Skip to content

Medical Image Formats#

Introduction#

A medical image is a digital representation of the internal structure or function of an anatomic region, typically presented as an array of picture elements called pixels (2D) or voxels (volume pixels, 3D). This representation is a discrete mapping of numerical values to positions in space.

Three panels explaining what pixels are. The first shows a low resolution black and white photo of Abraham Lincoln, the second has numerical pixel values overlaid as a matrix, and the third is just the pixel value matrix.

Image of Abraham Lincoln as a matrix of pixel values. (Source)

Subsection of an abodminal CT scan with a red square over an edge of an organ expanded out to a matrix to show the numerical values in that part of the image.

Section of abdominal computed tomography (CT) scan with a matrix of Hounsfield units (HU). (Source: Caryn Geady)

The numerical values, known as intensity values, vary based on:

  • Image type: MRI, CT, PET, etc.
  • Acquisition method
  • Reconstruction and post-processing

Medical Image Metadata#

Medical images often come with metadata, which provides additional information about the image. This metadata is usually stored at the beginning of the image file as a "header."

Common Metadata Fields#

  • Image dimensions: Width, height, depth
  • Voxel size: Spacing between voxels
  • Origin: Location of the first voxel
  • Orientation: Direction of x, y, and z axes
  • Pixel depth: Bytes used to represent each voxel intensity
  • Data type: Integer, floating-point, etc.
(0018,0015) CS BodyPartExamined = LUNG
(0018,0050) DS SliceThickness = 3
(0020,0037) DS ImageOrientationPatient = 1\0\0\0\1\0
(0020,1002) IS ImagesInAcquisition = 99
(0028,0010) US Rows = 512
(0028,0011) US Columns = 512
(0028,0030) DS PixelSpacing = 0.9766\0.9766
(0028,0100) US BitsAllocated = 16
(0028,0101) US BitsStored = 16
(0028,0102) US HighBit = 15
(0028,0103) US PixelRepresentation = 0

Pixel Data#

The pixel data in a medical image file represents the actual image values, stored in a format specific to the image file.

  • In fixed-size header formats, pixel data follows the header directly.
  • In other formats, a marker or tag indicates the start of pixel data.

Pixel Data Size#

\[ Pixel Data Size = \text{Rows} \times \text{Columns} \times \text{Pixel Depth (Bytes)} \times \text{Number of Frames} \]

Image File Size#

\[ Image File Size = \text{Header Size} + \text{Pixel Data Size} \]
Example Calculation:

For a DICOM image with the following parameters:

Parameter Value
Rows 512
Columns 512
Pixel Depth 2 bytes (16-bit image)
Number of Frames 32 (32 slices)
\[ \text{Pixel Data Size} = 512 \times 512 \times 2 \times 32 = 16,777,216 \text{ bytes (or 16 MB)} \]

Assuming the header size for this DICOM file is 1,024 bytes:

\[ \text{Image File Size} = 1,024 \text{ bytes} + 16,777,216 \text{ bytes} = 16,778,240 \text{ bytes (or 16.01 MB)} \]

Medical Image File Formats#

Categories of Medical Image Formats#

  1. Standardization Formats: Standardize images from diagnostic modalities.

    • Example: DICOM

  2. Post-Processing Formats: Facilitate and strengthen post-processing analysis.

    • Examples: Analyze, NIfTI, MINC

Configurations for Storing Medical Images#

  • Single File: Contains both metadata and image data, with metadata stored at the beginning.

    • Examples: DICOM, MINC, NIfTI

  • Two Files: Metadata and image data stored separately.

    • Example: Analyze (.hdr and .img)