Case Study

Heart Failure Diagnostics

A leader in Healthcare

A hospital located in the Benelux region that houses almost 2000 beds and over 9000 employees.

The Challenge

The goal of this project was to predict if a patient needed a pacemaker, based on an echocardiogram of the heart. An echocardiogram (or ‘echo’) transmits high-frequency sound waves that create echoes when they bounce off certain parts of the body, generating a video of the area. An experienced Doctor will look for two visual cues from a heart echo in order to make a diagnosis: apical rocking and septal flash. Unfortunately, misdiagnosis is common so the challenge was whether computer vision could help detect these visual cues.

The Solution

Using the latest computer vision technology, ML6 built a video classifier to recognise the presence of septal flash and apical rocking to assist the Doctor in making a diagnosis.

Results

Initial results demonstrated an accuracy of 74% which is a very promising start. Given more time and a larger data set, more feature engineering could be integrated to further improve the results.

Partner