Model-based registration of preprocedure MR and intraprocedure US of the lumbar spine

Abstract

The joint framework benefits from the complementary features in both modalities, leading to significantly smaller TREs compared to a model-to-US registration approach. The s [Formula: see text] p [Formula: see text] s model also outperforms our previous shape [Formula: see text] pose model of the lumbar spine, as separating scale from pose allows to better capture pose and guarantees equally-sized vertebrae in both modalities. Furthermore, the simultaneous visualization of the patient-specific models on the MR and US domains makes it possible for clinicians to better evaluate the local registration accuracy.

Ultrasound-based navigated pedicle screw insertion without intraoperative radiation: feasibility study on porcine cadavers

“…So far, most of the work emphasizes on evaluating the registration quality using plastic phantoms [8], [9], [10], ex vivo spine anatomy [11], or animal cadavers including sheep [12], [13], [14] and pigs [15], [16], [17], [18], [19]. The few papers evaluating feasibility on clinical data have focused on needle guidance applications [20], [22], [22], [23]. Although Winter et al [24] evaluated their method on data collected on 5 patients undergoing surgery, their validation was limited to assessing the registration robustness.…”

Spinal pain relief procedures with the assistance of the MRI‐updated statistical shape model

“…Nonetheless, since MRI images visualize the soft tissues and nerves, most clinicians prefer to use this modality for diagnosing the cause of lumbar disorders. Behnami et al used the complementary features of preoperative MR and intraoperative US to benefit from the diverse features of different imaging modalities using a weighted joint registration algorithm, considering that US images of the spine can usually display the posterior bone surfaces while MRI images provide better viewing of the anterior structures . They reported a larger target registration error (TRE) value compared with CT‐US registration.…”

Registration of 3D freehand ultrasound to a bone model for orthopedic procedures of the forearm

“…In [16], Gaussian Mixture Models (GMMs) were used for modeling pelvic bone surfaces [17], where the CT surface was manually cropped to a region of interest prior to registration, and the US surface extracted using 3D local phase features. Statistical methods for the registration have also been proposed, e.g., using Unscented Kalman Filters [18,19], statistical shape models and GMMs of vertebrae [10], as well as statistical shape + pose + scale models of the lumbar spine [20]. But these require many observations to build model statistics and may not generalize well to other populations or to pathologies.…”