Quantification of pulmonary involvement in COVID-19 pneumonia by means of a cascade of two U-nets: training and assessment on multiple datasets using different annotation criteria - Summary - MDSpire

Quantification of pulmonary involvement in COVID-19 pneumonia by means of a cascade of two U-nets: training and assessment on multiple datasets using different annotation criteria

  • By

  • Francesca Lizzi

  • Abramo Agosti

  • Francesca Brero

  • Raffaella Fiamma Cabini

  • Maria Evelina Fantacci

  • Silvia Figini

  • Alessandro Lascialfari

  • Francesco Laruina

  • Piernicola Oliva

  • Stefano Piffer

  • Ian Postuma

  • Lisa Rinaldi

  • Cinzia Talamonti

  • Alessandra Retico

  • October 26, 2021

  • 0 min

Share

Objective:

To develop a fully automated deep learning system for segmenting lung and COVID-19 lesions from CT scans, quantifying the lung volume affected by infection, and improving diagnostic accuracy.

Key Findings:
  • LungQuant effectively segments lung and COVID-19 lesions, providing quantitative analysis.
  • Segmentation performance is significantly influenced by the diversity of datasets and annotation styles.
  • Utilizing larger and more diverse datasets enhances the model's generalization ability.
Interpretation:

The LungQuant system demonstrates the potential of deep learning for automated segmentation in varied data environments, underscoring the critical role of dataset quality and annotation consistency.

Limitations:
  • The study faces challenges due to the limited availability of high-quality annotated data for COVID-19 lesions, which may affect the model's performance.
  • Variability in data acquisition protocols can lead to inconsistencies in segmentation accuracy.
  • Choices made during DICOM to NifTI conversion may compromise data quality, impacting the overall analysis.
Conclusion:

The study highlights the necessity for standardized annotation and data acquisition practices to improve the performance of deep learning models in medical imaging, paving the way for future advancements.

Original Source(s)

Quantification of pulmonary involvement in COVID-19 pneumonia by means of a cascade of two U-nets: training and assessment on multiple datasets using different annotation criteria

  • by Francesca Lizzi, Abramo Agosti, Francesca Brero, Raffaella Fiamma Cabini, Maria Evelina Fantacci, Silvia Figini, Alessandro Lascialfari, Francesco Laruina, Piernicola Oliva, Stefano Piffer, Ian Postuma, Lisa Rinaldi, Cinzia Talamonti, Alessandra Retico

  • October 26, 2021

Related Content

Volumetric Image Registration Techniques for Rigid and Nonrigid Models in Image-Guided Interventions

AI Scribes Lag Clinicians on Note Quality

A VHA study across 11 vendors finds AI-generated primary care notes score lower than clinician-written notes, with the largest deficits in thoroughness, organization, and usefulness

Rethinking the evidence for intensive surveillance after renal tumor ablation