Scope: Recent advances in fluorescence microscopy have enabled unprecedented progress in cellular and developmental biology. Imaging has become a component of nearly every cell-biological investigation at all scales, from single molecules to whole tissues, and research modalities, from reconstitution experiments to genome-wide screening;
However, these fast-paced developments in imaging technology have remained unmatched by developments in image analysis software. Most of the published image data in biology are still processed by hand and interpreted by qualitative visual inspection.
Currently, there are very few curricula, in universities and national laboratories, dedicated to the mathematical, statistical, software, and machine learning methods required to transform raw microscopy image data into rigorous biological knowledge. This course is targeted to fill this void by training a next generation of life scientists in the mathematical foundation and implementation of algorithms for image analysis.
Brief synopsis: Topics covered in this course include: image enhancement, segmentation, tracking, feature extraction, image classification, machine learning, noise analysis and uncertainty prediction, and statistical hypothesis testing. All topics will be covered in theory lectures and computer exercises. Exercises will be done on images from ongoing research projects in the instructors labs and will target actual research questions. An important subject in the course will be software design, addressing both the implementation of optimized algorithms and sharable code, including programming in teams.
Student requirements and course structure: The course will admit graduate students and junior postdocs with backgrounds in mathematics and physics, who are currently conducting research in cellular and developmental biology. Students with no formal training in the quantitative sciences may also be considered if space is available and if initial experience in the practice of computer image analysis in microscopy is documented in the application. We will accept a maximum of 12 students. This size has been proven ideal in the first implementation of the course in 2011, and is also defined by the funds made available through an NIH grant. The course will be free for all admitted students, with the exception of travel expenditure to and from Woods Hole.
The course will contain two theory lectures per day. The rest of the course time will be primarily devoted to computer exercises under the guidance of the four course instructors and two TAs. To broaden the perspectives and encourage discussions, there will also be several evening seminars by invited faculty and by the students themselves.
The software projects will be done in small teams. This course structure is designed to train the students in the practicalities of code sharing and collaborative problem solving, allowing them to tackle complex image analysis problems when they return to their home institutions. At the end of the course, students will be encouraged to take home their own projects and the library code they integrated. The course will use MATLAB as the programming language, one of the most widespread platforms for scientific computing.
Importantly, this is not a course for students who wish to get familiar with MATLAB programming. On the contrary, we expect students to have basic knowledge of MATLAB (or solid expertise in another modular and/or object-oriented programming language), allowing them to start using Matlab on Day 1 for solving image and data analysis problems related to cell and developmental biology. Once the student body is identified, we will publicize tutorial materials for self-study.
Acknowledgment of funding: NIH/NIGMS R25 GM103792-01
Guest lectures and Course Dinner sponsored by: