Analysis of Cervical Spinal MRI With Deep Learning

Bezmialem Vakif University

Overview

The aim of this study is analyzing the pathologies in cervical spinal MRI images by using image processing algorithms. Determination of these pathological cases which taught to the system with deep learning and determination of their levels. Finally; verification of the system by comparing radiologist reports and automated system outputs.

Neck pain is a very common health problem with a worldwide prevalence ranging from 16.7% to 75.1%. The source of neck pain is often considered - although there is no strong evidence - the cervical intervertebral disc. Radiological imaging methods are used for the detection of degeneration of the discs and the end plaque changes in the vertebral body corresponding to this degeneration.Magnetic Resonance Imaging (MRI) gives information about the structure of intervertebral disc, width of spinal canal and tissues outside the canal. However, there is no standardization in the identification and evaluation of radiological images, and interobserver variability is high. Studies have been initiated on automated systems that analyze MRI images to increase the accuracy and consistency of reporting procedures. Examining MRI images with deep learning can lead to the production of systems that help clinical decision making and also allows the evaluation of large data in a short time.

Study Type

OBSERVATIONAL

Conditions

Cervical Disc Disease Cervical Spine Disease

Interventions

Cervical Spinal MRIDIAGNOSTIC_TEST

Cervical Spinal MRI images of 500 patients will be entered into the system for modeling

Eligibility

Sex: ALLMin age: 18 YearsMax age: 75 Years

Medical Language ↔ Plain English

Inclusion Criteria: * 18-75 years of age * Having result of a cervical spinal MRI, which was performed for neck pain in the hospital records in the last 5 years. Exclusion Criteria: * Malignancy * Signs of active infection * Significant spinal vertebral deformity (advanced scoliosis, congenital vertebral defects) * Spinal surgery

Locations (1)

Bezmialem Vakif University Hospital

Istanbul, Turkey (Türkiye)

Outcomes

Primary Outcomes

Accuracy rate of the model as assessed by cross validation of the data set

We will randomly divide the dataset into 4 subsets. In each sub-experiments, MRI slices from 3 subsets will be trained and slices in the other subset will be tested. We will perform totally 4 sub-experiments, so each slice in the dataset will be tested once.

Time frame: Through study completion, an average of 1,5 years

Reliability of the model as assessed by comparing the reports of the model and radiologist.

Kappa statistics and reliability coefficients will be use.

Time frame: Through study completion, an average of 1,5 years

Data from ClinicalTrials.gov

This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.