Crop / ROI extraction

Crop pixel data to a region of interest and export the trimmed image.

Problem

You have a DICOM image with a large field of view (e.g. 512×512 whole-abdomen CT) but only need a specific region — a 100×100 pixel ROI around the liver for a focused analysis or AI model input. Cropping reduces file size, focuses analysis on the region of interest, and removes extraneous anatomy.

Steps

1. Open the file (⌘O) and record the original Image Position (Patient) (0020,0032) and Rows/Columns (0028,0010/0011) in Edit mode.

2. Open Pixel Operations — ⌘K → “Pixel operations…” or access from the top bar.

3. Select Crop from the pixel operation picker.

4. Enter ROI coordinates:

   • X (left): 20
   • Y (top): 20
   • Width: 100
   • Height: 100

5. Click Apply transform — the cropped image is derived and staged in memory.

6. Verify — filter for Rows (0028,0010) and Columns (0028,0011). Both should be 100. The Image Position (Patient) (0020,0032) is shifted to reflect the new spatial origin.

Validation

• Width and Height inputs enforce a minimum of 1. Entering 0 snaps back to 1 (zero-area crop is structurally impossible).
• The default crop dimensions are the full image dimensions (non-zero, pre-filled).

Expected Result

• Output image is 100×100 pixels.
• Image Position (Patient) is shifted to account for the crop offset.
• The output is a derived copy — original file is unchanged.
• SOP Instance UID is regenerated (2.25.* prefix).
• Zero-area dimensions are rejected (inputs clamp to min=1).