Search for: All records

Segmentation of multiple surfaces in optical coherence tomography (OCT) images is a challenging problem, further complicated by the frequent presence of weak boundaries, varying layer thicknesses, and mutual influence between adjacent surfaces. The traditional graph-based optimal surface segmentation method has proven its effectiveness with its ability to capture various surface priors in a uniform graph model. However, its efficacy heavily relies on handcrafted features that are used to define the surface cost for the “goodness” of a surface. Recently, deep learning (DL) is emerging as a powerful tool for medical image segmentation thanks to its superior feature learning capability. Unfortunately, due to the scarcity of training data in medical imaging, it is nontrivial for DL networks to implicitly learn the global structure of the target surfaces, including surface interactions. This study proposes to parameterize the surface cost functions in the graph model and leverage DL to learn those parameters. The multiple optimal surfaces are then simultaneously detected by minimizing the total surface cost while explicitly enforcing the mutual surface interaction constraints. The optimization problem is solved by the primal-dual interior-point method (IPM), which can be implemented by a layer of neural networks, enabling efficient end-to-end training of the whole network. Experiments on spectral-domain optical coherence tomography (SD-OCT) retinal layer segmentation demonstrated promising segmentation results with sub-pixel accuracy. 

Determining how ant colonies optimize foraging while mitigating pathogen and predator risks provides insight into how the ants have achieved ecological success. Ants must respond to changing resource conditions, but exploration comes at a cost of higher potential exposure to threats. Fungal infected cadavers surround the main foraging trails of the carpenter antCamponotus rufipes, offering a system to study how foragers behave given the persistent occurrence of disease threats. Studies on social insect foraging behavior typically require many hours of human labor due to the high density of individuals. To overcome this, we developed deep learning based computer vision algorithms to track foraging ants, frame-by-frame, from video footage shot under the natural conditions of a tropical forest floor at night. We found that most foragers walk in straight lines overlapping the same areas as other ants, but there is a subset of foragers with greater exploration. Consistency in walking behavior may protect most ants from infection, while foragers that explore unique portions of the trail may be more likely to encounter fungal spores implying a trade-off between resource discovery and risk avoidance.