**6. Ultrasound-based navigation accuracy**

The delicacy, precision and extent of the work the surgeon can perform based on image information rely on his/her confidence in the overall clinical accuracy and the anatomical or pathological representation. The overall clinical accuracy in image-guided surgery is the difference between the location of a surgical tool relative to some structure as indicated in the image information, and the location relative to the same structure in the patient. This accuracy is difficult to assess in a clinical setting due to the lack of fixed and well-defined landmarks inside the patient that can be accurately reached with a pointer. Common practice is therefore to estimate the system's overall accuracy in a controlled laboratory setting using precisely built phantoms. In order to conclude on the potential clinical accuracy, the differences between the clinical and the laboratory settings must be carefully examined.

#### **6.1. Error sources and key points**

navigation systems, where the tip of the tracked instrument determines the three extracted slices. The slices can also be orthogonal relative to the tracked instrument or the surgeon's view (i.e., oblique slicing relative to the volume axis or patient), and this is becoming an increasingly popular option in navigation systems. When a surgical tool cuts through multiple volumes several slices are generated. These slices can then be combined in different

**Figure 18.** Multimodal visualization. Orthogonal (A) and oblique (B) slicing, the position as well as the position and the orientation of the tool are used to extract the slices respectively. The three basic visualization types are shown in each image. The head is volume rendered in a 3D view that also shows geometric representations of both the tool and slice indicators. Corresponding slices are shown in a 2D view at the right. C) Display during freehand 3D ultrasound acquisition: Real-time 2D ultrasound to the left and an indication of the us-scanplane relative to MR data in a 3D and 2D view to the top and bottom right respectively. D) Overview of probe relative to head. E) Detailed view of real-time 2D ultrasound relative to MRA (read) and 3D power Doppler data (gray). F) Slice from ultrasound (top part) and MR (bottom part), surface model in red from MR (middle part). Mismatch between US (slice) and MR (tumor model) is clearly visible. G) 3D ultrasound (gray) is used to correct MRA (moved from red to green position) during an aneurysm

**•** *Direct volume rendering*: Volume- and geometric rendering techniques are not easily distin‐ guished. Often the two approaches can produce similar results, and in some cases one approach may be considered both a volume rendering and a geometric rendering technique. Still, the term volume rendering is used to describe a direct rendering process applied to 3D data where information exists throughout a 3D space instead of simply on 2D surfaces defined in (and often extracted from) such a 3D space. The two most common approaches to volume rendering are volumetric ray casting and 2D/3D texture mapping (figure 17 A, B, D, E, G). In ray casting, each pixel in the image is determined by sending a ray into the volume and evaluating the voxel data encountered along the ray using a specified ray function (maximum, isovalue, compositing). Using 2D texture mapping, polygons are generated along the axis of the volume that is most closely aligned with the viewing direction. The data is then mapped onto these quads and projected into a picture using

ways using various overlay and fusion techniques.

50 Advancements and Breakthroughs in Ultrasound Imaging

standard graphics hardware.

operation.

A comprehensive analysis of the error sources involved in neuronavigation based on intrao‐ perative ultrasound as well as preoperative MRI can be found in Lindseth et al. [97]. The overall accuracy is often referred to as the Navigation System Accuracy (NSA) and the essential points to remember can be summarized like this:

**•** The accuracy associated with navigation based on pre.op. MR/CT is independent of the accuracy associated with navigation based on intraoperative ultrasound, and vice versa.

In contrast, probe calibration, the major error source associated with ultrasound-based navigation, is included in the NSA resulting from accuracy evaluations using a rigid phantom in a laboratory setting. Furthermore, the surgeon is in control of the amount of tissue shift and deformation that is acceptable in a particular clinical case. A new scan can be acquired whenever needed in order to navigate using an updated image map (se figure 19, green line). As a consequence, the NSA found in a controlled laboratory setting will also be valid in the clinical case given that navigation is based on a recently acquired ultrasound scan (real-time 3D ultrasound being the extreme case) and that the speed of sound used in the ultrasound

Ultrasound-Based Guidance and Therapy http://dx.doi.org/10.5772/55884 53

A common mistake is to interpret a mismatch between MR/CT and Ultrasound in correspond‐ ing or fused displays as tissue shift. An observed mismatch between MR/CT and Ultrasound can only be interpreted as brain shift if 1a) navigation based on pre.op. data is accurate in the rigid case, 1b) the NSA, after the patient registration process, has been verified to be low, 2a) the NSA of ultrasound-based navigation in a controlled setting is low and 2b) the ultrasound

Preoperative MR/CT data can be "corrected" for brain shift using intraoperative ultrasound and advanced image-to-image registration techniques [85] as can be seen in figure 19. However this is a challenging task introducing additional error sources. Therefore the NSA associated with corrected preoperative MR/CT will not be as good as the NSA for ultrasound (see figure 19, blue lines). In addition, the independence between the NSA based on MR/CT and Ultra‐

The overall clinical accuracy of a navigation system will be determined by the contribution from all the individual error sources involved [97]. The net effect will not be the sum of all the error sources, but rather a stochastic contribution from all the terms. Stochastically independ‐

As stated previously, the most important parameter for the surgeon is the overall clinical Navigation System Accuracy (NSA). Although this parameter is difficult to assess, we believe that for ultrasound-based navigation an estimate can be made, based on a comprehensive laboratory evaluation and a thorough understanding of the significant additional error sources that occur in the clinical setting. Table 1 summarizes how such a calculation can be carried out assuming that a comprehensive evaluation of the system gives a NSA below 1.4 mm in a controlled laboratory setting. The error sources are assumed to be stochastically independent

data shown originate from an ultrasound volume that has recently been acquired.

sound will be broken (NSA for MR/CT will be dependent on NSA for Ultrasound).

ent contributions are summed using the following equation: ∑ (...)2

so that their contributions can be added on a sum-of-squares basis.

NSA using a phantom in the lab < 1.4 mm + Calibration and position tracking of rigid surgical tool < 0.5 mm + Interpolation of a 2D slice from a 3D volume / tool cross indication < 0.1 mm

**6.2. Clinical navigation system accuracy**

scanner corresponds to the average speed of sound in the tissue.


These points have major implications for the rational behind testing a navigation system in the lab using a phantom, and make a statement about the interesting parameter to the surgeon: the overall clinical navigation system accuracy. A lab test of a system based on preoperative MR/CT using a rigid phantom will give a very good navigation system accuracy (NSA<0.5mm, se figure 19, red line). Such a test will have limited validity in the general clinical situation, but is important to make sure that the system works as expected. The next phase in the evaluation of such a system would be to conduct a clinical study to investigate the system's ability to deal with a variety of different patient registration problems. Documenting that the system performs well in the rigid case and can deal in a satisfactory way with difficult patient registration cases is the best a system vendor can do. This does not give any information about the NSA experienced during a clinical case though. The surgeon must verify that the accuracy is acceptable after he has performed the patient registration procedure and anatomical landmarks inside the patient must be used to gain an impression about the amount of tissue shift and deformation. This shift and deformation makes systems based on preoperative MR/ CT of limited use during the procedure.

**Figure 19.** Navigation System Accuracy (NSA) based on preoperative (p) MR (red line) and intraoperative (i) US (green line). iUS can be used to correct pMR using various image-to-image registration techniques (blue line).

In contrast, probe calibration, the major error source associated with ultrasound-based navigation, is included in the NSA resulting from accuracy evaluations using a rigid phantom in a laboratory setting. Furthermore, the surgeon is in control of the amount of tissue shift and deformation that is acceptable in a particular clinical case. A new scan can be acquired whenever needed in order to navigate using an updated image map (se figure 19, green line). As a consequence, the NSA found in a controlled laboratory setting will also be valid in the clinical case given that navigation is based on a recently acquired ultrasound scan (real-time 3D ultrasound being the extreme case) and that the speed of sound used in the ultrasound scanner corresponds to the average speed of sound in the tissue.

A common mistake is to interpret a mismatch between MR/CT and Ultrasound in correspond‐ ing or fused displays as tissue shift. An observed mismatch between MR/CT and Ultrasound can only be interpreted as brain shift if 1a) navigation based on pre.op. data is accurate in the rigid case, 1b) the NSA, after the patient registration process, has been verified to be low, 2a) the NSA of ultrasound-based navigation in a controlled setting is low and 2b) the ultrasound data shown originate from an ultrasound volume that has recently been acquired.

Preoperative MR/CT data can be "corrected" for brain shift using intraoperative ultrasound and advanced image-to-image registration techniques [85] as can be seen in figure 19. However this is a challenging task introducing additional error sources. Therefore the NSA associated with corrected preoperative MR/CT will not be as good as the NSA for ultrasound (see figure 19, blue lines). In addition, the independence between the NSA based on MR/CT and Ultra‐ sound will be broken (NSA for MR/CT will be dependent on NSA for Ultrasound).

The overall clinical accuracy of a navigation system will be determined by the contribution from all the individual error sources involved [97]. The net effect will not be the sum of all the error sources, but rather a stochastic contribution from all the terms. Stochastically independ‐ ent contributions are summed using the following equation: ∑ (...)2

#### **6.2. Clinical navigation system accuracy**

**•** The accuracy associated with navigation based on pre.op. MR/CT is independent of the accuracy associated with navigation based on intraoperative ultrasound, and vice versa.

**•** The main error sources associated with preoperative MR/CT-based navigation are related to the patient registration process in a clinical setting, and the fact that the image maps are

**•** In contrast, intraoperative ultrasound volumes are acquired in the same coordinate system as navigation is performed. Patient registration is therefore not necessary, and a new ultrasound volume can be acquired to reflect the current patient anatomy whenever needed. However, navigation based on ultrasound is associated with its own error chain. The main error source in this chain is the ultrasound probe calibration process. In addition, small variations in the speed of sound in different tissue types are a potential problem [97].

These points have major implications for the rational behind testing a navigation system in the lab using a phantom, and make a statement about the interesting parameter to the surgeon: the overall clinical navigation system accuracy. A lab test of a system based on preoperative MR/CT using a rigid phantom will give a very good navigation system accuracy (NSA<0.5mm, se figure 19, red line). Such a test will have limited validity in the general clinical situation, but is important to make sure that the system works as expected. The next phase in the evaluation of such a system would be to conduct a clinical study to investigate the system's ability to deal with a variety of different patient registration problems. Documenting that the system performs well in the rigid case and can deal in a satisfactory way with difficult patient registration cases is the best a system vendor can do. This does not give any information about the NSA experienced during a clinical case though. The surgeon must verify that the accuracy is acceptable after he has performed the patient registration procedure and anatomical landmarks inside the patient must be used to gain an impression about the amount of tissue shift and deformation. This shift and deformation makes systems based on preoperative MR/

**Figure 19.** Navigation System Accuracy (NSA) based on preoperative (p) MR (red line) and intraoperative (i) US (green

line). iUS can be used to correct pMR using various image-to-image registration techniques (blue line).

not updated to reflect the changing patient terrain as surgery proceeds.

CT of limited use during the procedure.

52 Advancements and Breakthroughs in Ultrasound Imaging

As stated previously, the most important parameter for the surgeon is the overall clinical Navigation System Accuracy (NSA). Although this parameter is difficult to assess, we believe that for ultrasound-based navigation an estimate can be made, based on a comprehensive laboratory evaluation and a thorough understanding of the significant additional error sources that occur in the clinical setting. Table 1 summarizes how such a calculation can be carried out assuming that a comprehensive evaluation of the system gives a NSA below 1.4 mm in a controlled laboratory setting. The error sources are assumed to be stochastically independent so that their contributions can be added on a sum-of-squares basis.



**Table 1.** Overall clinical NSA estimates

As can be seen from table 1 it is possible to achieve an overall clinical NSA close to the NSA found in the laboratory under favorable conditions, i.e., when the speed of sound used in the scanner is close to the average speed of sound in the tissue imaged, and the ultrasound volumes are frequently updated. The need for updates can be determined by real-time 2D imaging. If these conditions are not met, the accuracy becomes poorer.

#### **6.3. Method for assessing ultrasound-based navigation accuracy**

As we have seen the ultrasound-based NSA found in the lab using a phantom is valid in the OR (Operating Room) as well, under normal conditions. This makes it very interesting to develop a method that can measure the NSA automatically. We have previously suggested a method based on a phantom with 27 wire crosses and correlating an ultrasound sub-image of each cross to a synthetic template of the cross [98], and the method has been used in a thorough accuracy evaluation of a commercial navigation system [97]. We have since that developed a method that seems to be even more robust, in addition to being more flexible and more convenient to integrate in a navigation system (see figure 20). The method can be used for substantially different ultrasound probes and the phantom is easier to build and to measure accurately. The technique is based on sweeping over the single wire cross with the ultrasound probe, reconstruct all the frames into a volume containing the cross, segment and extract the centerline of the cross and register it to a centerline representation of the accurately measured physical cross, acting as a gold standard, using a modified version of the ICP algorithm [86].

A slightly more advanced technique for 2D ultrasound guidance includes a needle guide mounted on the ultrasound probe. The guide will ensure that the needle tip is in the image plane at a given depth depending of the ultrasound image sector and the angle of the needle guide. The angle of the needle guide has to be adapted to the depth of the target. Even though this system provides assistance in keeping the needle in the image plane, the operator has to do imaging and puncturing at the same time. In addition, the orientation issues concerning the ultrasound image relative to the patient is not solved and the anatomical overview is restricted to the current real time 2D image. However, the method is fast, does not require specialized equipment or complicated logistics, and provides sufficient guidance for a number of applications such as biopsy of thyroid nodules [107], placement of ventricular catheters in

**Figure 20.** Automatic method for evaluating the accuracy in ultrasound-based navigation. A) The phantom with a sin‐ gle wire cross in the middle of the water tank and a reference frame in the front. B) Physical wire cross in green and an ultrasound volume of the wire cross in gray. C) The ultrasound data is segmented (red) and a small mismatch to the gold standard in green can be observed, i.e. small inaccuracies exist. D) Centerlines of the green and red wire crosses. E) Iterative closest point (ICP) registration between the two centerlines, initial correspondence shown. F) After some iterations. Final results showing the centerlines (G) and the wire crosses (H). The displacement is equal to the NSA.

Ultrasound-Based Guidance and Therapy http://dx.doi.org/10.5772/55884 55

Neuronavigation is the term used to describe the use of computer-assisted methods to guide or navigate instruments within the confinements of the scull (or spinal column) during surgery. A neuronavigation system should ideally provide high navigation accuracy throughout the surgical procedure. However, the anatomy of the brain is known to shift position after opening of the skull and dura due to drainage of cerebrospinal fluid (CSF), gravity effects and/or removal of tumor masses or hematomas. This shift in the position of the anatomy is often referred to as *brain shift* and has been shown to occur in the early stage of the surgery with displacement values ranging up to several centimeters [111-113]. The brain shift may therefore significantly impair the accuracy of navigation based on preoperative images as the surgery proceeds. Intraoperative ultrasound imaging provides a solution to the brain shift problem.

the brain [108, 109] and amniocentesis [110].

**7.1. Ultrasound-based navigation in neurosurgery**
