This project aims to develop a machine learning model capable of reconstructing a 3D volume using the ultrasound images of a forearm, tracked probe data, and the calibration matrix as inputs. The model should be able to compensate for inaccuracies in the calibration matrix, thereby improving the accuracy of the reconstructed 3D volume.
What you will do:
- Research on existing approaches for reconstructing volumes from 2D and 3D ultrasound images.
- Generate data for ultrasound probe calibration.
- Perform 2D and 3D probe calibration to derive the calibration matrix.
- Develop a machine learning model to reconstruct the 3D ultrasound volume.
What you should already know:
Machine Learning
Python
What you will deliver:
- An end-to-end pipeline for reconstructing 3D volumes from ultrasound images acquired with both 2D and 3D probes.
- Extended goal: Develop a method for trackerless 3D volume reconstruction using the model trained on the data from the tracking system.
Nice to have:
Prior experience with robotics and image processing