MPHY0043: Artificial Intelligence for Surgery and Intervention

UCL Module | MPBE | UCL Moodle Page

Term 1 (Autumn)

Contacts

Name	Email	Role
Yipeng Hu	[email protected]	Module Lead
Athena Reissis	[email protected]	Tutor
Weixi Yi	[email protected]	Tutor
Wen Yan	[email protected]	Tutor

1. Programming and development

Python, numerical computing

All practical tutorials, group work and coursework projects in this module are based on Python, with a number of common libraries, including NumPy and Matplotlib. For a refresher or relevant materials in medical image analysis, please have a look at the UCL Module MPHY0030.

Machine learning, deep learning

This module uses two deep learning libraries, TensorFlow and PyTorch. Guide and tutorial materials for the deep learning libraries are widely available, for example, from the UCL Module COMP0197, with relevant materials designed for medical imaging applications in the UCL Module MPHY0041.

MONAI is also used, with many dedicated deep learning functionalities designed for medical applications.

Development tools

Micromamba is used for tutorials and may be required for the assessed group work and coursework, together with Jupyter Notebook, by following the Development Tools.

Although not required, it is encouraged to use Git with this repository. Tutorials for its basic uses are also widely available, e.g. Work with Git.

2. Tutorials

| tools | envs | learning type | applications | remarks |

Go to individual tutorial sub-directories and read the readme.md file to get started.

Surgical Data Regression

Tutorial directory
Keywords: Classical machine learning, linear algebra, optimisation, NumPy, TensorFlow and PyTorch
Devlopement tools: Jupyter Notebook

Surgical Gesture and Skill

Tutorial directory
Keywords: supervised classification, PyTorch, 3D CNN, JIGSAWS
Devlopement tools: PyTorch

3D Medical Image Segmentation

Tutorial directory
Keywords: PyTorch, segmentation, MONAI U-Net, clinical imaging data
Devlopement tools: MONAI, PyTorch with Jupyter Notebook

Medical Image Registration

Tutorial directory
Keywords: PyTorch, Unsupervised registration, MONAI, MedNist dataset
Devlopement tools: MONAI, PyTorch with Jupyter Notebook

Intraoperative Motion Classification

Tutorial directory
Keywords: TensorFlow, Keras, PointNet, simulated dataset
Devlopement tools: Keras with TensorFlow

Vision and Workflow

Tutorial directory
Keywords: TensorFlow Keras, Supervised classification, "off-the-shelf" networks, endoscopic video data
Devlopement tools: Keras with TensorFlow

3. Reading list

A collection of books and research papers, applying artificial intelligence to surgery and intervention, is provided in the Reading List.