Semantic Rules for AI-ReadinessFor a truly robust spec sheet, I recommend adding these "Semantic Rules" to your logic:Pixel Spacing (0028,0030): Add a rule that Value 1 must equal Value 2 (Square Pixels). Most AI models are trained on square pixels. If a site uploads non-square pixels ($0.5\text{ mm} \times 0.6\text{ mm}$), your AI might produce distorted results.Slice Thickness (0018,0050) vs. Spacing Between Slices (0018,0088): * Your table lists Thickness. However, if there is a Gap between slices, the AI's 3D reconstruction will be "stretched."Recommendation: Add Spacing Between Slices to your MR Specific checks as Required for 3D/Volumetric AI.
For SEG files: Keep (0020,0052) Frame of Reference UID as Required for SEG.
Add SEG-specific checks (or a SEG profile section) that validates presence of:
Pixel Measures functional group
Plane Position (Patient) functional group
Plane Orientation (Patient) functional group
And that the SEG references source images/series.
Semantic Rules for AI-ReadinessFor a truly robust spec sheet, I recommend adding these "Semantic Rules" to your logic:Pixel Spacing (0028,0030): Add a rule that Value 1 must equal Value 2 (Square Pixels). Most AI models are trained on square pixels. If a site uploads non-square pixels ($0.5\text{ mm} \times 0.6\text{ mm}$ ), your AI might produce distorted results.Slice Thickness (0018,0050) vs. Spacing Between Slices (0018,0088): * Your table lists Thickness. However, if there is a Gap between slices, the AI's 3D reconstruction will be "stretched."Recommendation: Add Spacing Between Slices to your MR Specific checks as Required for 3D/Volumetric AI.
For SEG files: Keep (0020,0052) Frame of Reference UID as Required for SEG.
Add SEG-specific checks (or a SEG profile section) that validates presence of:
Pixel Measures functional group
Plane Position (Patient) functional group
Plane Orientation (Patient) functional group
And that the SEG references source images/series.