Yes, you can.

One simple method is to use known dose rates with various sources and then develop your own conversion factors for bands of energy.

There are more detailed methods, for example

https://hps.org/publicinformation/ate/q8371.html

http://www.ncbi.nlm.nih.gov/pubmed/26952947

http://web.ornl.gov/info/reports/1981/3445600490877.pdf

http://www.aber.ac.uk/temp-ancient-tl/issue25_1/mercier_atl25(1)_1-4.pdf

You can find these and others by searching the internet.

Assuming that you measure photons, the theory tells that the absorbed dose is, at electronic equilibrium, equal to the kerma, which itself is connected to the spectral distribution of the fluence (see formula in attachd file). As fluence is the number of particles per unit area, the spectral ditribution of the fluence is somehow the spectrum per unit area. But this is the spectrum of the incoming photons, which is not the measured spectrum. It is possible to unfold the measured spectrum for reconstructing the  photon spectrum, if you know the response matrix of your detector (usually calculated by a Monte Carlo calculation). Then you can calculate the Kerma in air or the absorded dose in tissues by using the appropriate µtr factor (it is the same as µen in  .http://physics.nist.gov/PhysRefData/XrayMassCoef/tab4.html )

Equivalently, fluence-to-dose conversion factors can be found on the Internet, but they also apply to the incoming photons.

It is of course possible to calculate directly the absorbed dose in your detector by simply multiplying each channel of your spectrum with its energy and summing on all channels (this will give you the total energy transferred to the detector) , then divide by the mass. But there is probably no interest in the dose to CsI; this method would only be useful for a tissue-equivalent detector.

Inorganic scintillators like NaI are sometimes used as approximate dosimeters. In order to obtain an approximate proportionality between the doserate and the countrate in an limited energy range, the dosimeter should be "compensated", i.e. the thickness and composition of the metal wall around the crystal must be specifically designed for this use.

D = A x Γ / d2

Where,

D is the exposure dose rate from a radionuclide source (mSv h-1)

with activity A (MBq) at a distance d (cm).

Γ is the specific exposure dose constant of the radionuclide at a

distance of 1 m (i.e. for 137Cs is1.03×104 mSv h-1m2 MBq-1).

The dose is the integral of the spectrum times the flux-to-dose conversion factor.  The latter is available in the literature.

it's so simple to measure the activity (Bq/kg or other unit  )of samples first. Then you can convert the activity to dose . Then depending on the type of radionulide you can convert to dose .for food water samples the dose depends on the uptake.