Stenosis segmentation of lumbar spine using dilated fully convolutional neural network

Abstract

Human spine or backbone may be a distinctive and complicated structure . The body part spine
refers to the lower back, wherever the spine curves inward towards the abdomen . Due to the
bone over growth, the formation of bone spurs occurs which cause spinal stenosis . Diseases
such as osteoarthritis , rheumatoid arthritis and paget’s disease are the main cause of spinal
stenosis . Stenosis refers to the abnormal narrowing of bone in the body . Lumbar stenosis may
cause chronic lower back pain . Magnetic Resonance Imaging scan of lumbar spine is
segmented to analyse any change in bone . In this paper we propose a method to segment the
spine based on using dilated convolutional neural network and pixel classification . Our
method provides an automatic segmentation of lumbar stenosis .