publishing calendar
click a month to filter the blog list.
- formatting
- images
- links
- math
- code
- blockquotes
- external-services
•
•
•
•
•
•
-
turning segmentation masks into 3d anatomical surfaces for navigation workflows
Why the step from mask to mesh is where a segmentation project starts becoming useful anatomy.
-
how to think about physics-informed learning without the hype
Why physics-informed learning is useful when it constrains the right thing, and annoying when it becomes branding.
-
why pathology ai is not just classification on patches
Why pathology models need context, scale, and careful evaluation before their scores mean very much.
-
what makes surgical computer vision different from standard vision benchmarks
Why surgical vision is not just object detection with more expensive images.
-
evaluation traps in biomedical ai: metrics that look good but say little
Why high scores in biomedical AI can still leave you with a model that explains very little and generalizes even less.
-
histopathology classification at multiple scales: why context changes everything
Why looking at pathology at more than one magnification can change the answer entirely.
-
neural anisotropic diffusion: unrolling a pde for medical image denoising
How an edge-aware PDE can be turned into a neural network that removes noise without destroying the anatomy that matters.
-
building a lung ct pipeline with monai, simpleitk, and vtk
Why putting segmentation, spatial preprocessing, and surface export in one pipeline is much more useful than training a model in isolation.
-
few-shot 2d echo to 3d cardiac reconstruction: what actually makes it hard
Why reconstructing a 3D heart from just a few echocardiography views is much more annoying than it sounds.
-
why deformation is the hard problem in image-guided robotic surgery
Why anatomy that moves, stretches, and bends turns guidance into a much harder vision problem than it first appears.