Pixel operations

Bit-depth conversion (16→8)

Down-convert 16-bit pixel data to 8-bit for sharing or display.

Problem

Many DICOM images store pixel data at 16 bits per pixel (Bits Allocated = 16, Bits Stored = 12-16). While this preserves the full Hounsfield Unit range for diagnostic use, 8-bit consumers (web viewers, PNG export, some ML frameworks) require 8-bit data. The conversion reduces bit depth from 16 to 8 while applying the current window/level to maximize contrast in the relevant tissue range.

Steps

Open the file (⌘O).
Open Pixel Operations — ⌘K → “Pixel operations…”.
Select Bake W/L — this operation performs both window/level baking and bit depth conversion in one step.
Click Apply transform.
Verify:
- Bits Allocated (0028,0100) → 8
- Bits Stored (0028,0101) → follows Bits Allocated
- High Bit (0028,0102) → 7 (0-indexed for 8 bits)
- Photometric Interpretation (0028,0004) → MONOCHROME2

Result

All bit-depth tags now read 8-bit, the pixels are windowed to the current W/L before truncation, and the Modality LUT tags are removed — roughly halving the file size. This is the same operation as Window/Level baking, of which the depth change is a consequence.