C.7.6.23.1 Description of the Enhanced Blending and Display Pipeline

The Enhanced Blending and Display Pipeline describes a scheme for blending of data frames of different Data Types and a color/grayscale mapping for display recommended by the provider of this information. There are no requirements upon an receiving application to utilize this recommendation in the processing and display of the referenced image.

The blending transformation model in Figure C.7.6.23-1 applies for mapping images derived from one, two, or three data frames with the same image position and orientation but of different values of Data Type (0018,9808) to grayscale P-values or color PCS-values for presentation. These inputs to the pipeline are frames of pixel values obtained from Pixel Data (7FE0,0010) and structured as described by attributes of the Image Pixel Module. Co-located pixels from each data frame are processed through the Pipeline, resulting in one output sample at that location.

The model utilizes up to two data paths called the Primary and the Secondary data paths. Each input data frame may be initially processed by a Modality LUT and a VOI LUT. If not explicitly specified, the Modality LUT and VOI LUT are assumed to be identity transformations. The Primary path may be used alone to obtain only grayscale transformation without blending. In this case, the Primary data path input after going through the Modality LUT and VOI LUT is mapped through a Presentation LUT to obtain device independent grayscale values (P-values) for presentation.

Either Primary or Secondary paths alone or both paths together may be used to obtain color transformation with blending. Up to one data frame input in the Primary data path and up to two data frames input in the Secondary data path are allowed for this blending transformation. These frames may be mapped to color using one-input or two-input palette color lookup tables depending on the number of data frames input to the data path. Alternatively if a path has one data frame, values may be converted to RGB (where R=G=B) before blending, if pseudo-color presentation of this data path is not desired. RGB values from Primary and Secondary data paths are combined via a “Blending Operation” in which the RGB color components are multiplied by a corresponding blending weight function (Weight 1 and Weight 2, respectively), and for each color component the two products are added together to produce the blended output value of that component. All inputs to the Blending Operation are normalized to the range 0.0 to 1.0, inclusive, even if they are fixed integer values that had been the entries in a lookup table.

Each of the Weight inputs to the blending operation is the output of a corresponding Blending LUT Transfer Function, which is selectable as either a constant or a derivation of the data frame values as described in C.7.6.23.4.

The RGB output from the Blending Operation is clamped to limit each color component (R, G, and B) to fall within the range 0.0 to 1.0, inclusive. The clamping function simply sets the value of any color component to 1.0 if the output from the Blending operation for that component exceeds 1.0. The RGB output of the clamping function is made available for use in rendering and slicing algorithms. Further, the output of the clamping function is transformed by the Profile Connection Space Transformation to device independent color values (PCS-values) for presentation.

If the input data frame values are representative of real world data, then they can be mapped to the real world values and units using the Real World Value Mapping Macro.

Note: PS3.17 “Enhanced US Data Type Blending Examples” describes a number of examples of the Enhanced Blending and Display Pipeline with specific attribute values for each example that invoke particular data flows through the pipeline.

[pic]

Figure C.7.6.23-1: Enhanced Blending and Display Pipeline