Week 3: CT Image Processing

Question 1

Prospective Reconstruction

Answer 1

Raw data automatically rendered as image data

Question 2

Retrospective Data

Answer 2

Raw data revisited and manipulated to create new image data

Question 3

Reformatting

Answer 3

Image data used to create new images.

Question 4

Filtered Back Projection

Answer 4

Projections are compiled back onto each other; data is all combined mathematically using Fourier Transformations. The more projections, the more complete picture of the object.

Question 5

Fourier Transformation

Answer 5

Takes waves spectates their function into frequency components. Ex: a music chord into individual notes.

Question 6

Iterative Reconstructions

Answer 6

Uses algebra to predict/determine the value based on other data in the set. Takes more computer power than back projection but needs less dose (50-70% less). Suppresses noise which increases contrast resolution and reduces artifacts. ASIR (adaptive statistical iterative reconstructions) are gaining popularity because they can reconstruct missing image data by comparing to an assumed image.

Question 7

Convolutions

Answer 7

Applied to the raw scan data, they affect the appearance of the displayed image data. Algorithms, kernels and filters. The choice of recon will be set in the protocol for that body part which affects how the raw data is processed. If you wish to change the algorithm you must reprocess the raw data = retrospective reconstruction.

Question 8

Manipulating DFOV

Answer 8

Thoracic spine can be pulled out of the whole body trauma scan by changing the DFOV and the start and end of the scan. This saves the patient for being re scanned.

Question 9

Reconstruction vs Reformat

Answer 9

Reconstruction: from raw data, axial orientation, any algorithm.

Reformat: from image data, any orientation, same algorithm as the image data.

Question 10

Multi Planar Reformats (MPR)

Answer 10

Best with thin slices, DFOV must be the same over the whole data set. Image centre must be the same, small amounts of motion are a problem.

Question 11

3D Reformats

Answer 11

Surface Rendering (SR), Shaded Surface Display (SSD), Volume Rendering, MinIPs and MaxIPs.

Manipulate and combine CT number values depending on the algorithm. Software bases the rendering on CT numbers, data will be included in the display or excluded based on the CT numbers of the voxel.

Question 12

Surface Rendering (SR) and Shaded Surface Display (SSD)

Answer 12

Shows tissue densities at the surface of the stutters within the volume of data. Ex virtual colonoscopy.

Question 13

Volume Rendering (VR)

Answer 13

Uses the entire volume of data to create 3D image. Semi-transparent representation of structure, all voxel data used so it shows multiple tissue types.

Question 14

Segmentation of Data

Answer 14

The ability to remove unwanted structures by removing specific parts of a 3D image. Manual: with a mouse and drawing around the unwanted area. Automatic: removes all CT numbers near bone to visualize vessels (ex).

Question 15

Maximum Intensity Projection (MIP)

Answer 15

Display only the highest attenuation values from the data. Best used when the objects you want to see are the brightest/most dense (contrast filled vessels, bones). Use very tick sections (10-20mm).

Question 16

Minimum Intensity Projection (MinIP)

Answer 16

Displays only the lowest attenuation values from the data. Not as common but can be used to see airways in the lungs better (least dense structures). Very thick sections (10-20mm). Work by not using volume averaging