Converting 2d points to 3d world coordinates and getting measurements using AR

Question

I'm developing a medical app that measures patients' wounds using computer vision for getting the wound 2d coordinates and AR for getting the depth by performing a hittest using the camera.

Lets say that my hitest result is

 {
  "type": "FeaturePoint",
  "transform": {
    "rotation": [0, 0, 0],
    "position": [0.34072238206863403, -0.017041677609086037, 0.09095178544521332],
    "scale": [1, 1, 1]
  }
}

and the inference returned by my computer vision model is

{
  "predictions": [
    {
      "x": 165.5,
      "y": 209.5,
      "width": 83,
      "height": 53,
      "confidence": 0.884,
      "class": "wounds",
      "points": [
        {
          "x": 140,
          "y": 182.809
        },
        {
          "x": 139.5,
          "y": 183.477
        },
        {
          "x": 138.5,
          "y": 183.477
        },
        {
          "x": 138,
          "y": 184.144
        },
        {
          "x": 137,
          "y": 184.144
        },
        {
          "x": 136.5,
          "y": 184.811
        },
        {
          "x": 136,
          "y": 184.811
        },
        {
          "x": 135,
          "y": 186.145
        },
        {
          "x": 134,
          "y": 186.145
        },
        {
          "x": 133.5,
          "y": 186.813
        },
        {
          "x": 133,
          "y": 186.813
        },
        {
          "x": 132.5,
          "y": 187.48
        },
        {
          "x": 132,
          "y": 187.48
        }
.....
}]

Can I use back projection equation,

that is, for each point 2d point (X,Y)
X = (x-cx)*Z/Fx
Y = (y-cy)*Z/Fy

where,
cx,cy =  Principal point (image center).
fx,fy = focal length in pixels
Z = depth from the hittest

to map the 2d coordinates from the computer vision model to 3d real-world points? If not, what else can I do to measure the wound accurately?

I would greatly appreciate any guidance that can point me in the right direction. Thank you!