How to Register a Live onto a Liver ? Partial Matching in the Space of Varifolds (2204.05665v1)

Abstract: Partial shapes correspondences is a problem that often occurs in computer vision (occlusion, evolution in time...). In medical imaging, data may come from different modalities and be acquired under different conditions which leads to variations in shapes and topologies. In this paper we use an asymmetric data dissimilarity term applicable to various geometric shapes like sets of curves or surfaces, assessing the embedding of a shape into another one without relying on correspondences. It is designed as a data attachment for the Large Deformation Diffeomorphic Metric Mapping (LDDMM) framework, allowing to compute a meaningful deformation of one shape onto a subset of the other. We refine it in order to control the resulting non-rigid deformations and provide consistent deformations of the shapes along with their ambient space. We show that partial matching can be used for robust multi-modal liver registration between a Computed Tomography (CT) volume and a Cone Beam Computed Tomography (CBCT) volume. The 3D imaging of the patient CBCT at point of care that we call live is truncated while the CT pre-intervention provides a full visualization of the liver. The proposed method allows the truncated surfaces from CBCT to be aligned non-rigidly, yet realistically, with surfaces from CT with an average distance of 2.6mm(+/- 2.2). The generated deformations extend consistently to the liver volume, and are evaluated on points of interest for the physicians, with an average distance of 5.8mm (+/- 2.7) for vessels bifurcations and 5.13mm (+/- 2.5) for tumors landmarks. Such multi-modality volumes registrations would help the physicians in the perspective of navigating their tools in the patient's anatomy to locate structures that are hardly visible in the CBCT used during their procedures. Our code is available at https://github.com/plantonsanti/PartialMatchingVarifolds.

• The paper introduces an asymmetric dissimilarity measure integrated with the LDDMM framework for partial liver registration in multi-modal imaging.
• The method robustly aligns intraoperative CBCT liver surfaces to complete CT volumes, achieving an average surface distance of 2.6mm.
• This approach enhances surgical precision by propagating deformations to map internal structures, including vessel bifurcations and tumor landmarks.

This paper tackles the pressing issue of partial shape correspondences prevalent in computer vision and medical imaging, particularly within the context of multi-modality medical imaging datasets. When processing data from different imaging modalities, such as CT and CBCT scans, variations in shapes and topologies due to different acquisition conditions pose significant challenges.

The researchers propose a novel approach that leverages an asymmetric data dissimilarity term tailored for various geometric constructs, including sets of curves and surfaces. This method evaluates the embedding of one shape into another without depending on explicit correspondences between them. The key advantage of this approach lies in its integration within the Large Deformation Diffeomorphic Metric Mapping (LDDMM) framework, which is renowned for its ability to model complex non-rigid deformations.

Specifically, the paper details how this methodology can be employed for robust multi-modal liver registration. The authors focus on aligning a truncated CBCT volume, typically captured during surgical procedures (referred to as "live"), with a comprehensive pre-intervention CT volume. The CBCT data usually provide incomplete visual information, whereas the CT offers a full anatomical picture.

The proposed method enables the non-rigid yet realistic alignment of the partial CBCT liver surfaces with the complete CT surfaces, achieving an impressive average surface distance of 2.6mm (± 2.2mm). Beyond surface alignment, the resultant deformations are propagated through the liver's volume, facilitating the alignment of internal structures such as vessel bifurcations and tumor landmarks. Here, the authors report an average distance of 5.8mm (± 2.7mm) for vessel bifurcations and 5.13mm (± 2.5mm) for tumor landmarks.

This registration capability holds significant clinical potential. It would assist physicians in navigating surgical tools within the patient's anatomy, identifying critical structures that are not sufficiently visible in the intraoperative CBCT scans. This advancement could enhance the precision and safety of surgical procedures.

The authors have also contributed to the scientific community by sharing their code on GitHub, thereby facilitating further research and application development in this domain.