EP-4738269-A1 - METHODS AND SYSTEMS FOR 3-D RECONSTRUCTION AND VISUALIZATION OF ANATOMY USING 2-D IMAGE SLICES

EP4738269A1EP 4738269 A1EP4738269 A1EP 4738269A1EP-4738269-A1

Abstract

The disclosure relates generally to methods and systems for 3-D reconstruction and visualization of anatomy using 2-dimensional (2-D) image slices. Conventional techniques lack in modelling the internal structures of the anatomy where mesh models of the internal structures are not visible when outer surface is visualized. According to the present disclosure, the 3-D reconstruction and visualization of the anatomy using the 2-D image slices is done in three steps. In first step, the multi-directional image information of the anatomy is combined in the form of a point cloud and converted into a voxel volume. In the second step, a 3-D mesh surface is created for each layered component associated with the anatomy. In third step, the layered mesh surfaces of the reconstructed anatomy are cut and opened in multiple directions representing multiple components to represent artifacts such as covering portions and details internal to the outer wall of the anatomy.

Inventors

CHANDEL, Vivek
BHATIA, DIVYA MANOHARLAL
Kanakatte Gurumurthy, Aparna
Khandelwal, Sundeep
GHOSE, Avik
SINHA, ANIRUDDHA

Assignees

Tata Consultancy Services Limited

Dates

Publication Date: 20260506
Application Date: 20251021

Claims (15)

A processor-implemented method (200), comprising: acquiring, via one or more hardware processors, a set of 2-dimensional (2-D) image slices of an anatomy at each time-instance of one or more time-instances within an operating cycle of the anatomy, using one or more image acquisition techniques, wherein the set of 2-D image slices of the anatomy acquired at each time-instance comprises one or more multi-directional 2-D image slices (202); generating, via the one or more hardware processors, a point cloud associated with the set of 2-D image slices acquired at each time-instance, wherein the point cloud comprises a plurality of points representing a plurality of pixels associated with the set of 2-D image slices of the anatomy acquired at each of the one or more time-instances (204); creating, via the one or more hardware processors, a coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, from the point cloud associated with the set of 2-D image slices, using a 3-dimensional (3-D) interpolation technique (206); creating, via the one or more hardware processors, a 3-D mesh surface for each layered component of one or more layered components associated with the anatomy, using (i) the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, and (ii) an iso-surface tracing technique (208); and creating, via the one or more hardware processors, a structurally complete sliced view of the 3-D mesh surface of the anatomy, from the 3-D mesh surface of each layered component of the one or more layered components associated with the anatomy created for each time-instance, using a mesh manipulation and rendering technique (210).
The processor-implemented method (200) as claimed in claim 1, wherein the set of 2-D image slices of the anatomy is acquired in parallel from one or more predefined directions at each time-instance and at a predefined inter-slice distance in each of the one or more predefined directions.
The processor-implemented method (200) as claimed in claim 1, wherein each of the one or more multi-directional 2-D image slices comprises the plurality of pixels and each of the plurality of pixels comprises (i) a 3-dimensional (3-D) location with reference to a fixed coordinate system, and (ii) one or more brightness values.
The processor-implemented method (200) as claimed in claim 1, wherein creating the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, comprises: defining a bounding box in a 3-D space, using the point cloud associated with the set of 2-D image slices acquired at each time-instance (206a); creating a voxel space based on the bounding box associated with the set of 2-D image slices acquired at each time-instance using the 3-D interpolation technique, wherein the voxel space comprises a plurality of voxels and wherein each of the plurality of voxels is defined as a cube in the 3-D space with a side of a predefined length (206b); and creating the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, using the voxel space associated with the set of 2-D image slices acquired at each time-instance (206c).
The processor-implemented method (200) as claimed in claim 1, wherein creating the 3-D mesh surface for each layered component of the one or more layered components associated with the anatomy, comprises: identifying the one or more layered components associated with the anatomy, from the coherent voxel volume, using a predefined set of annotated 2-D slice images of the anatomy (208a); extracting a scattered set of locations in the 3-D space associated with each layered component of one or more layered components, from the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance (208b); and creating the 3-D mesh surface for each of the one or more layered components based on the scattered set of locations associated with each layered component, using the iso-surface tracing technique (208c).
The processor-implemented method (200) as claimed in claim 1, wherein creating the structurally complete sliced view of the 3-D mesh surface of the anatomy, comprises: slicing the 3-D mesh surface associated with each layered component of the one or more layered components associated with the anatomy, along one or more directions defined by a plane in the 3-D space, using the mesh manipulation and rendering technique, to obtain a sliced view of the 3-D mesh surface associated with each layered component (210a); extracting one or more covering portions within each layered component and between the one or more layered components associated with the anatomy, based on the sliced view of the 3-D mesh surface associated with each layered component (210b); and creating the meaningful sliced view of the 3-D mesh surface of the anatomy by rendering the one or more covering portions within each layered component and between the one or more layered components with a covering mesh using the mesh manipulation and rendering technique (210c).
A system (100), comprising: a memory (102) storing instructions; one or more input/output (I/O) interfaces (106); one or more hardware processors (104) coupled to the memory (102) via the one or more I/O interfaces (106), wherein the one or more hardware processors (104) are configured by the instructions to: acquire a set of 2-dimensional (2-D) image slices of an anatomy at each time-instance of one or more time-instances within an operating cycle of the anatomy, using one or more image acquisition techniques, wherein the set of 2-D image slices of the anatomy acquired at each time-instance comprises one or more multi-directional 2-D image slices; generate a point cloud associated with the set of 2-D image slices acquired at each time-instance, wherein the point cloud comprises a plurality of points representing a plurality of pixels associated with the set of 2-D image slices of the anatomy acquired at each of the one or more time-instances; create a coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, from the point cloud associated with the set of 2-D image slices, using a 3-dimensional (3-D) interpolation technique; create a 3-D mesh surface for each layered component of one or more layered components associated with the anatomy, using (i) the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, and (ii) an iso-surface tracing technique; and create a structurally complete sliced view of the 3-D mesh surface of the anatomy, from the 3-D mesh surface of each layered component of the one or more layered components associated with the anatomy created for each time-instance, using a mesh manipulation and rendering technique.
The system (100) as claimed in claim 7, wherein the one or more hardware processors (104) are configured to acquire the set of 2-D image slices of the anatomy in parallel from one or more predefined directions at each time-instance and at a predefined inter-slice distance in each of the one or more predefined directions.
The system (100) as claimed in claim 7, wherein each of the one or more multi-directional 2-D image slices comprises the plurality of pixels and each of the plurality of pixels comprises (i) a 3-dimensional (3-D) location with reference to a fixed coordinate system, and (ii) one or more brightness values.
The system (100) as claimed in claim 7, wherein the one or more hardware processors (104) are configured to create the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, by: defining a bounding box in a 3-D space, using the point cloud associated with the set of 2-D image slices acquired at each time-instance; creating a voxel space based on the bounding box associated with the set of 2-D image slices acquired at each time-instance using the 3-D interpolation technique, wherein the voxel space comprises a plurality of voxels and wherein each of the plurality of voxels is defined as a cube in the 3-D space with a side of a predefined length; and creating the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, using the voxel space associated with the set of 2-D image slices acquired at each time-instance.
The system (100) as claimed in claim 7, wherein the one or more hardware processors (104) are configured to create the 3-D mesh surface for each layered component of the one or more layered components associated with the anatomy, by: identifying the one or more layered components associated with the anatomy, from the coherent voxel volume, using a predefined set of annotated 2-D slice images of the anatomy; extracting a scattered set of locations in the 3-D space associated with each layered component of one or more layered components, from the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance; and creating the 3-D mesh surface for each of the one or more layered components based on the scattered set of locations associated with each layered component, using the iso-surface tracing technique.
The system (100) as claimed in claim 7, wherein the one or more hardware processors (104) are configured to create the structurally complete sliced view of the 3-D mesh surface of the anatomy, by: slicing the 3-D mesh surface associated with each layered component of the one or more layered components associated with the anatomy, along one or more directions defined by a plane in the 3-D space, using the mesh manipulation and rendering technique, to obtain a sliced view of the 3-D mesh surface associated with each layered component; extracting one or more covering portions within each layered component and between the one or more layered components associated with the anatomy, based on the sliced view of the 3-D mesh surface associated with each layered component; and creating the meaningful sliced view of the 3-D mesh surface of the anatomy by rendering the one or more covering portions within each layered component and between the one or more layered components with a covering mesh using the mesh manipulation and rendering technique.
One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause: acquiring, a set of 2-dimensional (2-D) image slices of an anatomy at each time-instance of one or more time-instances within an operating cycle of the anatomy, using one or more image acquisition techniques, wherein the set of 2-D image slices of the anatomy acquired at each time-instance comprises one or more multi-directional 2-D image slices; generating, a point cloud associated with the set of 2-D image slices acquired at each time-instance, wherein the point cloud comprises a plurality of points representing a plurality of pixels associated with the set of 2-D image slices of the anatomy acquired at each of the one or more time-instances; creating, a coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, from the point cloud associated with the set of 2-D image slices, using a 3-dimensional (3-D) interpolation technique; creating, a 3-D mesh surface for each layered component of one or more layered components associated with the anatomy, using (i) the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, and (ii) an iso-surface tracing technique; and creating, a structurally complete sliced view of the 3-D mesh surface of the anatomy, from the 3-D mesh surface of each layered component of the one or more layered components associated with the anatomy created for each time-instance, using a mesh manipulation and rendering technique.
The one or more non-transitory machine readable information storage mediums as claimed in claim 13, wherein the set of 2-D image slices of the anatomy is acquired in parallel from one or more predefined directions at each time-instance and at a predefined inter-slice distance in each of the one or more predefined directions.
The one or more non-transitory machine readable information storage mediums as claimed in claim 13, wherein each of the one or more multi-directional 2-D image slices comprises the plurality of pixels and each of the plurality of pixels comprises (i) a 3-dimensional (3-D) location with reference to a fixed coordinate system, and (ii) one or more brightness values.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS AND PRIORITY The present application claims priority from Indian provisional application no. 202421082881, filed on October 29, 2024. TECHNICAL FIELD The disclosure herein generally relates to 3-dimensional (3-D) model generation, and, more particularly, to methods and systems for 3-D reconstruction and visualization of anatomy using 2-dimensional (2-D) image slices. BACKGROUND Reconstructing a 3-dimensional (3-D) model of an anatomy such as heart, lungs, kidney, and dental arch, of a living being such as humans and animals, plays a vital role for better analytics and diagnostics. Making use of a set of 2-dimensional (2-D) image slices of the anatomy that are acquired directly from imaging techniques such as Magnetic resonance imaging (MRI) and computed tomography (CT) imaging is one of most popular techniques for reconstructing the 3-dimensional (3-D) model of an anatomy. Conventional techniques for reconstructing the 3-D model of the anatomy make use of the set of parallel 2-D image slices of the anatomy for generating a 3-D voxel volume and a 3-D mesh which is further used for reconstructing the 3-D model. But with limited number of 2-D image slices which are acquired from a single direction, the conventional techniques results in inaccurate and incomplete 3-D mesh generation and hence the 3-D model reconstructed may not be visualized with greater details of the anatomy. Further, the conventional techniques concern with creating the 3-D mesh representation of a voxel volume to provide external structure reflecting outer morphology of the Anatomy. The anatomy possesses greater details layered in the inner portion of the outer hull, which remain un-constructed for example, inner walls, presence of blood, etc. Hence the conventional techniques lack in modelling the internal structures of the anatomy due to the reason that mesh models of the internal structures are not visible when the outer surface is visualized. It's imperative for a proper diagnostic activity involving detecting abnormalities in the internal structures (inner walls, blood, etc.) of the anatomy, or other congenital defects to not only look at the outer surface, but also be able to cut open the 3-D model from various directions and be able to visualize the inner details (structures) including inner walls, blood etc., together with the outer hull as 3-D mesh surfaces. SUMMARY Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. In an aspect, a processor-implemented method for 3-D reconstruction and visualization of anatomy using 2-D image slices is provided. The method comprising: acquiring a set of 2-dimensional (2-D) image slices of an anatomy at each time-instance of one or more time-instances within an operating cycle of the anatomy, using one or more image acquisition techniques, wherein the set of 2-D image slices of the anatomy acquired at each time-instance comprises one or more multi-directional 2-D image slices; generating a point cloud associated with the set of 2-D image slices acquired at each of the one or more time-instances, wherein the point cloud comprises a plurality of points representing a plurality of pixels associated with the set of 2-D image slices of the anatomy acquired at each time-instance; creating a coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, from the point cloud associated with the set of 2-D image slices, using a 3-dimensional (3-D) interpolation technique; creating a 3-D mesh surface for each layered component of one or more layered components associated with the anatomy, using (i) the coherent voxel volume associated with the set of 2-D image slices acquired at each time-instance, and (ii) an iso-surface tracing technique; and creating a structurally complete sliced view of the 3-D mesh surface of the anatomy, from the 3-D mesh surface of each layered component of the one or more layered components associated with the anatomy created for each time-instance, using a mesh manipulation and rendering technique. In another aspect, a system for 3-D reconstruction and visualization of anatomy using 2-D image slices is provided. The system includes: a memory storing instructions; one or more Input/Output (I/O) interfaces; and one or more hardware processors coupled to the memory via the one or more I/O interfaces, wherein the one or more hardware processors are configured by the instructions to: acquire a set of 2-dimensional (2-D) image slices of an anatomy at each time-instance of one or more time-instances within an operating cycle of the anatomy, using one or more image acquisition techniques, wherein the set of 2-D image slices of the anatomy acquired at each time-instance comprises one or more multi-directional 2-D image slices; generate a point cloud associated with the set o