Introduction to Medical Images

The choice of medical image and imaging modality will inform how it is processed. Before moving onto segmentation it is important to consider what determines the intensity of the image, as this will be the most important factor in what can be segmented from the image. Other parameters such as the resolution, contrast, and noise inherent to the image will affect how easy it is to extract target structures accurately. Here is a short list of some common 3D medical imaging technologies and their key attributes.

Modality	"Type"	Intensity from:	Resolution	Notes
X-Ray CT	Structural	​Photoelectric coefficient ≈ Density	​0.5 - 1.5 mm	​Due to the high contrast of hard tissues in these images, they are ideal for creating bone models
MRI	Structural/Functional	Water content & proton relaxation time	0.3 - 1.0 mm	Excellent soft-tissue contrast, so very useful for distinguishing neighbouring soft tissue structures
SPECT/PET	Structural/Functional	Particle emission from ingested/implanted radioisotope	3.0 - 8.0 mm​	Particularly useful for investigating brain activity and structures involved in a particular function.
Ultrasound	Structural	Reflections from changes in acoustic impedanc	0.3 - 0.5 mm	Cheap, portable, non-ionising BUT limited imaging depth, relatively poor tissue contrast and lots of noise. Difficult to segment.
Photoacoustics	Structural	Chromophores that absorb incident pulsed laser light and emit ultrasound	~ 0.1 mm (depth dependent)	Very high resolution possible,  but limited imaging depth, requires chromophore of interest and not yet in clinical use

​File Types:
DICOM – Universal image format and file sharing protocol, suitable for multiple image modalities and very widely used. Easy to import into most software.
NIFTI – Format designed specifically to store neuroimaging data. Compatible/viewable with several specific software packages
MINC – Image format used with certain brain imaging software

Next Stage...

SEGMentation