Patent Application
20170109922
Image Processing and Radiology.
In cases of suspected small bowel obstruction, it is desirable for the radiologist or another physician to quickly and easily find the abrupt transition point corresponding to the point of obstruction, between the dilated upstream bowel and the decompressed downstream bowel. Pre-existing methods of visual inspection are relatively time-consuming and can present varying degrees of difficulty, in part due to the presence of multiple air-fluid, air-stool, or air-PO-contrast levels within the lumen of convoluted bowel. Similar difficulties also slow down the visual evaluation of bowel in the context of determining post-surgical bowel course (such as determining whether Roux-en-Y, Billroth, or another procedure has been preformed), and in the context of suspected internal hernia.
While attempting to trace the bowel length for the above purposes, the radiologist must presently visually follow the bowel as its internal contents change multiple times from black (corresponding to air on CT, which floats up into some bowel loops), to gray-white (fluid or oral contrast in loops of the bowel that are relatively lower). Although visualization of air-fluid levels is desirable during initial evaluation, presence of these black-white transitions and resulting inherent perceptual edges within the bowel lumen slow down the process of tracking along the bowel length and make the process of finding a transition point unnecessarily difficult.
Pre-existing flood-fill and thresholding algorithms capable of converting large regions of digital images to another color have been widely used in the fields of graphic processing, but not in combination with radiology PACS system for the purpose of gastrointestinal tract evaluation.
The invention changes the heterogeneous appearance of the gastrointestinal tract lumen to a nearly-uniform appearance having similar or same color or gray-level throughout the bowel course, such as by means of converting appearance of air to resemble that of orally-administered contrast material or intrinsic bowel fluid, or by means of performing the the reverse conversion (such as changing voxels corresponding to oral contrast to resemble air). The invention consists of novel application of the pre-existing and widely used flood-fill and thresholding algorithms specifically applied to PACS and radiologic images for the purpose of gastrointestinal tract evaluation, wherein the flood-fill algorithm or thresholding algorithm is combined with a PACS display system (picture archiving and communication system), radiologic work station, or an intermediate processing computer, and provides the means to perform the following functions to the radiologist or another operator. The operator can specify a range of voxel intensities to be replaced with another color or shade such as by selecting a group of one or more voxels from the image or a palette, or means of utilizing an implicit pre-determined range of voxel intensities (such as a range of −600 to −2000 Hounsfields units which are assumed to represent air), thereafter referred to as “canvas substance.” The operator can select a desired new color or shade (thereafter referred to as “paint substance”), such as by clicking on another area of the image (example: fluid within bowel lumen), or selecting from a palette. Flood-fill or thresholding algorithm then processes the set of 3D or 2D radiologic images to cause all or subset of other voxels which have a similar intensity to the “canvas substance” to be changed to another, selectable color or grayscale-level of “paint substance”. This process makes the bowel easier to visually follow by causing the bowel contents to be nearly uniform in appearance.
Those skilled in the art of computer programming and radiologic image processing are able to implement the invention based on the description within the “Summary of Invention” and example herein, using pre-existing descriptions of flood-fill algorithms.
The following example is not meant to define the invention's scope, but merely serves to illustrate a sample implementation and demonstrate its usage:
