Annotation-efficient deep learning detection and measurement of mediastinal lymph nodes in CT - Takeaways - MDSpire

Annotation-efficient deep learning detection and measurement of mediastinal lymph nodes in CT

  • By

  • Alon Olesinski

  • Richard Lederman

  • Yusef Azraq

  • Jacob Sosna

  • Leo Joskowicz

  • September 13, 2025

  • 0 min

Share

  • 1

    Manual detection and measurement of lymph nodes in CT scans is time-consuming and prone to observer variability.

  • 2

    The SimU-Net pipeline offers a fully automatic solution for detecting and segmenting cancer lesions in CT scans.

  • 3

    The proposed semi-supervised method combines expert annotations with pseudolabeled data for lymph node analysis.

  • 4

    An anatomical filtering strategy is employed to reduce false positives in lymph node detection.

  • 5

    The method requires significantly less annotated data while achieving performance comparable to fully supervised approaches.

Original Source(s)

Annotation-efficient deep learning detection and measurement of mediastinal lymph nodes in CT

  • by Alon Olesinski, Richard Lederman, Yusef Azraq, Jacob Sosna, Leo Joskowicz

  • September 13, 2025

Related Content

Efficacy of interventions in peripheral arterial and venous malformations of the limbs in children: protocol for a systematic review

MimiVax and Roswell Park Open Phase 2 Trial of SurVaxM for Patients with Metastatic Neuroendocrine Tumors

Clinical study marks SurVaxM’s second phase 2 trial in a major cancer indication

  • by

  • May 7, 2026

  • 4 min
Partly intraoperative cell salvage in pediatric craniocerebral trauma: effects on coagulation function, allogeneic blood requirements, and clinical outcomes — a propensity score-matched retrospective cohort study