Enabling Augmented Segmentation and Registration in Ultrasound-Guided Spinal Surgery via Realistic Ultrasound Synthesis from Diagnostic CT Volume

TL;DR

An In silico bone US simulation framework that synthesizes realistic US images from diagnostic spinal CT volume is proposed that can achieve accurate and on-the-fly bone segmentation for spinal sonography and a Long-range Contrast Learning Module is proposed to fully explore the Long- range Contrast between the candidates and their surrounding pixels.

Abstract

This paper aims to tackle the issues on unavailable or insufficient clinical US data and meaningful annotation to enable bone segmentation and registration for US-guided spinal surgery. While the US is not a standard paradigm for spinal surgery, the scarcity of intra-operative clinical US data is an insurmountable bottleneck in training a neural network. Moreover, due to the characteristics of US imaging, it is difficult to clearly annotate bone surfaces which causes the trained neural network missing its attention to the details. Hence, we propose an In silico bone US simulation framework that synthesizes realistic US images from diagnostic CT volume. Afterward, using these simulated bone US we train a lightweight vision transformer model that can achieve accurate and on-the-fly bone segmentation for spinal sonography. In the validation experiments, the realistic US simulation was conducted by deriving from diagnostic spinal CT volume to facilitate a radiation-free US-guided pedicle screw placement procedure. When it is employed for training bone segmentation task, the Chamfer distance achieves 0.599mm; when it is applied for CT-US registration, the associated bone segmentation accuracy achieves 0.93 in Dice, and the registration accuracy based on the segmented point cloud is 0.13~3.37mm in a complication-free manner. While bone US images exhibit strong echoes at the medium interface, it may enable the model indistinguishable between thin interfaces and bone surfaces by simply relying on small neighborhood information. To overcome these shortcomings, we propose to utilize a Long-range Contrast Learning Module to fully explore the Long-range Contrast between the candidates and their surrounding pixels.

Figures (16)

• Figure 1: The figure shows examples of generated Ultrasound images which are the junctional surfaces of the two vertebrae, the plane where the vertebral plate could be entirely verified, the plane swept along the spinous, and an arbitrary scan plane, respectively.
• Figure 2: This figure illustrates the pressure simulation algorithm. As an enlarged part of the dashed box in the figure, the yellow arc represents the shape of the probe, and the area between the yellow arc and the green arc is the target space being squeezed into.
• Figure 3: Figure (a) shows the CT images after being processed by the pressure simulation, the skin and muscles are compressed into the probe's curved surface. Figure (b) shows the virtual US image produced with simulated extrusion. Figure (c) shows the reflection map after extrusion. Figure (d) shows that the processed image will not affect the propagation.
• Figure 4: The green part of the figure is the object to be scanned. Figure (a) shows the relationship between the entire feasible space (skin) and the scanned object(spine). Figure (b) shows the first type of movement, with the yellow dot representing the tip of the probe and the arrow representing the two-degree freedom of movement in the manifold. Figure (c) shows the second type of movement, the red line represents the discrete surface, the dashed arrow represents the probe can switch on different curves, and the arrow represents the probe can be moved along the curve direction.
• Figure 5: Illustration of propagation.