Tech

I sit somewhere between a clinician who codes and a sysadmin who reads brain MRIs. These are notes on the tools and systems I use in clinical work, research, and infrastructure.

AI in Clinical Radiology

At Turku University Hospital, we have been integrating commercial AI algorithms into emergency radiology workflows since 2022. Two areas where this has had the most impact:

Fracture detection. AI-assisted musculoskeletal trauma detection runs on emergency radiographs, flagging suspected fractures for the on-call radiologist. We published a comparative study of two commercial algorithms in 2025, finding that both improve sensitivity but differ in false positive rates. The real value is not replacing the radiologist but catching the subtle findings at 3 AM.

Automated MRI protocoling. Emergency brain MRI requests come with free-text clinical referrals. We built and validated machine learning and deep learning models that read the referral text and suggest the appropriate MRI protocol. This saves time for the neuroradiologist and reduces protocol errors, especially outside office hours. Published in Radiology: AI in 2025.

The harder problem is not building the models. It is integrating them into the PACS and RIS in a way that radiologists actually trust and use. Hospital IT moves slowly, and for good reason.

Self-hosted Infrastructure

I run my own servers for personal projects, research tooling, and learning. The setup:

• Daily driver: Framework Laptop 13 (Intel Core Ultra 7, 96 GB RAM) running Arch Linux with Hyprland (Wayland). I use Omarchy, a curated Arch distribution by DHH.
• Home server: KVM host running several VMs, including the one serving this website. Caddy for web, Soft Serve for private Git repos.
• Hospital workstation: i9-14900K with an RTX 5090, running Podman containers and KVM VMs for research. Used for local LLM inference with Ollama, DICOM processing, and AI model development.

Why self-host? Partly for privacy and control, partly because managing Linux servers is how I learned most of what I know about networking, security, and systems. When something breaks at 2 AM, you learn fast.

Tools

• Editor: Neovim
• Shell: Bash + Starship prompt
• Terminal: Ghostty
• AI assistant: Claude Code for sysadmin tasks, coding, and research
• Notes: Obsidian (GTD workflow)
• Presentations: Marp (Markdown to slides), exported to PowerPoint or PDF
• Medical imaging: pydicom, pynetdicom for DICOM scripting
• Version control: Git + GitHub CLI, self-hosted Soft Serve for private repos
• Database: PostgreSQL (production), SQLite (prototyping)

DICOM and Hospital IT

Medical imaging runs on DICOM, a protocol from the 1990s that somehow still works. Most of my technical work at the hospital involves connecting things to the PACS: routing studies to AI analysis VMs, building audit logging for image access, and automating data flows between systems that were never designed to talk to each other.

The challenge is always the same: hospital networks are locked down for good reason, vendor systems are black boxes, and every integration is a negotiation between security, usability, and the reality of a 24/7 clinical environment.