By: Connor Welch
The UW-Madison Laboratory for Optical and Computational Instrumentation (LOCI) is a computational imaging instrumentation development group based in the Center for Quantitative Cell Imaging (CQCI) in the Office of the Vice Chancellor for Research. Their mission is to “observe and study cells in a more natural environment to get a more accurate understanding of the true cell behavior,” as stated by Professor Kevin Eliceiri, the co-lead of LOCI along with new Cell and Regenerative Biology Assistant Professor Abhishek Kumar. This is achieved using innovations in optical instruments as well as developing computational methods that can analyze the optical data and draw useful conclusions from it. A major focus of LOCI is developing and applying novel computational imaging technology to aid in histopathology, including weakly supervised learning for pathology image classification and label-free imaging studies of the role of microglia in Alzheimer’s disease.
“Everything is cell based.”
-Professor Kevin Eliceiri, Director of the Laboratory for Optical and Computational Instrumentation
Histology Improvement
Histology is the study of diseased tissues on a microscopic level, and, according to Eliceiri, it is “the one of the most widely used method of identifying disease” and traditionally boils down to “taking human tissue, staining it, and look at it with a well-trained eye.”
One of the main recent innovations in histology methodology has been to use computational methods to extract additional image features out of the slides and complement the pathology expert. LOCI uses conventional imaging pathology and new label free imaging that can extract additional information from histopathology.
Weakly Supervised Learning for Image Classification
The second focus area of study is weakly supervised image learning for image classification. Weakly supervised learning means that most of the data is not annotated. This can range from under 1% of the data to just over 20% of the data. For example, there might be a set of tumor images but only a small portion will be marked as cancerous or benign. Being able to learn in a weakly supervised manner is important because LOCI’s data comes in the form of videos; having someone classify every image would be labor intensive and time consuming. Furthermore, the images are complicated and difficult to process compared to normal image classification, according to Eliceiri.
This is why he says that “deep learning is a necessary part of the process” of weakly supervised image classification. This weak supervision technique is enabled through the development of ImageJ and PyImageJ by LOCI which are adaptable software that engage effectively in weakly supervised image classification using Java and Python, respectively.
Microglia in Alzheimer’s Disease
Despite billions of dollars and decades of research, Alzheimer’s is still not understood well enough for effective treatment or a cure. It is also a “cell-scale disease, so we need optical imaging to see it,” says Professor Eliceiri. The important part of microglia is that they “shift into a different state with Alzheimer’s.” Understanding why this shift occurs is important to study for future steps in the treatment of Alzheimer’s.
Optical imaging is an effective way to study this shift because it is a way of observing the microglia over time without potentially changing cell behavior as there are ways of monitoring microglia using intrinsic optical cues. One of the main goals of studying Alzheimer’s optically is to observe glycolysis, the breakdown of glucose, in the brain because a malfunction in glycolysis has been shown to be indicative of Alzheimer’s at both the clinical and pre-clinical stages.
In sum, progress in histology areas of research allow for a better understanding of disease behavior leading to more effective treatment. Improvement in weakly supervised learning for image classification can help with the early identification of diseases like Alzheimer’s resulting in better outcomes. Finally, a better understanding of the role of microglia in Alzheimer’s disease could lead to improved therapies targeting metabolism.