Abstract
Needle insertion using flexible bevel-tip needles is a common minimally invasive surgical technique for prostate cancer interventions. Flexible, asymmetric bevel-tip needles enable physicians for complex needle steering techniques to avoid sensitive anatomical structures during needle insertion. For accurate placement of the needle, predicting the trajectory of these needles intraoperatively would greatly reduce the need for frequently needle reinsertions, thus improving patient comfort and positive outcomes. However, predicting the trajectory of the needle during insertion is a complex task that has yet to be solved due to random needle–tissue interactions. In this article, we present and validate, for the first time, a hybrid deep learning and model-based approach to handle the intraoperative needle shape prediction problem through leveraging a validated Lie-group theoretic model for needle shape representation. Furthermore, we present a novel self-supervised learning (SSL) and method in conjunction with the Lie-group shape model for training these networks in the absence of data, enabling further refinement of these networks with transfer learning (TL). Needle shape prediction was performed in single-layer and double-layer homogeneous phantom tissue for C- and S-shaped needle insertions. Our method demonstrates an average root-mean-square prediction error of 1.03 mm over a dataset containing approximately 3000 prediction samples with the maximum prediction steps of 110 mm.
Funding Information
  • National Institutes of Health (R01CA235134)
  • Johns Hopkins University internal funds

This publication has 66 references indexed in Scilit: