Robot-Assisted Prostatectomy: New Study Highlights Promising Deep Learning Strategies

Strategic Insights -

Researchers at the Polytechnic University of Bari in Italy have made significant strides in developing computer-assisted tools for evaluating surgical performance during robot-assisted radical prostatectomy. According to a new study, deep learning strategies have been employed to improve the accuracy of semantic segmentation in robot-assisted radical prostatectomy videos, paving the way for enhanced surgical outcomes.

Key Takeaways:

  • Robot-assisted radical prostatectomy (RARP) has become the most prevalent treatment for patients with organ-confined prostate cancer, but suboptimal vesicourethral anastomosis (VUA) may lead to serious complications, including urinary leakage, prolonged catheterization, and extended hospitalization.
  • A novel, annotated dataset collected from four VUA procedures was introduced to train deep learning models, with the nnU-Net 2D model achieving the highest class-specific metrics, including a Dice Score of 0.663 for the mucosa class and 0.866 for the needle class.
  • The research concluded that this work paves the way for computer-assisted tools that can objectively evaluate surgical performance during the critical phase of suturing tasks.
  • The two Deep Learning (DL) pipelines proposed and compared in the study included different architectures and training strategies to address the challenges of tissue distortions, changes in brightness, and instrument interferences in endoscopic videos.
  • The study involved researchers from the Department of Electrical and Information Engineering at the Polytechnic University of Bari, including Elena Sibilano, Claudia Delprete, Pietro Maria Marvulli, Antonio Brunetti, Francescomaria Marino, Giuseppe Lucarelli, Michele Battaglia, and Vitoantonio Bevilacqua.

Statistics:

  • The Dice Score for the mucosa class achieved by the nnU-Net 2D model was 0.663.
  • The Dice Score for the needle class achieved by the nnU-Net 2D model was 0.866.
  • The external validation video sequences used to evaluate the performance of the proposed deep learning models included four VUA procedures.
  • The study introduced a novel, annotated dataset collected from four VUA procedures to train deep learning models.

Sources:

  • "Deep Learning Strategies for Semantic Segmentation in Robot-Assisted Radical Prostatectomy." Applied Sciences 2025, 15(19): 10665. (Applied Sciences - http://www.mdpi.com/journal/applsci).
  • MDPI AG, publisher for Applied Sciences.
  • Polytechnic University of Bari, Department of Electrical and Information Engineering, Via Orabona 4, 70126 Bari, Italy.
  • NewsRx. Researchers at Polytechnic University of Bari Have Published New Data on Prostatectomy (Deep Learning Strategies for Semantic Segmentation in Robot-Assisted Radical Prostatectomy). Medical Devices & Surgical Technology Week. November 2, 2025; p 1682.