Automated measurements of lumbar lordosis in T2-MR images using decision tree classifier and morphological image processing

Lumbar spine’s lordosis is a very important parameter functionally and clinically; it is a key feature in maintaining the sagittal balance, in addition to its crucial role in evaluating the spinal deformities. The main objective of the current study is to present a fully-automated measurement of the lumbar spine’s lordotic curve angle in T2-MR images. This goal has been achieved by the automatic measurement of lordosis radius at the lumbar spine level by computer-aided methods utilizing data mining classification and image segmentation followed by morphological image processing. The spine has been segmented from the entire image using a machine-learning technique that is based on texture features for recognizing the lumbar-spine pattern. The extracted features were fed to C4.5 decision tree classifier for designing the lumbar-spine recognition system. The resultant classifier’s “if-then” rules have been employed for segmenting the spine region from the entire image. Multiple morphological image processes have been applied to the raw segmentation result to enhance the true positive rate and suppressing the false positive rate. The mean radius of lumbar spine’s curvature has been evaluated by fitting the contours average to the closest circle using least-square fitting algorithm which was followed by calculating the lumbar lordosis curvature angle. The proposed approach has been tested and validated on normal and pathological T2-MR spine images and found to perform effectively. The calculation of lumbar lordosis angles showed a strong correlation with the Cobb angle measurements (R = 93.2%). Keywords: Spine, Lumbar lordosis curvature, T2-MRI, Data mining, Decision trees, Texture features, Image segmentation, Morphological image processing, Cobb angle

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.