Machine Learning Audit of Radiological Images

The challenge

The client needed to automate and strengthen the audit of radiological images, a process that relied on manual review.

The approach

Kumulus prepared the data, developed computer vision and machine learning models on Microsoft Azure and carried out technical validation of the prioritised process.

The outcome

Auditors gained a consistent way to identify dental radiographs that may need further review. The computer-vision workflow flags possible manipulation, including contrast changes, rotation, resizing and re-photographed images.