Ultrasound image analysis for identifying blood in an existing effusion in knee joint

Ultrasound, an inexpensive, accessible and portable device is gaining popularity in various disease diagnoses. In this project, we aim to analyze the ultrasound images generated for knee joint for diagnosing hemarthrosis (joint bleeding), a common clinical event in patients with severe hemophilia. We aim to analyze the images using machine learning and deep learning techniques. As the major challenge in this case is the scarcity of number of cases with hemophilia, we plan to use one-shot learning techniques, which use neural networks to learn the similarity features between images and need only few training images. The learned model will be deployed in 16 Bit software for diagnosing hemarthrosis in hemophilic patients.

Faculty Supervisor:

Vassilios Tzerpos

Student:

Rajshree Daulatabad

Partner:

16 Bit Inc.

Discipline:

Engineering - computer / electrical

Sector:

Professional, scientific and technical services

University:

York University