As mentioned above, the most direct way to measure absorbed dose is with an absorbed-dose calorimeter. The first generation of primary standards for absorbed dose were based on graphite calorimeters in which a small disk of graphite was thermally isolated from the surrounding graphite phantom and its temperature rise was measured when irradiated. The measuring system is calibrated by injecting a known amount of electrical energy into the same graphite disk and measuring the temperature rise. This system works very well because all of the energy absorbed in graphite is transformed into heat and hence into temperature rise. Also the heat diffuses very quickly so that there are no hot spots and one or two thermistors can get an accurate estimate of the average temperature rise. The only problem with this system is that the measured absorbed dose to graphite must be converted to an absorbed dose to water, but this can be done in several ways with an accuracy of better than 1%.
In 1980, Steve Domen of NIST (the US standards lab) proved that calorimetry can be done directly in a water tank[7]. Water has a very low thermal diffusivity, which means that the heat does not rapidly flow away from where it is created. Thus, absorbed dose to water measurements can be done at a point in a water tank without providing thermal isolation. This system has the obvious advantage of measuring the quantity of interest. There is one significant problem: irradiated water undergoes chemical reactions which can be endothermic or exothermic. This leads to what is called a thermal heat defect, i.e. a difference between the energy absorbed by the material from the radiation and the thermal energy released. The magnitude of the defect depends sensitively on the (sometimes) trace impurities in the system and on the accumulated dose and dose rate. However, in the last 15 years the understanding of the heat defect of irradiated water has improved dramatically and the heat defect in water-based systems is known to within a few tenths of a percent. This has allowed several primary standards based on water calorimetry to be developed at NRC and elsewhere[8].
To summarize, there are several major approaches to primary standards for
absorbed dose to water (besides the two described)
and various on-going comparisons between national primary standards
labs suggest that they are in agreement at the 1% level or
better[10][9].
These standards apply in 
beams and accelerator beams as
used in radiotherapy.
Thus, in principle a clinical physicist will send an ion chamber to NRC and have it directly calibrated in the type of beam in which it will be used. Then:
where 
is the absorbed dose to water calibration factor. The
only problem is that
clinically there may be many beams and each accelerator
calibration factor is very expensive to determine (likely $3,000 or
more).
For this reason, the AAPM is developing a new dosimetry protocol which
starts from an absorbed-dose calibration factor in a
beam and introduces a factor, called 
, which takes into account the
variation of with beam quality, Q, so that the calibration
factor in a beam of quality Q is given by:
While equations for can be derived in terms of the various
quantities introduced in previous protocols, it is conceptually simpler
to recognize it as a quantity which can be experimentally measured using
primary standards for beams of different quality Q. It varies by less
than 5% for different photon beam qualities and all clinical ion
chambers made of a given wall material are predicted to
have the same (within 0.4%) values of as a function of beam
quality[6].
Thus the new generation of protocols will consist of assigning dose based on:
where 
is measured at the primary standards
laboratory for each clinical chamber, 
is measured in the clinic
and is taken from a protocol but is based on measurements with
primary standards for absorbed dose. There are no other ``corrections''
needed. This is much simpler than the current procedures based on
eqn(6) and (7) which actually include between four and
six further ``corrections'' which were not included here to keep the
presentation ``simple''. The new approach is more coherent
because it uses only the quantity absorbed dose and ties the
clinically-assigned absorbed dose to primary standards in a corresponding
accelerator beam. In contrast, the exposure-based system is very
complex because: it requires conversion from exposure to absorbed dose;
it uses a protocol based on theory which
uses many factors which are not rigorously understood; and it
requires externally-determined values of parameters such as 
and
.
While the new approach appears to be giving nearly the same dose estimates in clinical photon beams as the old system, the added clarity and simplicity of the new absorbed-dose-based system should lead to fewer mistakes in the clinic, and should also allow more time to be spent developing innovative solutions to the myriad other problems faced by physicists delivering cancer radiotherapy.
