Computer imaging techniques

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.


In the last years the use of computer imaging techniques in the field of medicine is growing. In hospitals, the use of such techniques, take place already in a large proportion of the medical images in order to have the images directly in digital form. Some of these techniques are: Magnetic Resonance Imaging, Computed Tomography, mammography, ultrasound. A great amount of digital images are produced annually in this way. If the images are in digital form there are many benefits. They can be really flexible because they can be processed easily, transferred through the internet, they can be saved in databases, making medical records of the patients and can be really helpful in the diagnosis of the disease and therefore to find the appropriate treatment.

The compression of medical images has an important place in the field of telemedicine. It is a significant research objective because many improvements must be done in order to transfer medical images among data networks. In that way for one case it is possible to have diagnosis from several hospitals and in less time. That is a big step for medicine.

For the better results in the process of medical images and easier transfer and storage, the compression of the medical images must achieve reduction of the data to the degree that the correct diagnosis is feasible. On the one hand a great level of compression of the medical images is desirable, so they can transfer fast, but on the other hand some specific points of the images must remain the same so as to have the right diagnosis of the diseases. This is made with lossless compression methods. Lossy compression methods achieve better results on the transmission of medical images and storage but there is no guarantee of the right diagnosis because of the changes in the images.

The compression of medical images is constituted by three basic steps: image transformation, quantization, entropy coding. The first is used for the reduction of the breadth of the signal, for the limitation of the information that is not essential and it gives the appropriate presentation of the medical image in order to have good entropy coding. The transformation must have three standards and all the three standards are about the transform coefficients. They must not depend on other coefficients, the amount of coefficients of the energy of the image must be limited to the smallest and finally their frequency must be kept in the smallest amounts.

2.1 Lossless compression

There are several kinds of methods that are used in order to compress medical images and they are based on predictive model or multi resolution model or a comparison of them, in order to decrease the plentitude. Afterwards they choose a particular code to encode the images. Huffman is a very common choice for coding medical images.

Some lossless methods are:

  1. Hierarchical Interpolation, it takes the initial image in a low resolution form and it adds noise in steps as it increases the resolution. This method does not offer a progress in development because of the noise that the images contain.
  2. Difference Pyramid, is also a kind of method that uses the resolution of the image. It builds an average pyramid and then it makes a mathematical computation of a deviation pyramid, which is constituted from the significant changes between the stages of the average pyramid.
  3. Bit Plane Encoding is a kind of method that uses the resolution of the image too. It divides the pecking order of the image in certain and unchangeable analysis and so the decoded image equal dimensions with the initial image.
  4. Differential Pulse Code Modulation, compress the image from 1,5 to 3 times. This method cannot achieve great levels of compression and it can't provide continuous sending.

All lossless methods cannot achieve big compress levels.

Moreover wavelet transformation that is used for compression of medical images is lossless (9 apo wavelet compression of medical images) because all the images when decoded can be in the initial form. In order to calculate a certain wavelet transform two filters must be utilized that are proposed in (9 apo wavelet compression of medical images) and what is given must be divided by a multiple of two. Each of the 2 filters divide the image in two other images of the same size with the initial image, so they give four images that contain information about the frequency of the directions. This is performed again and again on one certain image of them k times so a k level of wavelet transformation is made. In Figure 1 there is a five level wavelet transformation of a Magnetic Resonance.

The most regular method that wavelet compression uses is to apply the separated wavelet modification, then divide the wavelet factors that are gives the previous step and finally encode properly the wavelet factors. In the second part of the wavelet compression, the division of the wavelet factors, is the removal of redundant information. There the numerical quantities measured are substituted for estimations that are easier to compress and have better results in compression. The amount of compression depends on the amount of division of the factors that can take. In case the amount of division is big more values of the factors become zero and the amount of compression is getting bigger. On the other if the amount of division is small the amount of compression is small. Eventually the factors are changed with values in order to send or store them properly.