Towards Real-time Intrahepatic Vessel Identification in Intraoperative Ultrasound-Guided Liver Surgery

TL;DR

This work tackles the difficulty of real-time intrahepatic vessel identification during laparoscopic ultrasound-guided liver surgery by proposing a patient-specific AI pipeline that trains on preoperative 3D US liver volumes to identify portal branches in intraoperative US. It employs a six-step workflow including tracked US acquisition, 3D volume reconstruction, data augmentation, and a personalized Attention U‑Net trained on synthetic 2D US slices, with real-time deployment in 3D Slicer. Validation on ex vivo porcine livers shows a real-time inference speed of about 0.072 s per frame and a mean Dice score around 0.60, outperforming a brute-force baseline and demonstrating the approach’s potential to enhance intraoperative guidance. Preclinical validation with surgeons indicates the AI tool can achieve higher precision and recall in portal-branch identifications, suggesting meaningful clinical impact and a favorable learning curve for adoption in liver cancer surgery.

Abstract

While laparoscopic liver resection is less prone to complications and maintains patient outcomes compared to traditional open surgery, its complexity hinders widespread adoption due to challenges in representing the liver's internal structure. Laparoscopic intraoperative ultrasound offers efficient, cost-effective and radiation-free guidance. Our objective is to aid physicians in identifying internal liver structures using laparoscopic intraoperative ultrasound. We propose a patient-specific approach using preoperative 3D ultrasound liver volume to train a deep learning model for real-time identification of portal tree and branch structures. Our personalized AI model, validated on ex vivo swine livers, achieved superior precision (0.95) and recall (0.93) compared to surgeons, laying groundwork for precise vessel identification in ultrasound-based liver resection. Its adaptability and potential clinical impact promise to advance surgical interventions and improve patient care.

Figures (3)

• Figure 1: Overview of the proposed method for real-time portal branch identification in the liver. The method comprises 6 steps: 1) Electromagnetic (EM) tracked US acquisition, 2) 3D US volume reconstruction, 3) Semantic segmentation of the reconstructed volume, 4) Data augmentation to create a personalized dataset for diverse US scanning protocols, 5) Personalized model training using an Attention U-Net (A-UNet) architecture oktay2018attention, and 6) Real-time deployment of the model using 3D Slicer fedorov20123d.
• Figure 2: Validation experiment comparing ground truth, model prediction, and brute force method (model 2). (a) Ground truth projection on a US frame, model prediction, and brute force method overlay. (b) Overlay of AI model predictions (opaque) and 3D segmentation ground truth (transparent) from reconstructed 3D US volume.
• Figure 3: Comparison model architectures for the identification of intrahepatic vascular structures and the effect of data augmentation