Visualize 3D Medical Data in the Browser using Python

One of our latest experiments at PYCAD was to research how we can visualize 3D data on the browser by directly using Python.
​
This can help us build web apps for POCs in medical imaging.
​
One great tool that we recently discovered is called stpvista.
​
This is in fact a tool that allows you to integrate Pyvista plots, which is a great Python library for 3D visualization, directly inside your Streamlit web application.
​
Below you can see an example of visualizing an STL file of an intraoral scan on the browser using Python and Streamlit!
​
Pyvista is already a huge library for 3D visualization, so having the possibility to integrate its plots inside a Streamlit web app is just the cherry on top of the cake!

​

Medical Imaging Data Formats

I work daily on medical imaging problems. Here are 3 file formats that we deal with constantly.

Let’s demystify these 3 common formats: DICOM, NIFTI, and NRRD.

DICOM (Digital Imaging and Communications in Medicine)

• Industry Standard: Used primarily in radiology and often by imaging devices directly.
• Rich Metadata: Contains patient info, imaging parameters, and acquisition details.
• Structure: Packs multiple 2D slices to form a 3D image.
• Usage: Ideal for clinical applications and PACS integration.

NIFTI (Neuroimaging Informatics Technology Initiative)

• Brain Imaging: Primarily designed for the neuroimaging community.
• Structure: Single file storage of 3D or 4D data (includes time-series data).
• Header Info: Contains spatial and temporal metadata.
• Usage: Research-focused, especially functional MRI studies.

NRRD (Nearly Raw Raster Data)

• Flexibility: Supports a broad range of medical image data types.
• Header-Body Separation: Can keep metadata separate from image data.
• Extensibility: Open format with easy-to-adapt properties.
• Usage: Advanced visualization applications and research purposes.

In summary, the choice of format often depends on the application. DICOM is great for clinical setups, NIFTI shines in research, especially neuroimaging, and NRRD offers flexibility for diverse visualization needs.

Stay informed, and choose wisely based on your specific requirements!

Tweet of the Day

​

Pedro P. Sanchez @SnchzPedro_ 📢 New dataset on @huggingface! 🔬 Synthetic mammography (medical imaging) dataset. These patients do not exist. 🎯 100k synthetic mammograms. 512x512 size. 3 labels: - Low masking level (score ≤ 2) - Medium masking level (2 < score ≤ 6) - High masking level (score > 6) 🩺… https://twitter.com/i/web/status/1716208249815323043 11:41 PM • Oct 22, 2023 58 Retweets 349 Likes Read 6 replies

​