Search

EP-4742170-A1 - GENERATING A 2D IMAGE OF AN ANATOMICAL STRUCTURE FROM 3D MEDICAL DATA

EP4742170A1EP 4742170 A1EP4742170 A1EP 4742170A1EP-4742170-A1

Abstract

The present disclosure provides concepts for processing pixels of interest of a 3D volume of an anatomical structure to generate a 2D image of the anatomical structure. The method includes, for each of the pixels of interest: identifying one or more compound pixels in the vicinity of the respective pixel in the 3D volume based on one or more features of the anatomical structure included in the 3D volume; and determining an updated value of the respective pixel based on values of the compound pixels. The method further includes generating the 2D image based on the updated values of the pixels of interest. The method allows for adaptive compounding based on the shape of the anatomical structure included in the 3D volume, such that blurring and ghosting artifacts are reduced, while maintaining the benefits of compounding techniques (e.g., noise reduction).

Inventors

  • MORY, BENOIT
  • SOMPHONE, OUDOM
  • Attia, Emmanuel
  • FRADKIN, MAXIM

Assignees

  • Koninklijke Philips N.V.

Dates

Publication Date
20260513
Application Date
20241108

Claims (15)

  1. A method (100) for processing pixels of interest of a 3D volume of an anatomical structure to generate a 2D image of the anatomical structure, the method comprising: for each of the pixels of interest: identifying (120) one or more compound pixels in the vicinity of the respective pixel in the 3D volume based on one or more features of the anatomical structure included in the 3D volume; and determining (130) an updated value of the respective pixel based on values of the compound pixels; and generating (140) the 2D image based on the updated values of the pixels of interest.
  2. The method of claim 1, further comprising, for each of the pixels of interest, determining (122) a compounding direction associated with the respective pixel, the compounding direction based, at least in part, on a shape of the one or more features included in the 3D volume, and wherein identifying (120) the compound pixels is based on the compounding direction.
  3. The method of claim 2, further comprising, for each of the pixels of interest, identifying a gradient of a feature in the 3D volume in the vicinity of the respective pixel, and wherein determining (122) the compounding direction is further based on the identified gradient.
  4. The method of claim 2 or 3, further comprising, for each of the pixels of interest, identifying a projection direction at the location of the respective pixel, and wherein determining (122) the compounding direction is further based on the identified projection direction.
  5. The method of claim 4, wherein the projection direction is a direction perpendicular to the image plane of the 2D image at the location of the respective pixel.
  6. The method of any of claims 2-5, wherein identifying (124) the compound pixels based on the compounding direction comprises: identifying a ray in the 3D volume passing through the location of the respective pixel in the compounding direction, the ray having a predetermined length; identifying pixels in the 3D volume incident to the ray; and dentifying the compound pixels based on the identified pixels incident to the ray.
  7. The method of claim 6, further comprising determining the length of the ray based on the compounding direction and a gradient of an identified feature at the location of the respective pixel.
  8. The method of any of claims 2-7, wherein the compounding direction of each respective pixel is further based on the compounding direction of adjacent pixels of interest.
  9. The method of claim 8, further comprising a step of processing the compounding directions of each of the pixels with a regularization operator, to determine regularized compounding directions.
  10. The method of any of claims 1-9, wherein determining (130) the updated value of the respective pixel comprises processing (132), with a compound operator, the value of the respective pixel and the values of the compound pixels.
  11. The method of claim 10, wherein the compound operator is any one of an averaging operator, a weighted averaging operator, a minimum selector, or a maximum selector.
  12. The method of any of claims 1-11, further comprising, for each of the pixels of interest, identifying a location of the respective pixel in the 3D volume.
  13. The method of any of claims 1-12, further comprising a preceding step of identifying (110) the pixels of interest of the 3D medical image, each of the pixels of interest in a same image plane of the 3D medical image, and optionally wherein the image plane is based on a received indication of a slice of the 3D volume.
  14. A computer program comprising computer program code means adapted, when said computer program is run on a computer, to implement the method of any of claims 1-13.
  15. A system (200) for processing pixels of interest of a 3D volume of an anatomical structure to generate a 2D image of the anatomical structure, the system comprising: an analysis unit (220) configured, for each of the pixels of interest, to: identify one or more compound pixels in the vicinity of the respective pixel in the 3D volume based on one or more features of the anatomical structure included in the 3D volume; and determine an updated value of the respective pixel based on values of the compound pixels; and a graphics unit (230) configured to generate the 2D image based on the updated values of the pixels of interest.

Description

FIELD OF INVENTION The present invention relates to medical image processing, and more particularly to systems and methods for generating two-dimensional (2D) images from three-dimensional (3D) volumetric data of anatomical structures using adaptive compounding techniques. BACKGROUND OF INVENTION Three-dimensional (3D) medical imaging technologies, such as computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound, have revolutionized the field of medical diagnostics. These modalities provide detailed volumetric data of anatomical structures, allowing for comprehensive analysis and visualization. However, clinicians often need to examine specific two-dimensional (2D) slices or cross-sections of the 3D volume for diagnosis, treatment planning, and patient communication. Extracting 2D images from 3D volumetric data presents several challenges. Standard techniques for generating 2D slices, such as multi-planar reconstruction (MPR), often result in images with reduced quality compared to the original 3D data. Noise, artifacts, and loss of detail can occur during the extraction process, potentially impacting diagnostic accuracy. Additionally, when the desired 2D plane does not align with the native imaging orientation, the resulting image may suffer from reduced resolution and clarity. Compounding techniques, which combine multiple adjacent slices to form a single 2D image, have been developed to address some of these issues. While these methods can improve signal-to-noise ratio and enhance certain features, they often introduce new problems such as blurring, ghosting artifacts, and loss of spatial resolution. Thus, when performing compounding, there is a trade-off between noise reduction and image sharpness. Therefore, there is a present need for improved image processing techniques that reduce the noise present in 2D slices generated from 3D medical image data, whilst minimizing the presence of blur. SUMMARY OF INVENTION The invention is defined by the claims. According to an aspect of the present disclosure, a method for processing pixels of interest of a 3D volume of an anatomical structure to generate a 2D image of the anatomical structure is provided. The method comprises: for each of the pixels of interest, identifying one or more compound pixels in the vicinity of the respective pixel in the 3D volume based on one or more features of the anatomical structure included in the 3D volume; determining an updated value of the respective pixel based on values of the compound pixels. The method further comprises generating the 2D image based on the updated values of the pixels of interest. This method allows for adaptive compounding of pixels in the 3D volume, taking into account the underlying anatomical features. By identifying compound pixels based on the anatomical structure, the method can reduce blurring and ghosting artifacts typically associated with standard compounding techniques, while still benefiting from noise reduction. To be clear, the present invention provides concepts for generating high-quality two-dimensional (2D) images from three-dimensional (3D) volumetric data of anatomical structures using adaptive compounding techniques. The invention addresses the challenges associated with standard slice extraction and compounding methods, which often result in blurring, ghosting artifacts, and loss of spatial resolution. At the core of the invention is an adaptive compounding approach that takes into account the underlying features of the anatomical structure when processing pixels to generate a 2D image. For each pixel of interest in the desired 2D image plane, the method identifies compound pixels in the 3D volume based on local anatomical features. This approach allows for a more intelligent selection of pixels to be combined, reducing artifacts typically associated with standard compounding techniques. In existing techniques, pixels in the vicinity of each pixel of interest in the 3D volume are identified based on a compounding direction, for example. These pixels are then used to enhance the initial 2D image slice, to reduce the presence of artifacts, decrease noise, etc. This compounding direction may be the same for each of the pixels, and may not take into account the underlying anatomical features. Accordingly, the resultant image may include excessive levels of blur, particularly in regions on the edge/boundary of certain anatomical features. Accordingly, the present invention proposes to identify compounding pixels for updating values of the initial slice based on the underlying shape of the anatomical features. The proposed method offers several advantages over existing techniques. It provides improved image quality by reducing artifacts and enhancing feature preservation, potentially leading to more accurate diagnoses and improved patient care. The adaptive nature of the compounding process allows for better visualization of complex anatomical structures, which is partic