MORE | Spring 2022

Lesion Detection using Three-Dimensional CBCT Image Segmentation using Deep Learning

Health icon, disabled. A red heart with a cardiac rhythm running through it.

The goal of this project is to implement an efficient and accurate 3D semantic segmentation on Cone-Beam Computed Tomography (CBCT) dental imaging for accurate lesion detection using a 3D U-Net method. Applying this technology in diagnostic medical imaging can lead to early detection of various types of lesions, which is vital for medical diagnosis. In this project, the research team utilized high-resolution 3D CBCT images for lesion detection and performed a multi-class segmentation on the collected data from the medical school of the University of Pennsylvania. Finally, the research team used the deep 3D U-net architecture that presented a network and a training strategy that relies on the substantial use of the data augmentation techniques for the annotated samples.

Student researcher

Ajey Musuwati Rajkumar

Ajey Musuwathi Rajkumar

Industrial engineering

Hometown: Chennai, Tamil Nadu, India

Graduation date: Fall 2022