A common method for increasing energy deposition throughout an entire lesion has been to repeatedly insert multiple RF, laser, and microwave probes into the tissue to increase the diameter of induced coagulation [3, 5, 7, 36]. This approach is both time-consuming and difficult to use in the clinical setting, particularly because multiple overlapping treatments must be performed in a contiguous fashion (in all three dimensions) to destroy the entire lesion. Simultaneous application of energy using arrays can reduce the duration of therapy [42, 43]. However, the precise positioning of multiple probes can be technically challenging. The development of umbrella RF electrodes with multiple hooked arrays has overcome some of these problems and has enabled the creation of larger zones of coagulation [3, 9, 44].

Much recent development has centered on strategies that preferentially cool tissues nearest the probe in an attempt to increase overall energy deposition. Internally cooled electrodes have been used with RF, microwave, high-intensity sonography, and laser [18, 19, 20, 21]. For internally cooled devices, two internal lumens permit the delivery of chilled perfusate to the tip of the electrode and allow the warmed effluent to be removed to a collection unit outside the body. This procedure causes a heat-sink effect that removes heat closest to the electrode (Fig. 1). Pulsing of energy is another strategy that has been used with RF and laser to increase the mean intensity of energy deposited. When pulsing is used, periods of high energy deposition are rapidly alternated with periods of low energy deposition. If a proper balance between high and low energy deposition is achieved, preferential tissue cooling occurs adjacent to the applicator during periods of minimal energy deposition without significantly decreasing heating deeper in the tissue. Thus, even greater energy can be applied during periods of high energy deposition, enabling deeper heat penetration and greater tissue coagulation [45, 46]. Combination of both internal cooling and pulsing has been shown as synergistic with even greater tissue destruction observed than with either method alone [47].

Improved heat conduction within the tissues by injection of saline and other compounds has also been proposed [48, 49, 50]. The heated liquid spreads thermal energy farther and faster than heat conduction in healthy “solid” tissue. An additional potential benefit of simultaneous saline injection is that it increases tissue ionicity, thereby enabling greater current flow. Similarly, amplification of current shifts with iron compounds injected or deposited in the tissues before ablation has been used for RF and microwave [50].

Another primary factor that can alter the extent of coagulation necrosis is tissue composition because heat conducts through different tissues at various rates [4, 51]. For example, poor thermal conduction has been documented for bone compared with muscle [51]. This fact has been an advantage in the treatment of hepatocellular carcinomas and vertebral body lesions. Livraghi et al. [4] have described the “oven effect” in which cirrhotic tissue insulates hepatocellular carcinoma nodules and increases temperatures within the targeted tumor during RF therapy. Dupuy et al. [51] have shown that cortical bone also serves as an insulator, enabling treatment of vertebral body lesions without damaging the spinal cord.

Strategies that decrease tumor tolerance to heat have been proposed but are not yet well studied. Theoretically, previous insult to the tumor cells by cellular hypoxia caused by vascular occlusion or antiangiogenesis-factor therapy (i.e., endostatin) or prior tumor cell damage from chemotherapy or radiation could be used to increase tumor sensitivity to heat. Synergy between chemotherapeutic agents and hyperthermic temperatures (42-45°C) has already been established [27, 28].

Multiple imaging techniques (sonography, CT, and MR imaging) can be used to guide the percutaneous placement of thermal energy applicators into the selected target [1, 2]. Because in most cases adequate lesion conspicuity and visualization of the applicator can be achieved with any of these methods, the choice of imaging technique is often dictated by personal preference or research interests. Most imaging-guided thermal ablation procedures have thus far been performed with sonography (Fig. 2A, 2B, 2C). Benefits claimed for sonography include the real-time visualization of applicator placement, portability of the technology, nearly universal availability, low cost, and ability to target and guide ablation therapy with intracavitary endoluminal transducers (i.e., for transrectal or transgastric energy application to the prostate and abdominal organs). Limitations of sonography include occasional poor lesion visualization as a result of a lack of innate tissue conspicuity or overlying bone- or gas-containing structures. MR imaging generally provides greatest tumor-to-tissue conspicuity and the ability to use multiplanar guidance. However, this technology is relatively expensive, requires specialized ablation equipment that is compatible with a high magnetic field, and is the least available for general clinical use. CT and, more recently, real-time CT fluoroscopy have also been used to ensure adequate positioning of the energy applicator. Though CT fluoroscopy has not been extensively evaluated, it is fair to say that CT falls between sonography and MR imaging with respect to cost, tissue contrast, and complexity. In our clinical practice, we use a combined approach of CT fluoroscopy and sonography at the same setting to document optimal RF electrode positioning (Fig. 3B).

To prevent under- or overtreatment of a lesion, it is essential to have accurate and reliable methods for determining the adequacy of therapy. Thus, significant investigation into the development of imaging strategies that enable rapid assessment of the extent of tissue destruction induced by thermal ablation is being conducted. Despite initial enthusiasm, gray-scale sonographic findings observed during the thermal ablation procedure are not sufficiently accurate in predicting the extent of coagulation [7, 8, 53]. The progressively increasing hyperechogenic focus often seen surrounding the distal portion of the applicator during the application of energy represents microbubbles of gas that form in the heated tissue and does not represent tissue coagulation [54] (Fig. 2B). This hyperechogenic region can be variable in size, may be quite irregular in shape and contour, and often shows complete resolution within 1 hr of ablation (Fig. 2C). Additionally, this intense echogenicity can often obscure the energy applicator and tumor while increasing the difficulty of repositioning for further treatment.

Conventional color-flow and power Doppler sonography have similarly not been found useful in assessing the extent of induced coagulation [7, 8]. However, in one study contrast-enhanced color Doppler sonography with a synthetic microbubble sonographic contrast agent was able to achieve 92% accuracy in predicting the extent of coagulation in VX2 rabbit tumors immediately after RF ablation [55]. Additionally, sonographic contrast material has been used to direct a second energy application to residual enhancing (and presumably viable) foci within the treatment zone [56].

For solid organs such as the liver, unenhanced CT scans obtained immediately after ablation often reveal increased density at the center of the treatment zone, most often surrounded by a region of hypoattenuation [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 53, 57, 58]. With the exception of encapsulated lesions such as those of hepatocellular carcinoma, the margins of this outer hypodense zone are often too diffuse to be of sufficient sensitivity to assess therapy. However, contrast-enhanced CT is useful in discriminating between ablated and residual viable tumor immediately after thermal ablation because it shows regions of hypoattenuation devoid of characteristic tumorous or parenchymal enhancement in treated portions of the tumor. For intrahepatic metastases, the differentiation of coagulation necrosis from hypoattenuating tumor is usually easiest on images in the equilibrium phase of contrast enhancement (5-10 min after iodinated contrast administration). At this phase, persistent hypoattenuation is seen in coagulated tissues but not in viable tumor [31]. Hepatic arterial phase images are most useful for early-enhancing hepatocellular carcinomas (Fig. 3A, 3B, 3C, 3D). Imaging during the hepatic arterial phase can also show a thin rim of contrast material corresponding on histopathology to an early inflammatory reaction to the thermal damage (Fig. 3C). This inflammatory rim can be seen immediately after ablation and often regresses during the first month after treatment.

MR images characteristically reveal altered signal on both T1- and T2-weighted images [53, 57, 59] (Fig. 4A, 4B, 4C, 4D). Treated areas are devoid of gadolinium enhancement. Several studies have documented the particular usefulness of decreased signal on T2-weighted images as a marker for induced coagulation [59, 60]. Radiologic—pathologic correlation in both experimental and clinical studies has shown that CT and MR imaging findings predict the region of coagulation to within 2-3 mm [31].

One key advantage of MR imaging over other diagnostic imaging techniques is its ability to aid in determining the extent of coagulation during energy application. Heat-sensitive sequences have been constructed and permit tailoring of energy deposition [61, 62, 63]. Such a strategy is most useful in allowing the operator to limit energy deposition when heating adjacent to a critical structure (i.e., nerves) reaches cytotoxic temperatures. Pulsing switches were developed to overcome interference of RF and microwave usage during the acquisition of MR-RF encoded data [64].

Although initial imaging can serve as a good indication of the adequacy of therapy, the resolution and accuracy of current imaging techniques preclude identification of residual microscopic foci of malignancy, particularly at the periphery of a treated lesion (where blood flow is greatest). These viable tumor foci will inevitable continue to grow and, if untreated, will result in failed therapy. Additionally, considering issues of sampling error and the possible difficulty in differentiating between adequately treated and viable tumors with histopathologic techniques alone, we have not found the use of needle biopsy helpful. Thus, longterm imaging follow-up is necessary to find untreated regions of the tumor or to document complete treatment of a given focal malignancy.

Long-term follow-up of thermal ablation with sonography has limited value [7, 8]. Obscuration of the characteristic peritumoral halo observed before treatment is often seen, and the variability of gray-scale sonographic changes precludes accurate assessment of induced coagulation. Sonographic microbubble blood pool agents such as SH 508 A (Levovist; Schering, Berlin, Germany) may be helpful in differentiating treated tumor from the avascular coagulation at 6 months of follow-up [65].

Contrast-enhanced CT has been the mainstay of long-term imaging follow-up (Figs. 3A, 3B, 3C, 3D and 5A, 5B). Coagulated nonenhancing regions increase in conspicuity and develop sharper margins by 2 weeks after ablation [31, 53]. Imaging at 6-12 months can show marked regression of the lesion and the region of induced coagulation necrosis. Most commonly, the nonenhancing treatment focus shrinks less than 20% in volume. A peripheral rim that densely enhances on delayed contrast images often surrounds the region of coagulation. This finding should not be misconstrued as residual tumor, for experimental and clinical studies have shown this rim to represent an inflammatory reaction to the thermally damaged cells [53, 66]. A bulky irregular rim at the edge of a treatment site is the most common appearance of an incompletely treated lesion (Fig. 6).

When using MR imaging for long-term follow-up (>3 months), we have relied primarily on the presence or absence of gadolinium enhancement in the treated region [8, 53]. In comparison with MR images obtained within 3 days of ablation, we have observed heterogeneous alteration on unenhanced T1- and T2-weighted images (4A, 4B, 4C, 4D). This changing variability in signal intensity throughout the ablated region is most likely caused by an uneven evolution of the necrotic area and the host response to thermal damage. Hence, these images have been thus far too variable to be relied on as adequate proof of tumor destruction. The multiplicity of potential imaging sequences and parameters used for MR imaging has only further compounded this problem. Further research may ultimately lead to greater insight into the biologic mechanisms that account for such signal heterogeneity. For gadolinium-enhanced images, it is also common to detect a thin rim of enhancement after treatment. As for CT scans, only when this rim appears bulky is this finding to be interpreted as representing an untreated tumor.

Nuclear medicine has been used in a limited number of patients after ablation therapy. In one study, positron emission tomography scanning with a radioactive glucose analog (¹⁸F-fluorodeoxyglucose) was used to detect active foci of residual tumor after percutaneous ethanol instillation in intrahepatic metastases [67].

Our current imaging strategy after thermal ablation includes an initial contrast-enhanced CT or MR study on the day of treatment to determine whether the patient has residual gross viable disease that requires immediate retreatment. Follow-up imaging is then performed at 1 and 3 months, and every 3-4 months thereafter. These scans are helpful in documenting the presence or absence of residual tumor that often may be amenable to additional thermal ablation treatment. If no evidence of peripheral tumor regrowth is seen by 6-12 months, adequate treatment can be inferred.