Stereotactic surgery was first described by Horsley and Clarke in 1906. They developed a method of locating deep-seated brain lesions by assigning coordinates in three planes to neuroanatomical structures, based on cranial landmarks [1]. In 1947 Spiegel and Wycis introduced frame-based stereotaxy using a plaster head cap known as a stereoencephalatome, and a 3-D coordinate system relative to this [2].

Lars Leksell, a Swedish Neurosurgeon, was the first person to marry the two developing fields of stereotaxy and radiation therapy, and introduced the term “Radiosurgery” in 1951. He used a rigid metal stereotactic head frame fixed to the skull. Small intracranial targets were localised relative to the frame, and radiation was delivered in a single high-dose fraction [3]. The technique initially employed 250 KV x-rays, but in 1967 the first Gamma Knife prototype was developed, using 179 Cobalt-60 sources focused on the target. Since then, Gamma Knife has become widely used for intracranial stereotactic radiosurgery, with sub-millimetre total system accuracy [4].

The 1980s saw the adaptation of linear accelerators for intracranial stereotactic delivery, again using rigid stereotactic head frames, and specialist dosimetry software e.g. X-Knife (Radionics, Boston, MA).

In 1995, Hamilton et al. proposed a method of delivering linac-based stereotactic radiotherapy to spinal lesions using a prototype rigid “extracranial stereotactic frame” and associated 3-D coordinate system. Immobilisation was achieved by transcutaneous frame fixation to spinous processes superior and inferior to the target. They reported an overall treatment accuracy of 2 mm, but the technique was time-consuming, cumbersome, and limited to the delivery of single fractions [5].

Around the same time, Lax et al. developed a stereotactic body frame which, together with a vacuum bag, immobilised the patient from head to mid-thigh. They found the setup reproducibility for liver and lung lesions to be within 5–8 mm for 90 % of the patients. Diaphragmatic movements were reduced to 5–10 mm by applying pressure on the abdomen [6]. Many stereotactic radiotherapy systems today use a similar setup of body frame immobilisation, and some centres use corsets to limit diaphragmatic movement.

However for most extracranial sites the position of the tumour does not enjoy a fixed relationship relative to the external body contour, and can move both between and during each fraction of radiotherapy. An external body frame alone is therefore not sufficient to ensure accurate delivery of radiation to the target. Safe delivery of large fractions of radiotherapy requires sophisticated image guidance.

Image guidance in radiotherapy became a realistic concept with the development of the electronic portal imaging device (EPID) and software to aid quantitative evaluation of patient setup, thus allowing correction of translational errors. The next step was moving from “off-line” to “on-line” image guidance (ie adjusting patient position on the basis of imaging, before each fraction). Accurate identification of tumour position has improved with the use of inserted metal fiducial markers with planar images, or alternatively with the development of volumetric image guidance (eg cone beam CT, or in-room CT on rails). More recently, improved software, together with more sophisticated treatment couches, have meant that correcting for rotational setup errors is now possible. Finally, intra-fractional image guidance is now available, and is a key component of some of the stereotactic treatment systems described below.

Stereotactic systems which use planar imaging have great flexibility with respect to taking multiple intra-fraction images, but largely rely on implanted fiducials. Percutaneous fiducial insertion can be technically difficult, especially in the upper abdomen where it may be necessary to pass through other organs to reach the target lesion. See also Chap. 2.

The therapeutic benefit achieved with dose fractionation has been recognised for over 100 years [7]. Conventional fractionation has emerged from such early observations, with subsequent refinement as our knowledge of radiobiology has developed. The linear quadratic model [8] and its ability to describe cellular response to radiation, together with Withers’ “4 Rs” of radiotherapy—DNA Repair, Reoxygenation of the tumor, Redistribution within the cell cycle, and Repopulation of cells [9], have had a big influence on modern CFR regimes.

In contrast, intracranial radiosurgery exploits the ablative power of large single doses of radiation, which transcends the considerations proposed by Withers. Considerable dose inhomogeneity within the target volume is standard practice, due to the internal dose gradient achieved by using a low prescription isodose (commonly 40–60 % with gamma knife radiosurgery). There is some evidence to suggest that, rather than being a problem, target dose inhomogeneity may enhance the tumoricidal effect [10].

SBRT sits somewhere between the extremes of CFR and radiosurgery. Large doses per fraction are used, and a moderate internal dose gradient achieved, with a typical prescription isodose of around 60–80 %. However, unlike intracranial radiosurgery, inter- and intra-fraction target movement is a significant problem, increasing the risk of irradiating normal tissue, or missing the tumor, during treatment. Also we are treating at sites where the overwhelming clinical experience and evidence is with conventional fractionation. Conceptually, moving to a single large fraction is big step. The linear quadratic model and its derivatives can help clinicians to predict tissue response to altered fractionation regimes. However, there has been concern that it does not accurately predict tumor cell response at the higher doses per fraction (>10 Gy) seen with stereotactic treatment [11]. At these doses it is not clear to what extent modest fractionation (2–5 fractions) differs from a single fraction with respect to tumour response and normal tissue effects.

Unsurprisingly, therefore, there has been a large variation in dose and fractionation across SBRT series published to date. Whilst some SBRT centres adopt a “single large fraction” strategy for many patients, other centres would prefer to fractionate in similar cases. Current regimes have in many cases been derived empirically, often the result of cautious dose escalation.

See also Chap. 5.

A number of modern linacs with on-board imaging capabilities meet the basic image guidance requirements for delivering SBRT (e.g. Varian TruBeam, Elekta Synergy). A micro-multi leaf collimator can be added to produce the required degree of conformality for stereotactic plans.

More recently we have seen the introduction of linacs fully adapted as integrated stereotactic delivery systems. Novalis TX has a Varian Trilogy linac base with micro (2.5 mm) MLC, together with Brainlab ExacTrac image guidance and 6D Robotic Treatment Couch, and associated software. Elekta Axesse is a similar integrated system.

The TomoTherapy System (Accuray Inc, Sunnyvale, CA) has a ring gantry as used in diagnostic CT scanners, and delivers helical IMRT via thousands of small beamlets. Couch movement is continuous during radiation delivery. The system has on-board image guidance with megavoltage CT.

The Cyberknife (Accuray, Sunnyvale, CA) is an image-guided robotic radiosurgery system. A compact 6 MV x-band linac is mounted on a six joint robotic arm. This provides huge flexibility in beam pattern generation, allowing the system to produce very conformal, non-isocentric plans with a steep dose gradient around the target. Near real-time imaging is achieved using two diagnostic x-ray sources positioned orthogonally in the treatment room. Targets that move with breathing can be tracked using the Synchrony respiratory tracking system. Thus the system can correct for both inter- and intra-fraction target movement, which obviates the need for a stereotactic frame. Total system accuracy has been found to be less than 1 mm [12, 13].

See also Chap. 3.

With thoracic or upper abdominal SBRT we face the additional challenge of accounting for intra-fraction target movement with breathing. The majority of the published data come from centres using gantry-based linacs with vacuum and/or frame body immobilisation, and diaphragmatic pressure to reduce breathing movement. Respiratory gating methods such as Active Breathing Control (used in some series) have sought to reduce the volume of tissue irradiated by requiring a smaller GTV-PTV margin. The latest gantry-based stereotactic systems (eg Novalis TX) allow respiratory gating using infra-red chest wall tracking and intra-fraction x-ray imaging. 4-D CT planning allows the construction of an internal target volume which takes into account the tumour position at all phases of the respiratory cycle. It has been shown to increase the accuracy of, and reduce the volume of normal tissue irradiated in conformal radiotherapy, and is now being used in some lung SBRT centres.

For Cyberknife, the Synchrony Respiratory Tracking System monitors chest wall movement continuously via an infra-red camera and LEDs placed on the patient’s chest. Regular static KV images of the tumour are taken during setup and treatment, and correlated with chest wall movement. A regularly updated predictive model is generated which anticipates future tumour movement with breathing, and the robotic arm moves the radiation beam accordingly. The total accuracy for moving targets has been reported as 1.5 mm. “Xsight Lung” is a further software development for Cyberknife, which allows the tracking of certain peripheral lung lesions (within strict parameters) without the need for implanted fiducials.