There is no commercially available real-time dosimeter that can accurately measure output factors for field sizes down to 4 mm without the use of correction factors. Silicon diode detectors are commonly used but are not dosimetrically water equivalent, resulting in energy dependence and fluence perturbation. In contrast, plastic scintillators are nearly dosimetrically water equivalent. A fibre optic dosimeter (FOD) with a 0.8 mm(3) plastic scintillator coupled to an air core light guide was used to measure the output factors for Novalis/BrainLab stereotactic cones of diameter 4-30 mm and Novalis MLC fields of width 5-100 mm. The FOD data matched the output factors measured by a 0.125 cm(3) Semiflex ion chamber for the MLC fields above 30 mm and those measured with the EBT2 radiochromic film for the cones and MLC fields below 30 mm. Relative detector readings were obtained with four diode types (IBA SFD, EFD, PFD, PTW 60012) for the same fields. Empirical diode correction factors were determined by taking the ratio of FOD output factors to diode relative detector readings. The diodes were found to over-respond by 3%-16% for the smallest field. There was good agreement between different diodes of the same model number.

The recently commercialized PTW microDiamond detector (T60019) has been designed for use in small radiation fields. Here we report on the measurement of relative output ratios for small fields using five microDiamond detectors. All of the microDiamond detectors over-responded in fields smaller than 20 mm, by up to 9.3% for a 4 mm field. The over-response was independent of accelerator type and choice of collimation. The over-response was slightly larger than that observed in silicon diodes. Since all five microDiamond detectors showed the same over-response the corrections presented here should be transferable to other examples of the microDiamond detector, provided that the detector meets the manufacturing specifications and the beam characteristics are comparable.

Key wordsEquipment; pulse oximeter. Measurement.The performance and some potential errors of pulse oximeters under conditions of good and poor perfusion, and high and low saturation were reported before.'-5 This paper is a review of other potential errors and the clinical implications will be reviewed later.6 Effects of interferenceThe pulse oximeter signal can be corrupted by interference from ambient light, electromagnetic signals and motion. Emergency Care Research Institute (ECRI) tested 13 models of pulse oximeter for the effects of interference by electrosurgical units, phototherapy and surgical lamps, and shaking and ~i b r a t i o n .~ These results are summarised below and compared with other reports where possible. ElectrosurgeryElectrosurgical interference caused erroneous readings in six of the 13 units with no clear warning that the signal was unsatisfactory; three were unaffected and the rest displayed a clear interference warning signal. Interference can be reduced if the distance of the pulse oximeter probe from the surgical site is made as great as possible.

SummaryThere is no absolute reference for oxygen saturation, although multiwavelength in vitro oximeters are accepted as the ' Key wordsEquipment; pulse oximeter. Measurement.Many recent papers have dealt with the accuracy of pulse oximetry and comparisons of different brands under various conditions, but there are many potential sources of error that need to be considered for the evaluation and appropriate use of pulse oximeters. There is widespread confusion about what pulse oximeters actually measure and which reference oximeter to use. The use of different notations and statistical methods have made comparisons between studies difficult. This paper is one of a series: two companion papers (which compare and rank, under conditions of poor perfusion, 20 pulse oximeters with finger probes)' and 10 with ear, nose or forehead probes (in press). The series aims to review all the important sources of error in pulse oximetry and to discuss their clinical significance. Other papers consider the effects of changes in saturation and signal quality; the effects of dyshaemoglobins, dyes, other pigments and extraneous factors; and present a clinical overview of the significance of all the issues discussed. Theory of pulse oximetryTwo wavelengths of light are transmitted from light emitting diode (LED) sources and pass through an arterial bed, usually in a finger or earlobe. The attenuated light is received by a detector and converted into an electrical signal which is analysed by a microprocessor to give pulse rate and oxygen saturation readings. The two wavelengths used are differentially absorbed by oxygenated haemoglobin (HbO,) and de-oxygenated, or reduced, haemoglobin (Hb). Assuming there are no other major haemoglobin species, the amount that one wavelength is attenuated compared to the other gives the fraction of haemoglobin saturated by oxygen [HbO,]/([HbOJ i - [Hb]), where the square brackets denote concentration.Both wavelengths are also absorbed by venous blood and tissues. The attenuation of the light is analysed over a full pulse beat to make the saturation measurement independent of these factors. The total absorption of the light has a constant component from the tissue and from steadily flowing venous blood, and a changing component as a result of pulsation of arterial blood. The constant component is subtracted from the total, so that the net absorption of each wavelength can be attributed to arterial blood only. The two LEDs are cycled on and off 480 times per second (for a mains power frequency of 60 Hz) or 400 times per second (for a mains power frequency of 50 Hz), with only one being on at a time.2 This enables a single detector to be used to sample first one wavelength and then the other. The detector measures the background level of ambient light after the two LED 'on' periods. This is subtracted from the transmitted LED signals so that *

To derive accurate beam models for stereotactic radiosurgery (SRS) planning it is necessary to characterize the beam with dosimetric measurements. The aim of this study is to identify the best detectors for each task in the characterization process. Output ratios, beam profiles and percentage depth doses were measured for SRS cone diameters of 5-45 mm. Commercially available and emerging detectors were used: Gafchromic EBT2 film, an air-core fibre optic dosimeter (FOD) (developed at Royal Prince Alfred Hospital, Sydney), an IBA stereotactic field diode, a PTW 60012 electron diode and an IBA cc01 small volume thimble ion chamber. Analysis of the measured data supported by baseline Monte Carlo simulation data, led to the following recommendations: (1) water-equivalent detectors (Gafchromic EBT2 film or FOD) are the preferred choice for SRS dosimetry, (2) ion chambers (including small volume chambers with high-density central electrodes) should be avoided due to volume averaging effects and energy dependence, (3) if diodes are used, corrections must be made to account for their over-response in small fields.