Image-Guided Robotic Interventions for Breast Biopsy and Ablative Therapies: Challenges and Opportunities

In 2008, the American Cancer Society estimated that 182,460 women would be diagnosed with breast cancer and that 40,480 women would die of breast cancer in the USA alone. One in eight women born today is likely to be diagnosed with breast cancer during their lifetime. Although these statistics are discouraging, positive trends are evident as a result of innovations in the diagnosis and treatment over the past decade.

Recent, large-scale studies reported in The Lancet and The New England Journal of Medicine demonstrate the value of MRI as an effective tool in the diagnosis of breast cancer [1,2]. The results from these studies further stress the importance of early and accurate detection of breast cancer. This has also led to a renewed emphasis on less invasive procedures that maximize breast conservation while providing effective treatment strategies and optimal outcomes [3–5]. Emerging techniques for minimally invasive (and sometimes noninvasive) in situ treatments of breast cancer include cryoablation, radiofrequency ablation (RFA), microwave thermotherapy, interstitial laser ablation and focused ultrasound ablation [6–10]. Percutaneous RFA has been widely applied with safety and success in treatments of hepatocellular tumors and other liver lesions as well as in renal tumors [11,12]. In these applications, computed tomography (CT), ultrasound or MRI is used to guide the placement of needle(s) directly into the tumor for delivery of radiofrequency (RF) energy and achievement of local hyperthermia. Several pilot in vitro and in vivo studies of RFA techniques in breast tumors have demonstrated promise [13–17]. Ultrasound-guided RFA performed in patients immediately before surgical resection resulted in coagulative necrosis of 96% of resected tumor with a very low complication rate [18]. Similar success was reported by another study, which also noted that postablation MRI was predictive of histologic findings at delayed resection in patients scheduled for lumpectomy or mastectomy [19,20]. More recently, there have been reports of success with RFA of breast tumors, particularly when combined with external-beam radiation therapy or with adjuvant chemotherapy [15,21]. There are also encouraging reports of palliative effects and improvements in quality of life [21,22]. The majority of investigational studies of RFA in breast cancer have been conducted using ultrasound guidance for needle placement [23,24]. A significant limitation of this approach in any RFA application is that RF heating causes gas microbubbles to form in tissues, resulting in considerable acoustic noise/shadowing that impedes the physician's ability to evaluate the effect of the treatment – a crucial capability in achieving maximal extirpation of tumor. Moreover, ultrasound is limited in its ability to detect and assess the temperature changes in a tumor and its surrounding tissue that signal a tumoricidal effect during RFA, with resulting complications that range from incomplete tumor destruction to injury of adjacent structures (i.e., overlying skin) [12,25].

Having been used with RFA in hepatic and other cancers, MRI is not subject to these limitations [26,27]. MRI guidance has several advantages, including excellent soft-tissue contrast; near real-time visualization with no ionizing radiation burden [28,29]; interference-free MRI temperature-mapping techniques that provide the ability to directly visualize temperature changes in three dimensions, so that the extent of tumor destruction is apparent and the physician can iteratively modify treatment to ensure maximum effectiveness [30,31]. This approach is suitable in breast tissues, which offers RFA access and imaging and no interference with lungs or major vessels. While a significant amount of research has been conducted to improve the success rate and applicability of RFA, this research has concentrated on improving RF probes and procedures to increase the volume of ablated tissue, or to decrease the duration of the ablation procedure [32,33].

Despite these very important advantages, MRI is underutilized in the detection and management of breast cancer. For example, MRI-guided breast biopsies are currently performed under image guidance. The image guidance involves multiple steps. First, diagnostic quality MRI scans are obtained on the patient scheduled for biopsy where the lesions are identified. Second, the patient is pulled out of the magnet but remains on the table and is prepared for biopsy. Typically, a grid is placed laterally, which helps the interventionalist in positioning the needle in a manner that it can reach its target. At this juncture, the interventionalist is being guided by the images that were obtained a priori with the assumption that no movement of the patient has occurred between the time that the images were acquired and when biopsy is conducted. Furthermore, the trajectory of the needle is assumed to be straight while in reality it may take a curved path due to the resistance that the needle experiences as it traverses to its target. Once the needle reaches the supposed target, a biopsy is performed. Confirmation of the biopsy location is then made by moving the patient back into the bore of the magnet and reacquiring images. If the biopsy was not obtained from the intended target, the procedure is repeated again. Although widely used, this procedure is cumbersome, time consuming, leads to patient discomfort and could lead to significant tissue scarring. Performing the same procedure under continuous image guidance will help alleviate some of these problems and eliminate sampling errors that lead to false negatives.

Continuous interventional image-guided procedures require that the device being used for intervention be visible at all times. In the context of breast imaging, this means that the biopsy device or the therapeutic device be continuously tracked as it punctures the skin and makes its way towards the target. Therefore, additional functionality that allows one to change imaging parameters such as the imaging plane, field of view and contrast on the fly as the device is being tracked is desirable. Improvements in MRI technology has allowed the acquisition and reconstruction of MRI scans at rates of 30 frames per second or more [34]. However, the possibility of extending the technology to automating biopsy sampling and ablative therapies under continuous MRI has received less attention within the engineering and medical communities. In order to lower sampling errors for accurate diagnosis, and provide optimal ablation treatment for malignant breast tumors, the key question would be whether robotic technologies can be developed to enable needle guidance for biopsy and ablation while the patient is in the scanner and is being imaged continuously under MRI?

A key component for the successful implementation of continuous imaging of the breast for either biopsy sampling or delivering ablative therapies is the ability to harness the advances in robotics and to enable this process in the highly restrictive MRI environment without compromising the patient's comfort. The use of robots for medical applications has shown great promise in extending the surgeon's natural capabilities. In 1985, the first robotic surgical system was used in the operating room for neurosurgical biopsies with CT guidance [35]. This system utilized a programmable universal machine for assembly (PUMA) robot arm with a probe guide mounted on the end-effector and a CT scanner for accurate positioning during brain tumor biopsies. Since then, several other robotic systems have been developed for a variety of procedures. Some of the robotic systems available commercially and in the academic environment are position-controlled robots and do not have force-reflecting capabilities. Therefore, these systems are considered unilateral teleoperators. Bilateral teleoperators, on the other hand, offer the advantage of controlling contact forces with the environment and reflecting the contact forces to the operator. These advantages offered by bilateral teleoperators could allow surgical procedures to be performed more naturally and with minimal contact forces exerted on the soft tissue during the procedure [36]. A promising area of research in bilateral teleoperation is the use of stable force-feedback schemes that enhance the surgeon's capabilities by sensing remote forces, thereby improving task performance and efficiency. The use of virtual fixtures is one example of such an approach [37,38]. Virtual fixtures are forces applied to the operator that perform functions, such as preventing access to ‘forbidden’ regions, or constraining the motion of master or slave to move along a desired trajectory. A similar approach has been proposed by Turro et al., who suggest the use of virtual constraints, as well as attractive potential fields at the master and repulsive fields at the slave, to aid in such tasks as following a desired trajectory or avoiding manipulator singularities [39]. These approaches have demonstrated promising results in improving task performance and safety in medical teleoperation.

In addition to the development of general application haptic (i.e., sense of touch; force and tactile) devices, there has been significant research in the development of devices for the reflection of surgical forces to the surgeon. Many of the developed surgical tools and systems have also incorporated a force reflection component for a specific surgical procedure or task. One typical configuration is the addition of a force feedback handle that allows a surgeon to remotely control the surgical tool, as well as receive tool–tissue interaction forces as in conventional laparoscopic surgery [40–44]. Other researchers have used general application haptic devices to reflect the surgical forces to the surgeon. Researchers have also designed and developed haptic devices with specific application to surgical tasks or procedures.

There has also been limited progress in the development of robotic systems for use under an imaging modality such as CT, ultrasound and MRI. One effort at such an automated device has been made by Wood et al. at the NIH, who have designed and implemented a CT-compatible robot for needle insertion [45,46]. Activated by a pneumatic gripper, the device is capable of advancing an RF probe inside a CT scanner, through either a semiautomated or fully automated process[47]. While this device has been successful within the CT environment, it has not been designed or implemented for use with MRI. Since needle tracking, tumor visualization and thermographic mapping have been shown to be far superior with MRI [48,49], it is necessary to bring automation of biopsy and RFA probe placement to the MRI environment. A second effort at automating needle-based cancer treatments has been made by Fischer et al., who designed and implemented an MRI mammography-compatible device for breast biopsy and cryoablation [50]. A third robotic device was designed by Larson et al., who implemented a robot for automation of various needle-based interventions in the breast [51,52]. This design is fully automated; however, its components are not fully MRI-compatible. The ultrasonic motors cause image artifacts, resulting in long telescoping shafts that require the motors to be placed outside the imaging area. While there has been progress in the area of robotic systems for use in MRI, the progress has not been as rapid as one would expect. However, the component technologies for enabling the development of a bilateral teleoperated robotic system with haptic feedback capability for use in continuous MRI are in place [53].

It is envisioned that in the next 10–15 years, image-guided robotic systems will play a major role in accurately targeting breast lesions for both diagnosis as well as for treatment. The ultimate challenge would be to enable diagnosis and treatment as a one-sitting procedure rather than spacing it out over several months, which would lead to patient anxiety and repeat biopsy prior to treatment.