Fig. 4. Framework of the proposed method
3.1 Image Selection and Preprocessing
Selection of images is vital for the improvement of accuracy in identification of stroke. Four CT images of the brain affected by the stroke lesions are taken from the internet. Each image has a different shape, size and location of stroke lesion. Image preprocessing is one of the preliminary steps that are highly required to ensure the high accuracy of the consecutive steps. The CT brain images normally consist of some patient specific and equipment based artifacts. The Study shows that the CT images contain Gaussian noise. Thus, image preprocessing is needed to smoothen, enhance and remove noise from CT images that may be caused by defects of CT scanner so that segmentation would be easier and more effective. A Gaussian filter is used [25] for initial filtering the image with an impulse response h(t) given by equation (1) and it is applied to the image for two successive times to remove the noise in the CT image. …show more content…
Thus, the image is assumed to consist of two classes: background and object. Now, the intensity probability distribution function (pdf) of each of these regions denoted as〖 p〗_1^I and p_2^I are modeled using Gaussian distribution whose parameters are automatically updated during the evolution of the level set function. The segmentation method starts by initializing the level set function '∅^' [25] as the unit step function which takes a negative constant value -C_0 inside the contour 'C^(' )and a positive constant value C_0 outside it. Then, mean 〖'u〗_i' and variance 'σ_i' of the local Gaussian distribution corresponding to the pdf for the background and the object are calculated as
3.1 Image Selection and Preprocessing
Selection of images is vital for the improvement of accuracy in identification of stroke. Four CT images of the brain affected by the stroke lesions are taken from the internet. Each image has a different shape, size and location of stroke lesion. Image preprocessing is one of the preliminary steps that are highly required to ensure the high accuracy of the consecutive steps. The CT brain images normally consist of some patient specific and equipment based artifacts. The Study shows that the CT images contain Gaussian noise. Thus, image preprocessing is needed to smoothen, enhance and remove noise from CT images that may be caused by defects of CT scanner so that segmentation would be easier and more effective. A Gaussian filter is used [25] for initial filtering the image with an impulse response h(t) given by equation (1) and it is applied to the image for two successive times to remove the noise in the CT image. …show more content…
Thus, the image is assumed to consist of two classes: background and object. Now, the intensity probability distribution function (pdf) of each of these regions denoted as〖 p〗_1^I and p_2^I are modeled using Gaussian distribution whose parameters are automatically updated during the evolution of the level set function. The segmentation method starts by initializing the level set function '∅^' [25] as the unit step function which takes a negative constant value -C_0 inside the contour 'C^(' )and a positive constant value C_0 outside it. Then, mean 〖'u〗_i' and variance 'σ_i' of the local Gaussian distribution corresponding to the pdf for the background and the object are calculated as