DICOM Notes - Boundary of CT Image

        Common CT images are effective CT values ​​within the effective range.

        The processing of CT values ​​is limited to linear processing of slope and intercept;

         Can participate in commonly used DICOM tag information:

DICOM Notes - Summary of DICOM Commonly Used Tags_dicom Tag List_Black Mountain Laoyao's Blog Blog-CSDN Blog File Introduction MetaInfoGroupElementTag Description Chinese explanation VR00020000File Meta Information Group LengthMeta Info length UL00020001File Meta Information VersionMeta Info version OB00020002Media Storage SOP Class UI DSOP Class The UID is the same as the SOP class UID in the Data Set https://blog.csdn.net/liushao1031177/article/details/117696381         If you encounter the following ct images with obvious boundaries, how can you distinguish which are valid CT values ​​and which is the background value?

         

          We can distinguish whether the image has been filled according to the internal value of 0028,0103. If the pixel representation is 0, the CT image has not been filled, and its internal pixel values ​​​​represent effective CT values; if the pixel representation is 1, the CT image There are two parts of CT value in the image, one part is the effective CT value, and the other part is the filling value, just to display the image as a rectangular image;

        详细内容可见dicom tag的说明： Pixel Padding Value Attribute – DICOM Standard BrowserSingle pixel value or one limit (inclusive) of a range of pixel values used in an image to pad to rectangular format or to signal background that may be suppressed. See Section C.7.5.1.1.2 for further explanation.https://dicom.innolitics.com/ciods/rt-plan/general-equipment/00280120

        In the data I encountered, 0028,0103 is 1, then there is a value of 0028,0120, and its value is -2000.

        Since the slope of the image is 1.0 and the intercept is -1024, -2000 will be -3024 after linear transformation;         

         Find a background value, you can see that the value is -3024; then the pixel with a pixel value of -3024 is the filling pixel, which can be filtered out in post-processing;