A Novel Framework for Integrating 3D Ultrasound into Percutaneous Liver Tumour Ablation
Shuwei Xing ∗, Derek W. Cool, David Tessier, Elvis C.S. Chen Terry M. Peters, Aaron Fenster · 2025
Read A Novel Framework for Integrating 3D Ultrasound into Percutaneous Liver Tumour Ablation by Shuwei Xing ∗, Derek W. Cool, David Tessier, Elvis C.S. Chen Terry M. Peters, Aaron Fenster (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-21162.txt.
Introduction
Shuwei Xing and co-authors present a new way to bring 3D ultrasound into liver tumor ablation, a treatment that kills tumors through the skin with heat. Their 2025 framework tackles a known weak point: tumors are hard to spot in ultrasound images, so doctors struggle to line them up with clearer CT or MRI scans taken earlier.
The team uses 3D ultrasound as a bridge to align the scans, plus a clear display that mixes the image types for quick checks. Their method lined up key points within about 2 to 4 millimeters, in a fifth of a second per pair. A flexible form of alignment cut error by about 40 percent.
Catalog record
| CREATOR | Shuwei Xing ∗, Derek W. Cool, David Tessier, Elvis C.S. Chen Terry M. Peters, Aaron Fenster |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 19 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-21162
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Novel Framework for Integrating 3D Ultrasound into Percutaneous Liver Tumour Ablation Shuwei Xing ∗, Derek W. Cool, David Tessier, Elvis C.S. Chen Terry M. Peters, Aaron Fenster Abstract 3D ultrasound (US) imaging has shown significant benefits in enhancing the outcomes of percutaneous liver tumour ablation. Its clinical integration is crucial for transitioning 3D US into the therapeutic domain. However, challenges of tumour identification in US images continue to hinder its broader adoption. In this work, we propose a novel framework for integrating 3D US into the standard ablation workflow. We present a key component, a clinically viable 2D US–CT/MRI registration approach, leveraging 3D US as an intermediary to reduce registration complexity. To facilitate efficient verification of the registration workflow, we also propose an intuitive multimodal image visualization technique. In our study, 2D US–CT/MRI registration achieved a landmark distance error of ∼ \sim 2–4 mm \mathrm{mm} with a runtime of 0.22 s 0.22\text{\,}\mathrm{s} per image pair. Additionally, non-rigid registration reduced the mean alignment error by ∼ \sim 40% compared to rigid registration. Results demonstrated the efficacy of the proposed 2D US–CT/MRI registration workflow. Our integration framework advanced the capabilities of 3D US imaging in improving percutaneous tumour ablation, demonstrating the potential to expand the therapeutic role of 3D US in clinical interventions. 1 Introduction Percutaneous tumour ablation, including radiofrequency and microwave ablation, is an effective minimally invasive treatment for early-stage liver cancer [ 13 , 21 ] . This technique involves inserting one or more ablation applicators into the tumour’s centroid, generating a thermal ablation zone around the applicator active tip to induce irreversible cellular death. Clinically, ultrasound (US) is widely used for applicator guidance due to its real-time imaging capabilities and broad accessibility. However, achieving accurate applicator placement requires physicians to mentally reconstruct 3D anatomical structures, such as tumours and surrounding vessels, from series of 2D US images. This cognitive burden, combined with the reliance on extensive experience and training, presents challenges in ensuring consistent outcomes in US-guided ablation. 3D US imaging has gained considerable attention for enhancing tumour ablation. Compared to conventional 2D US, it reduces operator dependence and enables volumetric measurement [ 6 ] . Recent work has demonstrated that 3D US imaging provides better tumour coverage than 2D US alone, as evidenced by corresponding patient outcomes [ 27 ] . For effective integration into the standard workflow, 3D US is recommended for assessing intra-procedural tumour coverage, while 2D US remains the primary real-time applicator guidance modality. Nevertheless, challenges in identifying tumours in US images continue to hinder the widespread adoption of this workflow. Specifically , certain types of liver tumours have low conspicuity or are nearly undetectable when imaging small lesions or those in difficult-to-image regions ( e.g . , the liver dome). Moreover, tumour mimics, such as regenerative nodules in cirrhotic livers and prior ablation sites [ 5 ] , complicate tumour identification and procedural accuracy. To address these limitations, image registration approaches that align 2D US with CT or MRI, have been investigated [ 7 , 25 , 19 ] . For example, the “Linear Correlation of Linear Combination” (LC2) similarity metric was developed to facilitate automatic US–CT [ 25 ] and US–MRI [ 7 ] registration. Extrinsic spatial trackers, including optical [ 20 ] and magnetic [ 18 ] , have been used to assist US–CT/MRI alignment, although they often fail to account for the respiratory motion during the intervention. While these approaches demonstrated significant technological advancements, they have not yet been standardised for integration into the ablation workflow due to their complexity and computational demands. Recently, deep learning techniques have been developed to facilitate these registration challenges. Markova et al . [ 15 ] proposed a computationally efficient end-to-end registration approach that aligns extracted keypoints from US and MRI liver images using an adapted LoFTR [ 23 ] network. Azampour et al . [ 1 ] developed a deep regression model for US probe pose estimation. In this framework, a US synthesis module was integrated to address the domain gap between US and CT images. However, existing learning approaches are unable to achieve clinically acceptable registration accuracy and remain far from clinical deployment. Inspired by a strategy employed in prostate interventions [ 17 , 29 , 9 ] , 2D US–CT/MRI registration can be divided into two sequential registration steps: “3D US–CT/MRI” and “2D–3D US”, reducing alignment complexity. For example, Xu et al . [ 29 ] demonstrated that transrectal US images can be fused with pre-procedural MRI for prostate biopsy using 3D US as an intermediary to align 2D transrectal US with MRI. This registration decomposition shows promise for integration into the standard workflow; however, its applicability to liver interventions has yet to be explored in depth. This paper presents a 3D US system integrated with registration and visualization for percutaneous liver tumour ablation. We advances 3D US imaging beyond its diagnostic role to therapeutic applications in liver cancer. Our contributions include 1) an effective 3D US integration framework compatible with the standard ablation workflow, 2) a clinically viable 2D US–CT/MRI registration workflow optimized for accuracy, runtime and robustness, and 3) an intuitive multimodal image visualization tool for validating intra-procedural alignment. 2 Methods: 3D US Integration Framework Figure 1: Framework for 3D US integration into 2D US guidance. (a) 3D US system and (b) intra-procedural guidance. The role of 3D US is highlighted in a brown block. (Regi.: Registration, Recon.: 3D US reconstruction) Fig. 1 a depicts the proposed mechatronic 3D US system, including a counterbalanced arm equipped with joint encoders, and a motor-driven 3D US scanner at its end. This device can be integrated with any commercial 2D US machine compatible with a liver-imaging US probe. The 3D US scanner enables a 2D US probe to perform automated movements, including tilting ( 60\text{\,}\mathrm{\SIUnitSymbolDegree} ), translation ( 60\text{\,}\mathrm{mm} ), and hybrid rotation-translation motion. The acquired US frames are then used to reconstruct a 3D US image within 7\text{\,}\mathrm{s}12\text{\,}\mathrm{s} . The integration framework of our 3D US system consists of a pre-procedural spatial calibration (Sec. 2.1 ) and an optimized intra-procedural US guidance workflow (Fig. 1 b). The standard ablation workflow is enhanced with US–CT/MRI registration, multimodal visualization, and tumour coverage assessment, as highlighted in the brown block. Given our previous work on tumour coverage assessment [ 27 ] , this work primarily focuses on 2D US–CT/MRI registration and visualization (Sec. 2.2 ). Figure 2: System calibration and visualization. (a) US probe calibration, and (b) different visualization approaches. (MVR: Multi-planar and Volume Rendering; MPR: Multi-planar Reformation) 2.1 System Calibration Mechatronic Arm Calibration This calibration determines the pose of the physical end of the US probe ( F_{end} ) relative to the base of the arm ( F_{1} ) (Fig. 2 a). Following Li et al . [ 14 ] , we first modelled the kinematics of the mechatronic arm using the Denavit-Hartenberg approach [ 4 ] , then employed a Levenberg-Marquardt nonlinear regression model [ 16 ] to compute the calibration parameters. Notably, to enable tracking of the probe end using the optical tracker (Polaris Spectra, Northern Digital Inc., Canada), we designed a US probe-mimicking phantom that incorporated a dynamic reference body. US Probe Calibration This step establishes the spatial relationship between