Update README.md

README.md

@@ -22,6 +22,8 @@ The model is trained on the publicly available [TotalSegmentator dataset](https:
 available labels from TotalSegmentator. The classification labels were generated from the provided segmentation labels. 
 
 Note that the model expects one channel. If you create a multi-channel image using multiple CT windows, simply take the mean across channels.
+The model also expects 8-bit input (converted to float). Thus if your CT volume is in Hounsfield units, you can apply a standard window,
+such as soft tissue (level=50, width=400), before inputting it into the model. 
 
 ## Example Usage