EP-4740176-A1 - PULMONARY TRACTOGRAPHY APPARATUSES AND METHODS

Abstract

A medical image processing device includes a display device; and an electronic processor programmed to perform a medical image processing method including receiving a three dimensional (3D) medical image; computing least one 3D derivative image of the 3D medical image; identifying a set of fibers in the 3D medical image by tracing fibrous image features in the at least one 3D derivative image starting from respective seed locations in the least one 3D derivative image; and controlling the display device to display an anatomical representation comprising or derived from the set of fibers.

Inventors

• WIEMKER, RAFAEL
• SCHMIDT-RICHBERG, Alexander
• WIEBERNEIT, Nataly
• BUIZZA, Roberto
• CAROLUS, Heike
• HAARTSEN, JAAP, ROGER
• HENDRIKS, Cornelis, Petrus
• KOEHLER, THOMAS
• NEMPONT, Olivier, Pierre
• POLKEY, Michael
• SABCZYNSKI, Jörg

Assignees

• Koninklijke Philips N.V.

Dates

Publication Date

20260513

Application Date

20240626

Claims (15)

1. 1. A medical image processing device, comprising: a display device; and an electronic processor programmed to perform a medical image processing method including: receiving a three dimensional (3D) medical image; computing least one 3D derivative image of the 3D medical image; identifying a set of fibers in the 3D medical image by tracing fibrous image features in the at least one 3D derivative image starting from respective seed locations in the least one 3D derivative image; and controlling the display device to display an anatomical representation comprising or derived from the set of fibers.
2. 2. The medical image processing device of claim 1, wherein the 3D medical image is a 3D pulmonary image of at least one lung, and the anatomical representation comprises a representation of vasculature of the at least one lung and/or a respiratory tract of the at least one lung.
3. 3. The medical image processing device of claim 2, wherein the anatomical representation comprises at least a representation of vasculature of a respiratory tract, and the medical image processing method further includes: classifying the fibers of the set of fibers as belonging to the vasculature or to the respiratory tract based on contrast of the corresponding fibrous image features in the 3D pulmonary image.
4. 4. The medical image processing device of claim 1 , wherein the anatomical representation comprises a two-dimensional anatomical sheet containing the set of fibers.
5. 5. The medical image processing device of claim 1, wherein the at least one 3D derivative image includes: a first 3D derivative image having local eigenvectors representing directions of the fibrous image features; and a second 3D derivative image derived from the first 3D derivative image and having local eigenvectors directed toward centerlines of the fibrous image features.
6. 6. The medical image processing device of claim 1, wherein the computing of the at least one 3D derivative image includes: computing a first 3D derivative image as a 3D second-order derivative image of the 3D medical image; and deriving a second 3D derivative image from the first 3D derivative image by performing an orthonormalization on the first 3D derivative image to determine orthogonal vectors that are orthogonal to the local eigenvectors of the first 3D derivative image and weighting the orthogonal vectors based on local eigenvalues of the first 3D derivative image corresponding to the local eigenvectors of the first 3D derivative image.
7. 7. The medical image processing device of claim 5, wherein the tracing of fibrous image features in the at least one 3D derivative image includes: tracking the fibrous image features along the local eigenvectors of the first 3D derivative image that represent directions of the fibrous image features; and biasing the tracking toward centerlines of the fibrous image features using the local eigenvectors of the second 3D derivative image.
8. 8. The medical image processing device of claim 5, wherein the first 3D derivative image is computed by operations including applying a 3D Hessian matrix to the 3D medical image.
9. 9. The medical image processing device of claim 5, wherein the first 3D derivative image is computed by operations including applying a 3D structure tensor matrix to the 3D medical image.
10. 10. The medical image processing device of claim 1, wherein the 3D medical image is a computed tomography (CT) image.
11. 11. The medical image processing device of claim 1 , wherein the 3D medical image is a grayscale magnetic resonance imaging (MRI) image.
12. 12. A medical image processing method, comprising: computing a three-dimensional (3D) derivative image of a 3D medical image having fibrous image features, the 3D derivative image having local eigenvectors representing directions of the fibrous image features; identifying a set of fibers in the 3D medical image by tracing the fibrous image features in the at least one 3D derivative image starting from respective seed locations in the least one 3D derivative image; and displaying an anatomical representation comprising or derived from the set of fibers.
13. 13. The medical image processing method of claim 12, wherein the 3D medical image is a 3D pulmonary image of at least one lung, and the anatomical representation comprises a representation of vasculature of the at least one lung and a respiratory tract of the at least one lung.
14. 14. The medical image processing method of claim 12, further comprising: deriving a second 3D derivative image from the first 3D derivative image, the second 3D derivative image having local eigenvectors directed toward centerlines of the fibrous image features.
15. 15. The medical image processing method of claim 14, wherein the tracing of fibrous image features in the first 3D derivative image includes: tracking the fibrous image features along the local eigenvectors of the first 3D derivative image that represent directions of the fibrous image features; and biasing the tracking toward centerlines of the fibrous image features using the local eigenvectors of the second 3D derivative image.

Description

PULMONARY TRACTOGRAPHY APPARATUSES AND METHODS CROSS-REFERENCE TO RELATED APPLICATIONS This patent application claims the priority benefit under 35 U.S. C. § 119(e) of U.S. Provisional Application No. 63/525,180, filed on July 6, 2023, the contents of which are herein incorporated by reference. [0001] The following relates generally to the medical imaging arts, pulmonary imaging arts, medical image processing arts, pulmonary image processing arts, medical imaging driven medical diagnostic and treatment guidance arts, medical imaging driven pulmonary diagnostic and mechanical ventilation therapy arts, and related arts. BACKGROUND [0002] A quantitative and spatially resolved assessment of the lung’s ventilation and perfusion is desired to optimize the ventilator settings and other therapy parameters in mechanically ventilated patients. Usually, the aim is to match ventilation and perfusion (aka. V- Q matching) as ventilation and perfusion are equally important for proper gas exchange in the alveoli. [0003] Ventilation and Perfusion (V+Q) modeling of the lung requires in particular the anatomical structures of airways, veins, arteries. While the inference of these structures from 3D- image volumes (e.g., by convolutional neural networks (CNNs)) is known to be technically feasible, the model building (i.e., explicit analytical modeling or implicit Machine Learning) and the necessary annotation of training data is tedious and expensive, due to the delicate, intertwined nature of these tree structures, which manifest weak and incomplete on clinical image volumes, and close to or below the resolution limit. [0004] Local appraisal on a voxel-by-level basis is not only time-consuming but often unfeasible; even for an expert annotator it is often ambiguous whether a certain voxel group is part of an airway or rather an image artefact, because an airway is filled by low attenuation air voxels, and surrounded by low attenuation lung parenchyma voxels, only separated by an airway wall which is typically thinner than the image resolution as well as voxel grid spacing, yielding a poor image contrast. [0005] In contrast to airways, pulmonary vessels (i.e., that are blood-filled) have a good contrast to the surrounding lung parenchyma. However, pulmonary vessels cannot be reliably distinguished locally between belonging to the arterial or the venous tree, both of which are intimately intertwined (and the flow direction is not manifesting in CT, which on the other hand is the spatially highest resolving modality). [0006] Furthermore, attempting to follow the course of airways/vessels, and appraising the destination locations as well as the anatomical plausibility of the course is challenging, at least because the depictions of all three tree structures in the image cannot be expected to be complete and connected, due to image resolution, image noise, imaging artifacts (streaks, beam hardening), as well as disease-caused disturbances (mucus, clots, remodeling, tumors, etc.). [0007] The following discloses certain improvements to overcome these problems and others. SUMMARY [0008] In one aspect, a medical image processing device includes a display device; and an electronic processor programmed to perform a medical image processing method including receiving a three dimensional (3D) medical image; computing least one 3D derivative image of the 3D medical image; identifying a set of fibers in the 3D medical image by tracing fibrous image features in the at least one 3D derivative image starting from respective seed locations in the least one 3D derivative image; and controlling the display device to display an anatomical representation comprising or derived from the set of fibers. [0009] In another aspect, a medical image processing method includes computing a 3D derivative image of a 3D medical image having fibrous image features, the 3D derivative image having local eigenvectors representing directions of the fibrous image features; identifying a set of fibers in the 3D medical image by tracing the fibrous image features in the at least one 3D derivative image starting from respective seed locations in the least one 3D derivative image; and displaying an anatomical representation comprising or derived from the set of fibers. [0010] One advantage resides in improved diagnostic value of medical images that include complex vasculature. [0011] Another advantage resides in improved diagnostic value of pulmonary images that include complex and intertwined vasculature and airways. [0012] Another advantage resides in providing an improved patient-specific image-based representation of the pulmonary ventilation and perfusion systems. [0013] A given embodiment may provide none, one, two, more, or all of the foregoing advantages, and/or may provide other advantages as will become apparent to one of ordinary skill in the art upon reading and understanding the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS [0014] The disclosure m