Fabrication of a tissue characterization phantom from indigenous materials for computed tomography electron density calibration

Abstract

Background: Patient data for treatment planning are usually acquired from a computed tomography (CT) scanner. The acquired data are then downloaded into the treatment planning system (TPS). CT scanners use CT numbers in Hounsfield units to account for tissue inhomogeneities within the human body which are different from the parameters required by TPS to enable the dose computation algorithm of the TPS account for tissue heterogeneities in the dose computation process. TPS, however, requires radiological parameters such as relative electron densities (compared to that of water) of tissues to account for inhomogeneity corrections in radiation dose calculation. Such information can be entered in the treatment planning computers capable of reading CT images and used to enable accurate corrections for tissue heterogeneities on a pixel-by-pixel basis, if necessary. There is therefore a need to establish the correlation between the CT numbers and the relative electron densities, reDs empirically, by scanning a tissue characterisation phantom with CT scanner whose CT number to reD conversion is been determined.

This paper seeks to outline a procedure that can be use to fabricate tissue characterisation phantom from indigenous materials to minimize cost of purchasing a commercial one.

Materials and methods: A tissue characterisation phantom was constructed from 4 millimetre (mm) perspex (PMMA) sheets and 16 pieces of 20 millilitre (ml) plastic laboratory specimen collection containers. The tissue characterisation phantom was composed of two cylindrical phantoms designed to represent (mimic) the body and head of a standard adult human. The laboratory specimen collection containers were inserted into equally spaced holes arranged along rings on the circular surfaces of the phantoms, which were concentric to a central hole on each of the phantoms. This was done to ensure accuracy of the scanner's calibration. The centres of the holes on the rings were separated by 7.85 centimetre (cm) for the body phantom and 4.19 cm for the head phantom. The body phantom had nine holes; the head phantom had seven holes. The phantom assembly was air-tightly sealed but two holes with openings were created on each of the phantoms to facilitate filling of the phantoms with water. The effectiveness of the sealing was ascertained by subjecting the phantom through a pressure test to identify possible places of leakage along the bonded areas; leaking areas were amended. Locally available materials, which are rich in calcium, carbon, hydrogen and/or oxygen, and can simulate tissues found in the human body in terms of radiological properties, were sought. These materials were used to fill the specimen collection containers. The reDs of the materials were determined from CT scanning of the constructed phantoms filled with the materials with two different multi-slice CT scanners from different manufacturers. The reDs were confirmed with those obtained from measured linear attenuation coefficient with cobalt 60 beam for the materials. The constructed tissue characterisation phantom was used to calibrate General Electric (GE) LightSpeed® volume CT (VCT) scanners and the results compared to that of a commercial tissue characterisation phantom.

Result: Comparing the two approaches used for the determination of the reDs of the inserted materials of the fabricated tissue characterisation phantom, the reDs agreed with each other within ± 8 % (mean of 3.77 %; standard deviation of 3.01 %). The reDs of the constructed tissue characterisation phantom compared very well with that quoted by the manufacturer of the commercial phantom. The agreement of the reDs is within ± 6.6 % (mean of ± 3.27 %; standard deviation of ± 2.67 %).

Conclusion: The constructed tissue characterisation phantom compares favourably with the commercial one. In view of this the use of the constructed tissue characterisation phantom in clinical environment is recommendable.