Benjamin S Hopkins1, Kenneth A Weber2, Kartik Kesavabhotla1, Monica Paliwal1, Donald R Cantrell3, Zachary A Smith4. 1. Department of Neurological Surgery, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA. 2. Department of Radiology, Stanford University School of Medicine, Stanford, California, USA. 3. Department of Radiology, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA. 4. Department of Neurological Surgery, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA. Electronic address: zsmith1@nm.org.
Abstract
BACKGROUND: Cervical spondylotic myelopathy (CSM) severity and presence of symptoms are often difficult to predict based simply on clinical imaging alone. Similarly, improved machine learning techniques provide new tools with immense clinical potential. METHODS: A total of 14 patients with CSM and 14 controls underwent imaging of the cervical spine. Two different artificial neural network models were trained; 1) to predict CSM diagnosis; and 2) to predict CSM severity. Model 1 consisted of 6 inputs including 3 common imaging scales for the evaluation of cord compression, alongside 3 objective magnetic resonance imaging measurements. The outcome for model 1 was binary to predict CSM diagnosis. Model 2 consisted of 23 input variables derived from probabilistic volume mapping measurements of white matter tracts in the region of compression. The outcome of model 2 was linear, to predict the modified Japanese Orthopedic Association (mJOA) score. RESULTS: Model 1 was used in predicting CSM. The mean cross-validated accuracy of the trained model was 86.50% (95% confidence interval, 85.16%-87.83%) with a median accuracy of 90.00%. Area under the curve (AUC) was calculated for each repetition. Average AUC for each repetition was 0.947 with a median AUC of 1.0. Average sensitivity, specificity, positive predictive value, and negative predictive value were 90.25%, 85.05%, 81.58%, and 91.94%, respectively. Model 2 was used in modeling mJOA. The mJOA model predicted scores, with a mean and median error of -0.29 mJOA points and -0.08 mJOA points, respectively, mean error per batch was 0.714 mJOA points. CONCLUSIONS: Machine learning provides a promising method for prediction, diagnosis, and even prognosis in patients with CSM.
BACKGROUND:Cervical spondylotic myelopathy (CSM) severity and presence of symptoms are often difficult to predict based simply on clinical imaging alone. Similarly, improved machine learning techniques provide new tools with immense clinical potential. METHODS: A total of 14 patients with CSM and 14 controls underwent imaging of the cervical spine. Two different artificial neural network models were trained; 1) to predict CSM diagnosis; and 2) to predict CSM severity. Model 1 consisted of 6 inputs including 3 common imaging scales for the evaluation of cord compression, alongside 3 objective magnetic resonance imaging measurements. The outcome for model 1 was binary to predict CSM diagnosis. Model 2 consisted of 23 input variables derived from probabilistic volume mapping measurements of white matter tracts in the region of compression. The outcome of model 2 was linear, to predict the modified Japanese Orthopedic Association (mJOA) score. RESULTS: Model 1 was used in predicting CSM. The mean cross-validated accuracy of the trained model was 86.50% (95% confidence interval, 85.16%-87.83%) with a median accuracy of 90.00%. Area under the curve (AUC) was calculated for each repetition. Average AUC for each repetition was 0.947 with a median AUC of 1.0. Average sensitivity, specificity, positive predictive value, and negative predictive value were 90.25%, 85.05%, 81.58%, and 91.94%, respectively. Model 2 was used in modeling mJOA. The mJOA model predicted scores, with a mean and median error of -0.29 mJOA points and -0.08 mJOA points, respectively, mean error per batch was 0.714 mJOA points. CONCLUSIONS: Machine learning provides a promising method for prediction, diagnosis, and even prognosis in patients with CSM.
Authors: Jun Wei; Lubomir M Hadjiiski; Berkman Sahiner; Heang-Ping Chan; Jun Ge; Marilyn A Roubidoux; Mark A Helvie; Chuan Zhou; Yi-Ta Wu; Chintana Paramagul; Yiheng Zhang Journal: Acad Radiol Date: 2007-06 Impact factor: 3.173
Authors: Tom M Mitchell; Svetlana V Shinkareva; Andrew Carlson; Kai-Min Chang; Vicente L Malave; Robert A Mason; Marcel Adam Just Journal: Science Date: 2008-05-30 Impact factor: 47.728
Authors: Ron Pisters; Deirdre A Lane; Robby Nieuwlaat; Cees B de Vos; Harry J G M Crijns; Gregory Y H Lip Journal: Chest Date: 2010-03-18 Impact factor: 9.410
Authors: U Joseph Schoepf; Alex C Schneider; Marco Das; Susan A Wood; Jugesh I Cheema; Philip Costello Journal: J Thorac Imaging Date: 2007-11 Impact factor: 3.000
Authors: Kai-Uwe LewandrowskI; Narendran Muraleedharan; Steven Allen Eddy; Vikram Sobti; Brian D Reece; Jorge Felipe Ramírez León; Sandeep Shah Journal: Int J Spine Surg Date: 2020-12
Authors: Mark E Stephens; Christen M O'Neal; Alison M Westrup; Fauziyya Y Muhammad; Daniel M McKenzie; Andrew H Fagg; Zachary A Smith Journal: Neurosurg Rev Date: 2021-09-07 Impact factor: 3.042
Authors: Zachary A Smith; Kenneth A Weber; Monica Paliwal; Benjamin S Hopkins; Alexander J Barry; Donald Cantrell; Aruna Ganju; Tyler R Koski; Todd B Parrish; Yasin Dhaher Journal: World Neurosurg Date: 2019-11-11 Impact factor: 2.104