Clinical Recap

Congrats, this was a lot of material that hopefully put you on a good track to comprehend the context of the problems you would be working with! Now you are ready to start digging deeper into data and machine learning techniques.

In this lesson we have covered the following topics:

• What are 3D medical images?
• Who uses 3D medical images?
• Why are they being used?
• Some example clinical scenarios

After that, we did an exercise on picking a suitable problem for an AI project by tapping into publicly available medical resources.

Then, we jumped into some technical details and covered:

• Physical principles of CT scanners along with an exercise on computing a sinogram
• Physical principles of MR scanners
• Covered basic 3D imaging tasks:
  • Multi-planar reconstruction
  • 3D reconstruction
  • Windowing
  • Registration

Vocabulary

Let us leave you with a little vocabulary of the many terms that have been introduced throughout this lesson:

• Imaging modality: a device used to acquire a medical image
• Contrast resolution: the ability of an imaging modality to distinguish between differences in image intensity
• Spatial resolution: the ability of an imaging modality to differentiate between smaller objects
• CT scanner: computed tomography scanner
• Sinogram: “raw” data generated by CT scanner. Images need to be reconstructed from it
• MRI scanner: Magnetic Resonance Imaging scanner
• K-space data: “raw” data generated by an MRI scanner. Images need to be reconstructed from it
• Windowing: mapping high dynamic range of medical images onto the screen-space gray color scale
• MPR: multi-planar reconstruction - extraction of non-primary imaging planes from a 3D volume
• 3D reconstruction: constructing a 3D model from multiple slices of 3D medical imaging data
• Registration: bringing two different images into same patient-centric coordinate space