Limits...
Classifying dementia using local binary patterns from different regions in magnetic resonance images.

Oppedal K, Eftestøl T, Engan K, Beyer MK, Aarsland D - Int J Biomed Imaging (2015)

Bottom Line: The best result we achieved was in the two-class problem NC versus AD + LBD with total accuracy of 0.98 (0.04).The performance using 3DT1 images was notably better than when using FLAIR images.The results from the WM region gave similar results as in the WML region.

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical Engineering and Computer Science, University of Stavanger, 4036 Stavanger, Norway ; Centre for Age-Related Medicine, Stavanger University Hospital, Stavanger, Norway.

ABSTRACT
Dementia is an evolving challenge in society, and no disease-modifying treatment exists. Diagnosis can be demanding and MR imaging may aid as a noninvasive method to increase prediction accuracy. We explored the use of 2D local binary pattern (LBP) extracted from FLAIR and T1 MR images of the brain combined with a Random Forest classifier in an attempt to discern patients with Alzheimer's disease (AD), Lewy body dementia (LBD), and normal controls (NC). Analysis was conducted in areas with white matter lesions (WML) and all of white matter (WM). Results from 10-fold nested cross validation are reported as mean accuracy, precision, and recall with standard deviation in brackets. The best result we achieved was in the two-class problem NC versus AD + LBD with total accuracy of 0.98 (0.04). In the three-class problem AD versus LBD versus NC and the two-class problem AD versus LBD, we achieved 0.87 (0.08) and 0.74 (0.16), respectively. The performance using 3DT1 images was notably better than when using FLAIR images. The results from the WM region gave similar results as in the WML region. Our study demonstrates that LBP texture analysis in brain MR images can be successfully used for computer based dementia diagnosis.

No MeSH data available.


Related in: MedlinePlus

Overview of MR images and the ROIs used for feature extraction. (a) in the top left corner shows an example of an axial FLAIR MR image. The white matter lesions are possible to see as hyperintense areas. (b) in the top right shows the segmented voxels labelled as WML overlayed on the FLAIR MR image seen in (a). (c) in the bottom left corner shows the segmented WML voxels, found from the corresponding FLAIR, overlayed on the T1 MR image. (d) in the bottom right corner shows the segmented WM voxels overlayed on the T1 MR image.
© Copyright Policy - open-access
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4385607&req=5

fig3: Overview of MR images and the ROIs used for feature extraction. (a) in the top left corner shows an example of an axial FLAIR MR image. The white matter lesions are possible to see as hyperintense areas. (b) in the top right shows the segmented voxels labelled as WML overlayed on the FLAIR MR image seen in (a). (c) in the bottom left corner shows the segmented WML voxels, found from the corresponding FLAIR, overlayed on the T1 MR image. (d) in the bottom right corner shows the segmented WM voxels overlayed on the T1 MR image.

Mentions: For the 2D texture analysis approach, the LBP values as well as the C measure were calculated from every voxel in the selected ROI and MR image type for all subjects in the data set using Matlab [55]. Three different combinations of neighbourhood radius (R) and number of samples (P), namely, R = 1 and P = 8, R = 2 and P = 12, and R = 4 and P = 16, were used. Mean, standard deviation, variation, median, interquartile range, entropy, skewness, and kurtosis of the ROI-wise collected LBP and C values were calculated to be used as a descriptor of the distributions of the LBP and C values. These features were subjected to further selection and classification resulting in 8 features for each of the three combinations of R and P for both LBP and C resulting in a total of 48 features for each subject. See Figure 3 for an example of the FLAIR and T1 MR images and the WML segmentation results. See also Figure 4 for an example of LBP- and C-valued images based on the FLAIR and T1 MR images.


Classifying dementia using local binary patterns from different regions in magnetic resonance images.

Oppedal K, Eftestøl T, Engan K, Beyer MK, Aarsland D - Int J Biomed Imaging (2015)

Overview of MR images and the ROIs used for feature extraction. (a) in the top left corner shows an example of an axial FLAIR MR image. The white matter lesions are possible to see as hyperintense areas. (b) in the top right shows the segmented voxels labelled as WML overlayed on the FLAIR MR image seen in (a). (c) in the bottom left corner shows the segmented WML voxels, found from the corresponding FLAIR, overlayed on the T1 MR image. (d) in the bottom right corner shows the segmented WM voxels overlayed on the T1 MR image.
© Copyright Policy - open-access
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4385607&req=5

fig3: Overview of MR images and the ROIs used for feature extraction. (a) in the top left corner shows an example of an axial FLAIR MR image. The white matter lesions are possible to see as hyperintense areas. (b) in the top right shows the segmented voxels labelled as WML overlayed on the FLAIR MR image seen in (a). (c) in the bottom left corner shows the segmented WML voxels, found from the corresponding FLAIR, overlayed on the T1 MR image. (d) in the bottom right corner shows the segmented WM voxels overlayed on the T1 MR image.
Mentions: For the 2D texture analysis approach, the LBP values as well as the C measure were calculated from every voxel in the selected ROI and MR image type for all subjects in the data set using Matlab [55]. Three different combinations of neighbourhood radius (R) and number of samples (P), namely, R = 1 and P = 8, R = 2 and P = 12, and R = 4 and P = 16, were used. Mean, standard deviation, variation, median, interquartile range, entropy, skewness, and kurtosis of the ROI-wise collected LBP and C values were calculated to be used as a descriptor of the distributions of the LBP and C values. These features were subjected to further selection and classification resulting in 8 features for each of the three combinations of R and P for both LBP and C resulting in a total of 48 features for each subject. See Figure 3 for an example of the FLAIR and T1 MR images and the WML segmentation results. See also Figure 4 for an example of LBP- and C-valued images based on the FLAIR and T1 MR images.

Bottom Line: The best result we achieved was in the two-class problem NC versus AD + LBD with total accuracy of 0.98 (0.04).The performance using 3DT1 images was notably better than when using FLAIR images.The results from the WM region gave similar results as in the WML region.

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical Engineering and Computer Science, University of Stavanger, 4036 Stavanger, Norway ; Centre for Age-Related Medicine, Stavanger University Hospital, Stavanger, Norway.

ABSTRACT
Dementia is an evolving challenge in society, and no disease-modifying treatment exists. Diagnosis can be demanding and MR imaging may aid as a noninvasive method to increase prediction accuracy. We explored the use of 2D local binary pattern (LBP) extracted from FLAIR and T1 MR images of the brain combined with a Random Forest classifier in an attempt to discern patients with Alzheimer's disease (AD), Lewy body dementia (LBD), and normal controls (NC). Analysis was conducted in areas with white matter lesions (WML) and all of white matter (WM). Results from 10-fold nested cross validation are reported as mean accuracy, precision, and recall with standard deviation in brackets. The best result we achieved was in the two-class problem NC versus AD + LBD with total accuracy of 0.98 (0.04). In the three-class problem AD versus LBD versus NC and the two-class problem AD versus LBD, we achieved 0.87 (0.08) and 0.74 (0.16), respectively. The performance using 3DT1 images was notably better than when using FLAIR images. The results from the WM region gave similar results as in the WML region. Our study demonstrates that LBP texture analysis in brain MR images can be successfully used for computer based dementia diagnosis.

No MeSH data available.


Related in: MedlinePlus