ROI-based rendering for diagnostic image consistency
AssigneeCarestream Health, Inc.
InventorsZhimin Huo, Minjie Chen
Life of this patent
Estimating the term…
Abstract
A method for consistent rendering of two or more diagnostic images according to a region of interest. Image data as input code values are obtained for each of the two or more diagnostic images. The background image content is segmented from tissue image content for each of the two or more diagnostic images. Corresponding structures are identified within the tissue image content, and the region of interest is identified within the tissue image content for each of the two or more diagnostic images. The input code values are remapped to output code values for each of the two or more diagnostic images.
Claims24
A method carried out by a computer for providing consistent rendering of two or more diagnostic images obtained from substantially the same body tissue but taken at different times, according to a region of interest, the method comprising:
- obtaining image data as input code values for each of two or more diagnostic images;
- wherein each of the two or more diagnostic images (a) is an original image produced by a diagnostic imaging scan of a patient, (b) is different from any of the other diagnostic images, (c) is taken at different times, and (d) is of substantially the same body tissue of the patient;
- identifying at least one region of interest in each diagnostic image;
- identifying two points of from each region of interest;
- computing an output value of each of the two input points of interest based on an analysis of the two or more diagnostic images;
- and
- remapping the input code values for the region of interest to output code values for each of the two or more diagnostic images.
A method carried out by a computer for image segmentation of foreground image tissue content from background, the method comprising:
- a) obtaining an image histogram according to image data values;
- b) generating a difference histogram by subtracting a smoothed histogram from the image histogram;
- c) forming a bounded histogram by determining the histogram left and right bounds according to image activity and the difference histogram;
- d) classifying the image according to background content characteristics ascertained from the bounded histogram;
- into the following classes:
- Type I:
- no background;
- Type II:
- Background, but without a distinct background peak value;
- Type III:
- Background, having a background peak value and lower values between the background peak value and the values of pixels that provide the tissue image content;
- Type IV:
- Background, having a peak value and a higher “valley” between the background peak and tissue content, such that there is a clear separation between the background and the tissue content;
- e) calculating a segmentation threshold according to the classification obtained in step d) and characteristics of the bounded histogram near its left bound;
- and
- f) segmenting the image according to the segmentation threshold.
A method carried out by a computer for providing consistent rendering of two or more diagnostic images according to a region of interest, the method comprising:
- obtaining image data as input code values for each of the two or more diagnostic images;
- identifying corresponding structures within the tissue image content of the image data and identifying the region of interest within the tissue image content for each of the two or more diagnostic images;
- and
- remapping the input code values for the region of interest to output code values for each of the two or more diagnostic images with the steps of:
- calculating an upper remapping value that is conditioned according to relative spatial position of the region of interest within the diagnostic image content for each diagnostic image;
- calculating a lower remapping value that is conditioned by a normalization factor;
- and
- remapping the input code values for each pixel within the region of interest to a remapped value in the range defined between and including the upper remapping value and the lower remapping value.
A method carried out by a computer for providing consistent rendering of two or more diagnostic images according to a region of interest, comprising:
- obtaining image data as input code values for each of two or more diagnostic images;
- wherein each of the two or more diagnostic images (a) is an original image produced by a diagnostic imaging scan of a patient, (b) is different from any of the other diagnostic images, (c) is taken at different times, and (d) is of substantially the same body tissue of the patient;
- identifying corresponding structures within the tissue image content of the image data and identifying the region of interest within the tissue image content for each of the two or more diagnostic images;
- and
- remapping the input code values for the region of interest to output code values for each of the two or more diagnostic images including the step of calculating an upper remapping value that is conditioned according to relative spatial position of the region of interest within the diagnostic image content for each diagnostic image.
A method carried out by a computer for providing consistent rendering of two or more diagnostic images according to a region of interest, comprising:
- obtaining image data as input code values for each of two or more diagnostic images;
- wherein each of the two or more diagnostic images (a) is an original image produced by a diagnostic imaging scan of a patient, (b) is different from any of the other diagnostic images, (c) is taken at different times, and (d) is of substantially the same body tissue of the patient;
- identifying corresponding structures within the tissue image content of the image data and identifying the region of interest within the tissue image content for each of the two or more diagnostic images;
- and
- remapping the input code values for the region of interest to output code values for each of the two or more diagnostic including the step of calculating a lower remapping value that is conditioned by a normalization factor.
Dependencies are inferred from the claim text.