Automated Grading of Lumbar Disc Degeneration via Supervised Distance Metric Learning Xiaoxu He a , Mark Landis a , Stephanie Leung a , James Warrington a , Olga Shmuilovich a , and Shuo Li a a The digital imaging group of London, Dept. of Medical Imaging, Western University, London, ON N6A 3K7, Canada ABSTRACT Lumbar disc degeneration (LDD) is a commonly age-associated condition related to low back pain, while its consequences are responsible for over 90% of spine surgical procedures. In clinical practice, grading of LDD by inspecting MRI is a necessary step to make a suitable treatment plan. This step purely relies on physicians manual inspection so that it brings the unbearable tediousness and inefficiency. An automated method for grading of LDD is highly desirable. However, the technical implementation faces a big challenge from class ambiguity, which is typical in medical image classification problems with a large number of classes. This typical challenge is derived from the complexity and diversity of medical images, which lead to a serious class overlapping and brings a great challenge in discriminating different classes. To solve this problem, we proposed an automated grading approach, which is based on supervised distance metric learning to classify the input discs into four class labels (0: normal, 1: slight, 2: marked, 3: severe). By learning distance metrics from labeled instances, an optimal distance metric is modeled and with two attractive advantages: (1) keeps images from the same classes close, and (2) keeps images from different classes far apart. The experiments, performed in 93 subjects, demonstrated the superiority of our method with accuracy 0.9226, sensitivity 0.9655, specificity 0.9083, F-score 0.8615. With our approach, physicians will be free from the tediousness and patients will be provided an effective treatment. Keywords: Lumbar Disc Degeneration, Automated Grading, Supervised Distance Metric Learning, Image Classification 1. INTRODUCTION As the cartilaginous structures that lie between two adjacent vertebrae, lumbar discs have an important function in absorbing shock and providing flexibility to the spine. 1–3 This function is derived from their unique composition structured by a soft mucoid center, the nucleus pulposus, and the annulus fibrosus which are a strong but flexible ring of collagen fibers. With aging and due to mechanical loading, lumbar disc easily become degenerative, lead to chronic back pain, and are considered to be major causes of functional incapacity. 4, 5 This condition is summarized as lumbar disc degeneration (LDD). LDD is the leading cause of lower back pain and has the risk of disability. Besides, LDD is responsible for over 90% of spine surgical procedures. 1, 6 Grading of LDD is a necessary step for making suitable treatment plan. In clinical practice, MRI is the preferred modality for diagnosing disc due to its better visualization over other modalities. 1, 2 As shown in Fig. 1, a normal disc appears as a bright ellipse surrounded by a dark ring, while the degenerated disc appears darker, with the not clear boundary between the nucleus and the annulus. Existing methods rely on physicians’ visual inspection in spine MRI. This is a heavy burden for physicians and greatly lowers the timely treatment. Automated grading is highly desirable but faces big challenges in technical implementation. As shown in Fig. 2, these challenges, derived from large intra-class image variations and inter-class ambiguity, leads to a serious class-overlapping problem when grading the input image with several grading labels. To solve these problem, we present an automated and accurate method, based on supervised distance met- ric learning, for clinical grading of LDD. It creatively clinical LDD grading task as a multi-class problem shown in Further author information: (Send correspondence to Shuo Li) E-mail: slishuo@gmail.com, Telephone: 519-646-6000 (ext. 64624) Medical Imaging 2017: Computer-Aided Diagnosis, edited by Samuel G. Armato III, Nicholas A. Petrick, Proc. of SPIE Vol. 10134, 1013443 · © 2017 SPIE · CCC code: 1605-7422/17/$18 · doi: 10.1117/12.2253688 Proc. of SPIE Vol. 10134 1013443-1 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 03/05/2017 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx