Medical imaging informatics
Image analysis has become a powerful tool to diagnose, evaluate and follow up overall disease change. As clinical needs evolve, quantitative imaging techniques have been in continuous development to analyze and represent changes on imaging findings that can be used to better produce insights about treatment effectiveness and understand disease progression.
Imaging techniques such as MRI, CT and PET combined with modern computer-based image analysis techniques (computer vision, machine learning) can provide the tools for radiologists to assist in the interpretation of medical imaging studies, having the benefits of shortened interpretation time, improved accuracy and reduced interrater variability. With the help of computer algorithms (deep learning, random forest, SVMs, etc.) abstractions on the data can be obtained to identify the optimal features to represent the data and, therefore, achieve superior performance in difficult classification tasks. Recent advances including large datasets; faster computation; and insights into sparsity, regularization, and optimization have propelled automated imaging analysis forward.
Active research on UCLA's medical imaging informatics includes development of algorithms related to image processing such as segmentation, feature extraction, and classification. Collaborating with multiple experts across domains, imaging informatics research at UCLA spans multiple areas.