Efficient Stratified 3-D Scatterer Sampling for Freehand Ultrasound Simulation.

Area of Science:

• Medical Imaging
• Computational Ultrasound
• Image Simulation

Background:

• Ultrasound image simulation is crucial for developing computational imaging algorithms and training radiologists.
• Existing simulators typically involve scatterer generation, probe modeling, and RF signal generation.
• The first step, scatterer generation, requires high data volumes and computational time, especially for 3-D simulations.

Purpose of the Study:

• To explore novel scatterer spatial distributions for efficient ultrasound image simulation.
• To improve computational efficiency in generating realistic speckle patterns.
• To enable real-time ultrasound image generation from 3-D volumes.

Main Methods:

• Investigated pseudorandom scatterer distribution schemes using lazy evaluation.
• Developed a method for coherent, 3-D-aware scatterer generation.
• Integrated scatterer generation into an end-to-end simulation pipeline.

Main Results:

• Pseudorandom schemes demonstrated higher computational efficiency than uniform random distributions.
• The proposed method allows for multiple coherent 2-D slice simulations from 3-D volumes.
• Achieved real-time ultrasound image generation capabilities.

Conclusions:

• Lazy evaluation of pseudorandom schemes offers a computationally efficient alternative for scatterer generation in ultrasound simulation.
• The developed end-to-end method facilitates real-time, high-fidelity ultrasound image synthesis.
• This work advances the field of medical image simulation for both research and training applications.