Get a Second Opinion on an MRI Report Using Claude Code

What this is, in plain English

This workflow describes using Claude Code, an advanced AI tool from Anthropic, to get a second opinion on an MRI scan and its accompanying report. Unlike simpler chat AIs, Claude Code can run code and process complex data files, making it suitable for tasks like analyzing medical images.

The goal is to have the AI independently review your MRI results and compare them against a human doctor’s diagnosis. This can help you understand your condition better and identify any potential differences in interpretation.

This is an advanced workflow because it involves handling specialized medical files (DICOM format), using a developer-focused AI environment, and potentially writing or understanding code instructions. It’s not a simple copy-paste process for a beginner, as the exact steps for uploading and processing complex data can vary and require technical comfort.

What you can use it for

• Get an independent review of medical reports: Use AI to cross-check a doctor’s diagnosis against the raw MRI data and written reports.
• Identify potential discrepancies: The AI can highlight differences between human and AI interpretations, prompting you to ask more questions or seek further medical advice.
• Understand complex medical terminology: Ask the AI to explain medical terms from your MRI report in simple, easy-to-understand language.
• Research treatment options: Use the AI to find information on clinical guidelines or common treatments related to your diagnosis.

Tools you need

• Claude Code (paid): An advanced AI environment by Anthropic that can run code and process complex files, used here for MRI analysis.
• ChatGPT (freemium): A general-purpose AI chatbot by OpenAI, useful for initial checks and discussions about medical information.

How it actually works

1. Prepare your MRI data: The author used a standard DICOM export, which included hundreds of files totaling around 266 MB. You would typically get these files from your clinic, often on a CD/DVD or through a secure online portal. For easier upload to an AI tool, you would likely need to compress these files into a single archive, such as a .zip file.
2. Access Claude Code: Log in to your Claude Code environment. The author used a specific version called “Opus 4.8 (xhigh)” which is a powerful model within Claude’s developer tools.
3. Instruct the AI to prepare: Before uploading the data, you would tell Claude Code to install any necessary software packages it might need for medical image analysis. The author does not specify which packages were used, so this step is generic.
4. Upload and prompt for initial analysis: Upload your compressed MRI data to Claude Code. Then, give the AI a basic prompt describing your situation, such as “right shoulder pain for 2–3 weeks,” and ask it to analyze the MRI and generate its own report.
5. Compare and arbitrate: If the AI’s report differs from the human doctor’s diagnosis, provide both reports to Claude Code. You can also add any additional context, like discussions you had with other AIs (e.g., ChatGPT) or doctors. Then, ask Claude Code to compare the findings and help you understand the differences. The author mentions using “multiple subagents” for this, which suggests a sophisticated, multi-step process or a series of detailed prompts to guide the AI’s comparison.

Words you’ll see, explained

• MRI (Magnetic Resonance Imaging): A type of medical scan that uses strong magnets and radio waves to create detailed pictures of organs and structures inside your body.
• DICOM (Digital Imaging and Communications in Medicine): A standard file format used for storing and exchanging medical images and related patient information.
• Subscapularis tendon: One of the four muscles that make up the rotator cuff in your shoulder, which helps with arm movement and rotation.
• Partial-thickness tear: A type of injury where a tendon is torn, but not completely separated or ruptured.
• Shockwave therapy: A non-invasive medical treatment that uses acoustic waves to stimulate healing in injured tissues.
• Homeopathic medicine: An alternative medicine system based on the idea that highly diluted substances can trigger the body’s healing process.

Original source

This workflow was shared by engmarketer on Hackernews, detailing their personal experience using Claude Code Opus to analyze an MRI report and compare it with a human diagnosis. The full article provides more context on their journey.

Notes & variations

• Do you even need this?: For simpler medical questions or to understand a text-based report, a freemium chat AI like Claude.ai or ChatGPT might be sufficient. This advanced workflow is specifically for analyzing raw medical image data, which is a much more complex task.
• Free-tier limits: Claude Code is a paid service, and processing large files like a 266 MB MRI package will consume significant credits, which can be costly.
• Common pitfall: AI is not a substitute for professional medical advice. Never rely solely on AI for diagnosis or treatment decisions. Always consult a qualified medical professional for any health concerns. The author’s experience highlights that AI results can sometimes differ significantly from human doctors.