Student working on a research project.

Histology Image Annotation And Analysis

This project involves generating ground-truth datasets for use in medical image analysis applications. When we build automated recognition software to identify patterns in digital tissue images (i.e. whether or not an image of H&E-stained histology contains cancer, and if so, how aggressive the cancer is), we must first ``train'' the computer to recognize the patterns of interest. This is done by manually labeling the images according to the structures we're looking for -- cells, glands, tissue regions, and so on.

In this project, the student will learn how to identify structures of interest on biological images, particularly in digital pathology. The data generated by the student will be used to train computer vision algorithms to identify these structures on testing samples, and will be used on future studies in classification and analysis of high-dimensional, multi-modal patient data.

Depending on interest and motivation, students may design algorithms to extract features from these datasets, write programs to process and analyze images, and contribute to articles for publication in high-impact conferences and journal articles. The student will also get exposure to clinical scientists, experimental pathologists, and faculty members in a wide variety of fields. This work can easily be the foundation for a rewarding research project.

Research Project Information

Disciplines: Biomedical Engineering, Medical Image Analysis, Quantitative Histology, Machine Learning, Pattern Classification
Student Skill-Set Needed: Ability to work with image editing programs (e.g. Photoshop, ImageJ, Gimp, ImageScope, or ImageViewer); Ability to work independently, write progress reports, and participate in lab meetings; Programming ability in MATLAB or Python is useful but not requi
Compensation: Academic Credit, Volunteer, Work Study
Available: Fall, Spring, Summer


For further information on this opportunity, or to apply, contact:

Faculty Member: Scott Doyle
Title: Assistant Professor
Department: Pathology And Anatomical Sciences
Office: 208 Farber Hall, 3435 Main Street, Buffalo, NY 14214
Phone: 716-829-2005