ScoliosisX
- 0 Collaborators
Design of a computer system to support medical decisions based on the use of Machine Learning to identify the type of deformation of the spine and quantify its degree of curvature in adolescent patients. ...learn more
Project status: Concept
Groups
Student Developers for AI
Intel Technologies
Intel Python
Overview / Usage
Scoliosis is a malformation of the spine that affects 3% of the global population of adolescents. Because this malformation evolves progressively, its early and accurate diagnosis along with an early corrective treatment is of the greatest importance for patients suffering from it. To determine the degree of severity of a patient's scoliosis, clinical diagnostic techniques such as Cobb's Angle, spine balance, surface topography, computed tomography, ScolioScan and others are applied.
Based on the observation and evaluation of X-ray plates, the angle of deformation that the column presents can be evaluated. Based on this angle, three categories of spine deformation are identified: light, moderate and severe. The recommended treatments range from physiotherapy and the use of orthopedic supports to surgery. Other measurements of the degree of curvature of the spine are equally effective, although some of them are economically expensive or too risky because of the radiation levels to which they expose the patient.
The present project focuses on the exploration of an iterative and flexible computational method aimed at measuring the degree of curvature of the column. Initially, we resort to digital image treatment techniques (radiographic plates), then apply the convolutional neural network method to cover the need for a vast set of training images from only a limited availability of them; and finally, it makes use of K-means to calculate the degree of curvature of scoliosis. The computer system will be able to support the decision-making of the doctor responsible for the diagnosis and treatment of scoliosis.
Methodology / Approach
At present, the diagnosis for idiopathic scoliosis in our country follows traditional processes, which range from the use of measuring devices (scoliometer) to the manual calculation of the angle of curvature of the spine.
It is for this reason that the present project is of an exploratory nature, in which it is intended to address machine learning methodologies, in order to develop a computer system that manages to automate the process of measuring curvature of the spine. For better performance, recent advances in the literature regarding the subject and machine learning methods will be taken into account.
Repository
https://github.com/Fernando23296/scoliosisx