I think direct measurement with anthropomorphic phantom is the best way, but this method needs expensive phantoms and is time consuming according to calibrate and read TLDs. Is an alternative method to asses organ doses with less difficulties?
Polymer gel dosimetry due that X-ray CT has been used to measure dose response and dose distributions in 3D the resulting images, which are reconstructed
from linear attenuation data, have Hounsfield units (HU) which are directly proportional to physical density changes of the polymer gel dosimeters....only that you need a anthropomorphic phantom or a simple water phantom and its holder.
If the organ you are interested in calculating the dose is in the fully present in the image, you can approximate the measured CTDi to organ dose. As I mentioned, the organ has tone fully within the image. CTDi is easy to calculate with a ion chamber and simple phantom.
If the organ is not within the scan range, I would suggest monte carlo for calculating not measuring the dose and as you mentioned TLDs for measuring. Just remember that TLDs have one simple limitation of not being able to record very small doses.
There are numerous published tabulated conversion factors to relate recorded dose length product (DLP) or CTDI to organ doses or effective dose, for example Lee et al, Med. Phys. 39 (4), April 2012. These need to be patient size and beam energy specific. You could try an existing CT dose calculation tool such as CT-Expo (Stamm and Nagel 2011) or imPACT (http://www.impactscan.org/ctdosimetry.htm).
You might want to look at the method developed by Walter Huda and me. Using the free-in-air isocenter kerma, we measured the ratio to organ doses in multiple anthropomorphic phantoms. Very easy approach to organ dose:
I add two matching papers from Zankl et al about dose conversion coefficients for adults and child with tube current modulation published in IOP Science.