Prediction of tissue and clinical thrombectomy outcome in acute ischaemic stroke using deep learning - Summary - MDSpire

Prediction of tissue and clinical thrombectomy outcome in acute ischaemic stroke using deep learning

  • By

  • Marie-Sophie von Braun

  • Kristin Starke

  • Lucas Peter

  • Daniel Kürsten

  • Florian Welle

  • Hans Ralf Schneider

  • Max Wawrzyniak

  • Daniel P O Kaiser

  • Gordian Prasse

  • Cindy Richter

  • Elias Kellner

  • Marco Reisert

  • Julian Klingbeil

  • Anika Stockert

  • Karl-Titus Hoffmann

  • Gerik Scheuermann

  • Christina Gillmann

  • Dorothee Saur

  • January 18, 2025

  • 0 min

Share

Objective:

To develop a deep learning model that predicts individual responses to thrombectomy in acute ischemic stroke patients, enhancing personalized treatment strategies.

Key Findings:
  • Deep learning model achieved a mean Dice score of 0.48 on internal test data and 0.52 on external test data, outperforming baseline methods, indicating improved spatial accuracy in tissue outcome predictions.
  • Median absolute errors for NIHSS score predictions were 1.5 points (internal) and 3.0 points (external), demonstrating superior accuracy compared to other machine learning models and emphasizing the model's clinical relevance.
Interpretation:

The deep learning approach provides a more accurate and patient-specific assessment of potential outcomes from thrombectomy, significantly enhancing personalized treatment strategies and potentially improving patient care.

Limitations:
  • The study's reliance on a single dataset may limit generalizability; future studies should consider multi-center data to validate findings.
  • Potential biases in data collection and patient selection could affect outcomes; addressing these biases in future research is crucial.
Conclusion:

The proposed deep learning models represent a significant advancement in predicting individual benefits from thrombectomy, with implications for personalized stroke care and resource allocation, paving the way for future research in this area.

Original Source(s)

Prediction of tissue and clinical thrombectomy outcome in acute ischaemic stroke using deep learning

  • by Marie-Sophie von Braun, Kristin Starke, Lucas Peter, Daniel Kürsten, Florian Welle, Hans Ralf Schneider, Max Wawrzyniak, Daniel P O Kaiser, Gordian Prasse, Cindy Richter, Elias Kellner, Marco Reisert, Julian Klingbeil, Anika Stockert, Karl-Titus Hoffmann, Gerik Scheuermann, Christina Gillmann, Dorothee Saur

  • January 18, 2025

Related Content

Radiological predictors of shunt response in the diagnosis and treatment of idiopathic normal pressure hydrocephalus: a systematic review and meta-analysis

Diagnostic Performance of a Multiantigen Print ImmunoAssay (MAPIA) for Antibody Detection in Human Neurocysticercosis

Comparative analysis of GPT-4-based ChatGPT’s diagnostic performance with radiologists using real-world radiology reports of brain tumors