B5 + U-Net segmenter, Mask R-CNN detector, and ResNet-18 classifier via a cascaded ROI-narrowing strategy, packaged as containerized microservices for cloud or on-premise deployment.

In practice

• Deploy AI as containerized microservices.
• Combine multiple models for complex tasks.
• Integrate human review into AI workflows.

Topics

• KAYRA System
• AI-Assisted Karyotyping
• Microservice Architecture
• Clinical Cytogenetics
• Machine Learning Models

Best for: Computer Vision Engineer, AI Scientist, AI Engineer, MLOps Engineer, Research Scientist

Related on AIssential

• Aether: Network Validation Using Agentic AI and Digital Twin — Aether automates network change validation using agentic AI and a multi-functional digital twin.
• Microsoft accelerates telecom return on intelligence with a unified, trusted AI platform — Telecoms achieve significant ROI from AI by integrating a unified, trusted platform across operations, networks, and customer experiences.
• Operationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production — Microservice architecture for Document AI separates GPU inference from CPU orchestration, revealing OCR as the primary latency bottleneck.
• Unlocking the Future of Gene Therapy — AI is revolutionizing gene therapy by enabling precise cellular targeting and accelerating discovery-to-clinic pathways.
• AWS and Cerebras partner to advance AI inference performance in the cloud — Inference disaggregation optimizes generative AI by dedicating specialized hardware to prefill and decode stages.
• How a Nonprofit Transforms Data with Cloudera and AI — Nonprofits can leverage hybrid data and AI platforms to conduct complex research affordably.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.