omkarpshetty/abdomen-project ? reverse-engineered prompt

Build me a Python app that takes an abdominal CT DICOM study and turns it into a set of annotated slice images with the organs colored to match a reference legend.

I want it to be able to pull a series from an OsiriX export folder or a local DICOM folder, run an automatic segmentation model for the organs it can recognize, then generate one PNG per axial slice with the CT in grayscale and colored overlays on top. Please use the exact colors from the legend in color_map.py, and include a simple color key on the side unless I turn it off.

It should also handle the fact that some organs will not be available automatically, and clearly leave those for manual tracing later. Make it easy to adjust the CT window, opacity, and speed versus quality, and have a single command that runs the whole pipeline from input DICOMs to finished annotated images. If you need to check the latest docs for the segmentation model, look them up online first.