Скачать или смотреть From Scan to Screen: Visualizing Medical DICOM Data with Python

In this technical talk, I share how breaking my knee while bouldering led to an exploration of medical imaging visualization. After suffering a tibial plateau fracture (Schatzker type IV) requiring osteosynthesis with titanium plates and screws, I turned my recovery time into a deep dive into DICOM data processing and 3D visualization techniques.

*Technical Overview:*
This presentation covers the complete pipeline from raw medical imaging data to interactive 3D visualizations. We explore CT scan fundamentals, including Hounsfield units and their density mappings (air at -1000 HU, soft tissue 0-300 HU, bone 300-1000 HU, and titanium at ~3000 HU). The talk demonstrates practical Python implementations for parsing DICOM files, converting to NumPy arrays, and using NRRD datasets for volumetric rendering.

*Key Topics Covered:*
• DICOM file format structure and metadata extraction
• CT vs MRI imaging principles and data characteristics
• Volumetric rendering techniques: ISO surface extraction, Maximum Intensity Projection (MIP), and ray tracing
• WebGL shader implementation for real-time 3D visualization
• Transfer functions and opacity mapping for medical data
• Handling imaging artifacts from metal implants

*Technical Stack:*
Python (NumPy, pydicom), JavaScript (Three.js), WebGL shaders, NRRD format

*Chapters:*
00:00 Introduction
00:20 Bouldering and overhangs
02:17 What now?
04:22 Medical Imaging
04:49 CT imaging
10:58 NRRD format
17:00 Volumetric rendering
29:33 MRI imaging
31:47 Q&A
48:02 Conclusion
48:13 End

Code available at: [repository link]