Changes in Brain White Matter Assessed Via Textural Features Using a Neural Network

Changes in Brain White Matter Assessed Via Textural Features Using a Neural Network

R. Kalpana (Anna University Chennai, India), S. Muttan (Anna University Chennai, India) and B. Agrawala (Bharat Scans, India)
DOI: 10.4018/978-1-4666-1755-1.ch009
OnDemand PDF Download:
No Current Special Offers


Diffusion Tensor Magnetic Resonance Imaging (DTMRI) has proved useful for microstructure characterization of the brain. This technique also helps determining complex connectivity of fiber tracts. The brain white matter (BMW) changes with respect to age and corresponding appearance of white-matter lesions among the brain’s message-carrying axons affects cognitive functions in old age. In this paper, the observed morphology in BWM on ageing is analyzed using statistical parameters extracted from DTMR images of different age groups. The gray level co-occurrence matrix (GLCM) obtained from the segmented images gives 14 textural features, subsets of which are adopted as the input sets in a backpropagation neural network classifier. The network is trained to predict the age based on BMW details used as the inputs. The proposed method helps in understanding the age-related changes in white matter. This is useful for the physician in understanding miscorrelation in motor activities and relevant causes in elderly subjects.
Chapter Preview

Data Gathering

DTMR images obtained at the center of corpus collasum from 60 volunteers of age group between 55 and 80) are segmented for white matter using automated segmenting procedure and analysed for age-related changes in brain fiber tracts. The procedure is as follows:

  • Obtaining the DTMR images from the volunteers

  • Segmenting the image for white matter

  • Finding co-occurrence matrix of the quantised white matter image

  • Finding the appropriate statistical features from GLCM

  • Finding the most suitable features for classification

  • Using back propagation neural network, the image is classified against the age.


Image Acquisition

MRI acquisition was performed on a GE 3 Tesla Signa HDX system equipped with an 8-channel brain array coil using a Diffusion Tensor Imaging with Fourier transform. Protocol: Field-of-view (FOV) 240 × 240mm; matrix 256 × 256; TR=7400; Number of diffusion direction = 25; b value = 1000; voxel size 0.9375 × 0.9375 × 5 mm; slice thickness = 5mm; scanning time = 8 min. Scanning done in axial plane parallel to the long axis of the body of corpus callosum. Ten images under each age group, totaling 60 images (on subjects with mean age = 64.4483, SD = 8.7841) from both male and female (with no neurological problem) are considered for the purpose. Image analysis methods are implemented in MATLABTM (Version 7.6).

Complete Chapter List

Search this Book: