Optimizing Surgical Plans for Parenchyma-Sparing Liver Resections through Contour-Guided Resection and Surface Approximation

TL;DR

This study tackles the challenge of optimizing liver resections by introducing a contour-based, computer-aided planning pipeline that models curved, parenchyma-sparing resections (PSR) directly on the liver surface. It combines spherical and manual contour definitions, 3D elliptic Fourier analysis for contour smoothing, cubic B-spline contour filling, and bicubic Bézier surface fitting to generate realistic virtual resections, implemented as an open-source 3D Slicer extension. In a 14-case OSLO-CoMET dataset, PSR consistently reduced resected volume to $32.71\pm13.80$ mL versus $249.53\pm135.23$ mL for AR, while increasing remnant liver volume to $1922.77\pm442.86$ mL and remnant percentage to $98.16\pm0.81\%$ compared with $1716.87\pm403.00$ mL and $87.40\pm6.49\%$ for AR, with strong statistical significance ($p<0.0001$ for volumes and percentages). These results suggest that careful virtual planning with the proposed methods can optimize resection strategies and maximize healthy liver preservation, potentially enabling safer PSR implementations in clinical practice. The work provides a flexible, open, computer-aided framework for defining virtual resections that better reflect surgical paths and supports broader adoption through the Slicer-Liver ecosystem.

Abstract

Objective: This study introduces a novel method for defining virtual resections in liver cancer surgery, aimed at enhancing the adaptability of parenchyma-sparing resection (PSR) plans. By comparing these with traditional anatomical resection (AR) plans, we explore the potential for optimization in surgical planning. Methods: Leveraging contours and spline surface approximations directly from the liver's surface, our method aligns closely with actual surgical procedures, offering a more realistic representation of curved resection paths. This technique, tested against 14 cases from the OSLO-COMET study, incorporates surface deformation for versatile plan modeling, comparing volumetric outcomes of PSR and AR. Results: The study highlights significant benefits of PSR over AR, including reduced resected volume ($32.71 \pm 13.80$ ml for PSR vs. $249.53 \pm 135.23$ ml for AR, $p <0.0001$) and higher remnant liver volume ($1922.77 \pm 442.86$ ml for PSR vs. $1716.87 \pm 403.00$ ml for AR, $p <0.0001$). PSR also showed a considerably higher remnant percentage ($98.16 \pm 0.81%$) compared to AR ($87.40 \pm 6.49%$, $p <0.0001$). Conclusion: The proposed approach is able to define virtual resections accommodating a wide variety of resections (i.e., PSR and AR). Careful surgical planning using virtual resections can optimize the resection strategy. Significance: This study presents a novel computer-aided planning system for liver surgery, demonstrating its efficacy and flexibility for definition of virtual resections. Virtual surgery planning can be used for optimization of resection strategies leading to increased preservation of healthy tissue.

Figures (5)

• Figure 1: Visualization of the surgical resection line (A) alongside pre-operative representations using contour definition modeled with Spherical Contour (B) and Markups Closed Curve (C).
• Figure 2: Visualization of Harmonics through Elliptic Fourier Analysis on a 3D Closed Contour: The initial five diagrams illustrate the reconstructions for the first five harmonics, presented in green. In contrast, the final diagram showcases the 16th harmonic's contour reconstruction at the Nyquist frequency, also displayed in green. The original closed contour is depicted in orange.
• Figure 3: The figure depicts the progressive stages of our hepatic resection planning technique culminating in contour filling: (A) Initial contour delineation is performed using spherical extraction prior to the application of Elliptic Fourier Analysis (EFA); this stage includes artifacts (isolated regions generated by the liver's anatomical variations) that are subsequently eradicated and the contour refined through EFA as shown in (B); finally, (C) illustrates the implementation of the contour filling strategy.
• Figure 4: Comparison of Surgical Outcome Metrics in Patients Undergoing Parenchyma Sparing Resection (PSR) and Anatomical Resection (AR).
• Figure 5: Comparative visualization of hepatic resection strategies across three distinct patient cases: The first column presents three-dimensional reconstructions of the liver, vascular structures, and tumors for each patient. The second column illustrates the proposed resection surface (highlighted in white) and the corresponding parenchyma volume intended for resection (shown in purple) in the context of parenchyma-sparing hepatectomy (PSR). The third column depicts the resection surface (in green) and the resected liver parenchyma volume (in purple) for anatomical resection (AR).