Skip to content

Ultrasound Image Processing

1. get ultrasound image by screenshot to video

you need to find to geometry structure one for IJV systolic (IJV small), one for IJV diastolic (IJV large).

2. S1_ijv_model_find_tissue.py

labeling each tissue type (skin, fat, muscle, ijv, cca)

  1. for skin and fat, we find a boundary(straight line) to decide skin and fat.
  2. for ijv, using GUI to get the region of IJV
  3. for cca, we use circle equation to determine the region of cca

As the figure shown above, the from z=0 to green line is the skin area, from green line to blue line is fat area ... vice and versa.

Basically, how to run S1_ijv_model_find_tissue.py is according to blood_vessel_segmentation_line_new.json as shown below.

{
    "HW":{
        "20230903":{
            "bound": [3, 650],
            "length10mmEdge": [215, 445],
            "skin":{
                "__comment__": "axis-0 of imagee is x for line.",
                "x": 35
            },
            "fat":{
                "__comment__": "axis-0 of imagee is x for line.",
                "x":70
            },
            "IJVLarge":{
                "cca":{
                    "__comment__":"[x, y, radius]",
                    "v": [470, 500, 100]
                }
            },
            "IJVSmall":{
                "cca":{
                    "__comment__":"[x, y, radius]",
                    "v": [470, 500, 100]
                }
            }
        }
    }
}

** json file is a dictionary structure, if you don't know you can see the link https://tw.alphacamp.co/blog/json

** bound --> clip the image to what we want.

** length10mmEdge --> scale bar

S2_ijv_model_seg_construct.py

  1. make sure subject and date is synchronous with S1_ijv_model_find_tissue.py
  2. run the program

Then you will success to get the voxel file (*vol.npy)

S3_plot_3D_ijv_model.py

  1. make sure subject and date is synchronous with S1_ijv_model_find_tissue.py
  2. run the program

TODO :
oblique IJV model --> https://stackoverflow.com/questions/48818373/interpolate-between-two-images

https://stackoverflow.com/questions/59690451/how-to-turn-ct-segmentation-into-3d-model-in-python