David Staub Email & Phone Number

San Francisco, Ca, Us

San Francisco, California, Us

San Mateo, Ca, Us

San Mateo, Ca, Us

Dallas, Texas, Us

The goal of my postdoc work was to perform novel research exploring applications of machine learning in the fields of Medical Physics and Radiation Oncology. Projects included using machine learning to automate intensity-modulated radiation therapy planning as well as predict gene expression within tumors based on medical imaging features. Specific research components included:• Developing regression and classification models to improve cancer treatments.• Extracting features from medical images using self-written software.• Writing software in Python using numpy, scipy, and scikit-learn packages and storing on local Git server for developing machine learning models.• Using Matlab for prototyping.• Visualizing modeling results using matplotlib Python package.• Storing extracted data in self-designed SQL database.• Using machine learning algorithms including SVM, gradient boosting, and random forests to model cancer outcomes as a function of image features.• Using PCA matrix decomposition to reduce data dimensionality.• Validating and comparing models using train/test splits and statistical t-tests.• Working with team of faculty and postdocs to integrate predictive models into radiation therapy cancer treatment planning software.

Assisted in performing IMRT QA analysis with the MatrixX phantom. Investigated the possible clinical use of the ArcCheck phantom for IMRT QA use.

The focus of my graduate school career was the development of a novel 4D cone-beam CT reconstruction algorithm whose purpose was to produce high quality, artifact free cone-beam CT images at the frame rate of the raw projection data acquisition. The end goal was to enable accurate tracking and reconstruction of moving thoracic anatomy for the purposes of improving lung cancer radiotherapy. Important aspects of my research included:• Developing a novel method for reconstructing CT images by incorporating a patient specific breathing model.• Using multivariate calculus and linear algebra to model the physics of x-ray imaging systems.• Writing libraries in C to implement my reconstruction algorithm.• Using Matlab for prototyping and visualization.• Porting sections of my code to the GPU using CUDA to increase speed by 30x.• Optimizing the breathing model using numerical optimization routines from the GSL C library.• Incorporated sensor data from a patient monitoring device into the breathing model.• Presenting my work at international meetings.• Publishing papers in a top level peer reviewed journal.

My role as research intern was to assist in research investigating the possible role of apoptosis, or programmed cell death, in wound healing and force production using the process of dorsal closure in Drosophila melanogaster as a model system.

Doctor Of Philosophy (Ph.D.), Medical Physics

Bachelor Of Science (B.Sc.), Physics