CNN-Based Automatic Diagnosis for Knee Meniscus Tear in Magnetic Resonance Images

The meniscus tear is common for athletes and the elderly, and because of its irresistible nature, early diagnosis and treatment is particularly important. Magnetic Resonance (MR) Imaging is noninvasive and has a high diagnostic accuracy of 98%, which has been considered as an ideal method to diagnose meniscus tear. However, it is a time consuming process for radiologists to diagnose meniscus tear by comparing dozens of MR images and to diagnose the tear grade. Here, the convolutional neural network (CNN) is used to accomplish the aim of automatic diagnosis of tear grade. At first, we apply the Hough transformation to preprocess the data, during which we shrink the image to about 1/10 of its original size so that the meniscus local candidate frame (LCF) is formed. Then, to validate the method, 3062 actual clinical MR images were used and an accuracy of 89.5% is achieved. The experimental result has demonstrated the validity of our proposed method, which can meet the needs of radiologists for computer-aided diagnosis of meniscus tear classification.

[1]  W. S. Lee,et al.  Identification and determination of the number of immature green citrus fruit in a canopy under different ambient light conditions , 2014 .

[2]  J. Katz,et al.  Associations among meniscal damage, meniscal symptoms and knee pain severity. , 2017, Osteoarthritis and cartilage.

[3]  Dana H. Ballard,et al.  Generalizing the Hough transform to detect arbitrary shapes , 1981, Pattern Recognit..

[4]  Dali Chen,et al.  Optic disc segmentation method based on low rank matrix recovery theory , 2018, 2018 Chinese Control And Decision Conference (CCDC).

[5]  Wen-Jia Kuo,et al.  Two-Stage Road Sign Detection and Recognition , 2007, 2007 IEEE International Conference on Multimedia and Expo.

[6]  L. Bassett,et al.  Meniscal injuries: detection using MR imaging. , 1986, Radiology.

[7]  Fan Jun,et al.  Recognition of mature oranges in natural scene based on machine vision , 2008 .

[8]  A. Inderster,et al.  Influence of Magnetic Resonance Imaging on Indications for Arthroscopy of the Knee , 1996, Clinical orthopaedics and related research.

[9]  M. Axe,et al.  Knee Pain and Mobility Impairments: Meniscal and Articular Cartilage Lesions Revision 2018: Clinical Practice Guidelines Linked to the International Classification of Functioning, Disability and Health From the Orthopaedic Section of the American Physical Therapy Association , 2018, The Journal of orthopaedic and sports physical therapy.

[10]  H. Thomsen,et al.  Mr Imaging before Arthroscopy in Knee Joint Disorders? , 1996, Acta radiologica.

[11]  Chein-I Chang,et al.  An Automatic Computer-Aided Detection System for Meniscal Tears on Magnetic Resonance Images , 2009, IEEE Transactions on Medical Imaging.

[12]  Jurjen Caarls,et al.  Fast and Accurate Robot Vision for Vision Based Motion , 2000, RoboCup.

[13]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[14]  M G Myriam Hunink,et al.  MR imaging of the menisci and cruciate ligaments: a systematic review. , 2003, Radiology.

[15]  Cemal Köse,et al.  An automatic diagnosis method for the knee meniscus tears in MR images , 2009, Expert Syst. Appl..

[16]  Richard O. Duda,et al.  Use of the Hough transformation to detect lines and curves in pictures , 1972, CACM.

[17]  E. Panagiotopoulos,et al.  A computer-based system for the discrimination between normal and degenerated menisci from Magnetic Resonance Images , 2008, 2008 IEEE International Workshop on Imaging Systems and Techniques.

[18]  Songül Albayrak,et al.  An efficient and fast computer-aided method for fully automated diagnosis of meniscal tears from magnetic resonance images , 2019, Artif. Intell. Medicine.