Учебники / Computer-Aided Otorhinolaryngology-Head and Neck Surgery Citardi 2002

is usually incorrect selection of the point in the preoperative imaging data set and/or in the operating field volume. Both of these are user errors.

User errors rarely influence the accuracy of automatic registration, since the registration process simply localizes the fiducial markers on the relocatable headset in the preoperative imaging data set. As a practical matter, user error can compromise the next step in these systems. If the relocatable headset is not correctly placed on the patient, then differences between the intraoperative placement and the placement used at the time of preoperative imaging leads to obvious errors. For this reason, radiology personnel and operating room personnel (including the surgeon) must be familiar with the relocatable headset.

Registration protocols are designed to support surgical navigation. If registration fails, then surgical navigation will be impossible. In some cases, registration may seem successful, but the surgical navigation accuracy will be poor. For this reason, surgeons must continually check the fidelity of surgical navigation by localizing against known anatomical landmarks in the operating field volume throughout the case. When surgical navigation accuracy is compromised, the cause usually can be tracked to one of the following problems:

1.The initial registration was not as robust as it initially appeared.

2.The DRF has shifted relative to the operating field volume.

3.The CAS pointer is damaged, or the CAS pointer is no longer calibrated due to even a slight shift in the alignment of the tip relative to the ILD.

4.The tracking system has failed. The actual hardware can fail. Transient errors are more common. If the line of sight between the ILDs and the camera array is not maintained for optical systems, then the system cannot track. Similarly, distortions of the electromagnetic field can disrupt tracking based on electromagnetic sensors.

5.The CAS software can freeze/crash. These bugs can be obvious, but they can be more subtle. If the latter is true, then detection of this error can be much more difficult.

4.7CONCLUSION

Registration refers to the process through which corresponding fiducial points in the preoperative imaging data set volume and the intraoperative operating field volume are aligned. As a result, registration serves as the foundation for surgical navigation. Paradigms for registration may be divided into the categories of manual point mapping, automatic registration (with a relocatable headset), and contour mapping. Each of these approaches offers specific advantages and disadvantages. Because of the complexity of registration and its impact on clinical applications, surgeons should be familiar with registration principles. This will serve to enhance the overall clinical effectiveness of CAS.

separate but communicating cryostat, which provides a homogeneous spherical imaging volume approximately 30 cm in diameter. Surface flexible transmission and receiving radio frequency coils are placed over the area of interest for image acquisition. The system is also designed to allow differing patient positions, including the sitting position, as well as allowing the operate surgeon to be positioned sitting or standing at either the patient’s head or side (Figure 5.2). A highquality two-way audio system provides communication between the operating surgeon and radiology personnel during the procedure (Figure 5.3).

In addition to the magnet, the operating suite is specially designed to merge features of a radiology suite with the essential features of a traditional operating room. In order to do so, multiple special innovations were conceived to allow the introduction of many of the basic instrumentation, anesthesia, and monitoring devices into and around the magnetic field. The relevant concerns include safety of both the patient and health-care personnel and the compatibility of instruments, monitoring devices, and anesthesia delivery, as well as proper functioning of various image display and integration systems.

FIGURE 5.2 Open magnet interventional magnetic resonance imaging unit. The surgeon’s access to the operative field is demonstrated.

FIGURE 5.3 Interventional MR console. The radiologists and MR technicians control the I-MRI unit through this interface. Direct interaction with the surgical team influences the progress of the surgical procedure.

5.2.2 Anesthesia Considerations

All of the anesthesia equipment employed in the I-MRI suite must be magnetic resonance (MR) compatible. The anesthesia workstation (XL10 MRI Compatible Anesthesia Station: Ohmeda, Madison, WI), ventilator (Omnivent, Topeka, KS), and monitoring equipment systems, including pulse oximetry, electrocardiogram, automated noninvasive blood pressure and invasive pressure transducer, were specially designed to be MRI compatible. Most of the other anesthesia equipment such as intravenous lines, needles, and syringes are standard, and no modifications were required. Other items meeting modification included aluminum oxygen tanks and IV poles as well as nonmagnetic lithium batteries. Unfortunately, MRI-compatible infusion pumps and intubating devices are not currently available. This requires that patients be intubated outside the operating enclosure, followed by subsequent transfer into the I-MRI procedure room [8].

One of the major concerns for patient safety involves the inability to perform a standard 12-lead EKG within the procedure room because of the ST wave distortion caused by the magnetic field. A limited monitoring EKG system was developed with EKG leads composed of carbon-fiber or fiber optics to provide intraoperative cardiac monitoring. In the event of a potential intraoperative cardiopulmonary event, the patient would be brought outside the procedure room

known as ‘‘interactive scan plane control’’ has been developed to allow interactive scanning with the plane to be scanned determined by the operating surgeon. The scan plane is determined by the surgeon at discrete times during the procedure by using a single-purpose pointer brought into the surgical field [8]. The position of this pointer is recognized by a detection system utilizing infrared LED that are mounted above the vertical operating gap within the magnet. The position of the LED is then calculated by computer algorithm providing coordinates in the sagittal, axial, and coronal planes. The pointing device is then visualized by the surgeon as a small crosshair on the MRI console and the liquid crystal display, and the position of the pointing device is verified. The main purpose of the pointing device is to dictate the three-dimensional planes to be imaged by the realtime scan. Once the desired position has been confirmed, a real-time MR scan may be obtained, and multiplanar MR images centered on the desired position in the surgical field can be visualized. This allows active verification of surgical position within the operating field and shows changes in anatomy produced by the procedure itself [8].

Static pointing devices have also been developed for use in surgical localization. Such pointers can be constructed of gadolinium-filled or labeled tubes and instruments. When real-time scans are conducted with these instruments in the surgical field, these pointers appear as high-intensity objects on T1-weighted MR images. This type of pointing device is non interactive and does not dictate the plane to be imaged, but rather serves to verify surgical position, even when anatomy has changed due to the surgical procedure.

5.2.4.3 Acquisition Times

With the current technology, images may be acquired at discrete intervals during surgery with a mean acquisition time of 14 seconds using a ‘‘fast spin echo’’ sequence, with a relaxation time (TR) of 400 msec and an excitation time (TE) 26 msec. This acquisition time compares quite favorably to conventional spin echo sequences, which require minutes for acquisition. Even faster acquisition times of images during the procedure can be achieved with ‘‘fast gradient sequences.’’ However, though these images only require 1–2 sec to compose, these sequences provide less signal-to-noise quality, and thus provide significantly less anatomical resolution. They have been found to afford good intraoperative instrument localization.

5.2.5 Training of Personnel

Accompanying the development of any innovative technology is a learning curve required to employ the technology effectively. This is even truer in the I-MRI suite because of the complex interactions between the operating surgeon, radiology personnel, and anesthesia personnel. Prior to utilizing the I-MRI suite in

clinical trials, extensive training was conducted with cadavers [8]. Such training was conducted not only to familiarize operating surgeons with the spatial limitations of the open magnet system, but also to verify the MR compatibility of the surgical instrumentation. In addition, studies were conducted to verify the accuracy and precision of the optical tracking system and the intraoperative MR images. The three-dimensional optical tracking system has been found to be linear to within 1% through the 30 cm imaging volume. The tracking accuracy of the instrument within the imaging volume exhibits a 1 mm tolerance. During training, it was evident that one problem with the optical tracking system is that it is extremely sensitive to changes in instrumentation angle, which may obscure optical tracking of the LED. Such considerations must be carried over into clinical preparations and planning.

In addition to the surgeon’s operative training, extensive training for support staff and personnel is required to orient them to the unique safety and efficiency considerations required by the open magnet system. A special team of anesthesia personnel familiar with the I-MRI and its limitations was assembled and trained. Similar considerations were employed in assembling the nursing staff for head and neck I-MRI procedures. Patients and operating room personnel with cardiac pacemakers or aneurysm clips that have not been verified as MR compatible must be excluded from the operating suite. Special protocols for emergencies, equipment failure, and unforeseen circumstances must be constructed and implemented in order to ensure proper patient safety [8].

5.3 APPLICATIONS

5.3.1MRI-Guided Biopsy of Head and Neck Lesions

Fine needle aspiration biopsy (FNAB) is now commonplace in the diagnosis of head and neck lesions. It may be used to provide an initial cytological diagnosis for primary lesions arising in major salivary glands, submucosal lesions in the upper aerodigestive tract, and cervical lymphadenopathy. It has been shown to be highly sensitive, specific, and reliable [13,14].

Extensive work has been done with coupling FNAB to various imaging modalities. Such a linkage offers the attractive benefit of ensuring that the desired tissue is actually sampled. Increasing accuracy for FNAB has been demonstrated when ultrasound or CT guidance is used to direct and confirm the biopsy [15,16]. Sensitivity and specificity may be increased to up to 95% with these adjunctive techniques. Image-guided biopsy techniques are essential when the lesion is not palpable, situated within the deeper structures of the head and neck, or when vital structures lie between the scan and the lesion of interest [16].

MR-guided FNAB may provide distinct advantages over other image-

guided biopsy techniques. First, with the capacity to image in multiple planes, three-dimensional confirmation that the target tissue has been sampled may be realized. Second, of the major imaging techniques, MRI provides the best resolution and distinctions of the various soft tissues within the head and neck [17]. Since in many cases it is desirable to follow the biopsy needle from its entry point continuously until it reaches the target tissue, multiple scans are often required. In this situation, iterative CT scans will result in additional radiation exposure to the patient (and health-care personnel), and such CT scans may require administration of intravenous contrast with potential attendant side effects. Although ultrasound imparts no radiation to the patient, it is less efficient at delineating soft tissue/soft tissue interfaces than MRI. Because of its favorable imaging characteristics (Table 5.1) and the lack of radiation exposure, MR-guided biopsy in the head and neck has recently been explored [3,6].

Fried and associates [3] reported on their initial patient experience with MR-guided needle biopsies within the head and neck. In their series of seven patients, none of the lesions requiring biopsy could be palpated or visualized on routine examination. Target lesions included four parotid gland lesions, two tumors of the parapharyngeal space, and one lesion of the second cervical vertebra. Using intraoperative fast-gradient echo pulse sequences, with 2–4 sec acquisition times, interactive MR-guided biopsies were performed. The authors were able to successfully localize the desired tissue for biopsy using both the previously described optical tracking system as well as sequential intraoperative scanning. Notably, the authors found that the needle trajectory required alteration several times based on the intraoperative images. In five of seven cases the MR-guided

TABLE 5.1 Summary of Important Distinctions Between Potential Intraoperative Imaging Modalities

aThe third (sagittal) imaging plane may be viewed from reconstructed images, but this consumes additional time and is not yet widely employed intraoperatively.