Amgen Scholars: Announcements of Opportunity
Related Pages 
Below are Announcements of Opportunity posted by Caltech faculty for the Amgen Scholars program.
Announcements of Opportunity are posted as they are received. Please check back regularly for new AO submissions! Remember: This is just one way that you can go about identifying a suitable project and/or mentor. For additional tips on identifying a mentor click here.
Please remember:
- Students pursuing Amgen must be U.S. citizens, U.S. permanent residents, or students with DACA status.
- Students pursuing Amgen must complete the 10-week program from June 21 - August 25, 2023. Students must commit to these dates. No exceptions will be made.
- Accepted students must live in provided Caltech housing.
<< Prev
Record
34 of
59
Next >>
Back To List
| Project: | Understanding Misclassifications - a data Science Approach | ||||||||
| Disciplines: | Multidisciplinary, Data Science, Medical Science | ||||||||
| Mentor: |
Ashish Mahabal,
Lead Computational Scientist, (PMA),
aam@astro.caltech.edu, |
||||||||
| Mentor URL: | http://www.astro.caltech.edu/~aam (opens in new window) | ||||||||
| Background: | A set of lung images with tumors is being annotated by radiologists prior to and after being trained on a set of defined images that portray features of interest. For each image there will be a set of attributes that the radiologists must score. At baseline the radiologists will be presented for each attribute with the lexicon of possible responses, presented through a drop-down box supplied through the Zooniverse software. A fraction of the images will be repeated in the pre- and post training images. The primary outcome is the change in accuracy for the radiologists compared to a gold standard as marked by experts on the same set of images. The detailed annotations will lead to interesting possibilities for not only detecting lung cancer, but also detailed classification based on features like density, margin, shape, size and so on. | ||||||||
| Description: | The project will include the following tasks: (1) Comparison of pre-, post- and gold annotations of the image data to understand possible biases, (2) Creation of a larger, comprehensive labeled dataset using similar 3D datasets, (3) Building machine learning applications, in particular convolutional neural networks (CNNs) that are trained on the annotated data and run on additional similar data for detecting tumors and classifying them, (4) If possible, using longitudinal data, determine nascency of the tumors. An emphasis in this entire process will be to understand misclassifications, and ambiguous classifications, and ways to disambiguate them. 5) Build/automate summary reports on model performance based on different annotations. | ||||||||
| References: |
Zooniverse: https://www.zooniverse.org/ CNNs: https://en.wikipedia.org/wiki/Convolutional_neural_network Lung nodule features: Lobar location, Conspicuity, Margin, Cavitation, Calcification, Fibrosis, … |
||||||||
| Student Requirements: | Proficiency in python, jupyter notebooks (Google Colab), and git. Conversant with basics of machine learning and statistics, knowledge about linux/unix. Basic biology knowledge will be a plus. Some experience with deep learning, GPUs. | ||||||||
| Programs: |
This AO can be done under the following programs:
|
<< Prev Record 34 of 59 Next >> Back To List