Non-rigid point cloud registration for middle ear diagnostics with endoscopic optical coherence tomography - Scorecard - MDSpire

Non-rigid point cloud registration for middle ear diagnostics with endoscopic optical coherence tomography

  • By

  • Peng Liu

  • Jonas Golde

  • Joseph Morgenstern

  • Sebastian Bodenstedt

  • Chenpan Li

  • Yujia Hu

  • Zhaoyu Chen

  • Edmund Koch

  • Marcus Neudert

  • Stefanie Speidel

  • June 17, 2023

  • 0 min

Share

Clinical Scorecard: Flexible point cloud alignment for diagnosing middle ear conditions using endoscopic optical coherence tomography

At a Glance

CategoryDetail
ConditionMiddle ear disorders including deformation, discontinuation of tympanic membrane and ossicle chain, effusion, cholesteatoma
Key MechanismsNon-invasive imaging of middle ear morphology and function via endoscopic OCT combined with neural network-based point cloud registration
Target PopulationPatients with middle ear disorders such as acute otitis media, chronic otitis media, otitis media with effusion, and trauma-induced conditions
Care SettingClinical otolaryngology settings utilizing endoscopic optical coherence tomography and advanced image analysis

Key Highlights

  • Endoscopic OCT enables high-resolution, depth-resolved, non-invasive imaging of middle ear structures in vivo but suffers from noise and incomplete data especially for deeper ossicles.
  • A novel two-stage neural network pipeline (C2P-Net) registers ex vivo micro-CT middle ear models to noisy, partial in vivo OCT point clouds for improved interpretation.
  • Synthetic data generation pipeline simulates realistic shape variants and noise to train the neural network for robust, non-rigid registration of middle ear point clouds.

Guideline-Based Recommendations

Diagnosis

  • Use endoscopic OCT to acquire high-resolution volumetric images of middle ear morphology and function.
  • Apply advanced point cloud registration techniques to align patient-specific OCT data with ex vivo micro-CT templates for enhanced structural interpretation.

Management

  • Integrate improved imaging interpretation to better localize and characterize middle ear pathology for targeted treatment planning.

Monitoring & Follow-up

  • Utilize repeated endoscopic OCT imaging combined with point cloud alignment to monitor morphological changes in middle ear disorders over time.

Risks

  • Be aware of limitations in OCT image quality due to noise and shadowing effects, particularly affecting visualization of deeper ossicles.

Patient & Prescribing Data

Patients undergoing middle ear evaluation for suspected or confirmed middle ear disorders

Enhanced imaging interpretation via point cloud alignment may improve diagnostic accuracy and guide appropriate clinical interventions.

Clinical Best Practices

  • Combine endoscopic OCT imaging with computational registration methods to overcome limitations of noisy and partial data.
  • Train neural networks on synthetic datasets simulating realistic anatomical variability and noise to improve generalizability to patient data.
  • Use non-rigid registration approaches to accommodate anatomical differences and deformations between template and patient-specific middle ear structures.

References

Original Source(s)

Related Content