Development of disease-specific software lesions to evaluate and optimise the performance of image-based dosimetry in radionuclide therapy

Abstract

Background: Since 1940, iodine-131 has been used in radionuclide therapy to treat differentiated thyroid carcinoma. In recent years, nuclear medicine scans are being used to determine the dose delivered to the lesion. Internal dosimetry involves a series of acquisitions and conversions which introduce errors in the dosimetry results. Correction algorithms and software are aimed at making dosimetry more accurate but assessing the improved accuracy is not straightforward. Objectives: This study aimed at establishing a methodology to test imaging processing techniques and software used in image-based dosimetry. Retrospective data about the characteristics of differentiated thyroid carcinomas was collected. Artificial lesions will be created based on the retrospective tumour characteristics. One software and one correction method used in planar image based dosimetry was evaluated as proof of concept. Research Design: The study involves the generation of chest lesions using MATLAB. The methodology involved the generation of the ROI, count value, maximum count and the uniformity of the lesion. For the second part of the project, an anthropomorphic phantom study was conducted having small lesions inside the lung with known activity. Results: The retrospective tumour characteristics were presented and discussed. The artificial tumour generation code was tested by comparing the retrospective tumours with the generated tumours. The results show that the generated tumours are in general, clinically representative. The phantom images were used to establish the sensitivity. The assessment of a dosimetry software was done using the phantom images. Also, a scatter correction method was assessed, comparing the non-corrected with the corrected results. Conclusions and recommendations: The study showed that the majority of metastasised thyroid carcinoma is within the chest and that synthetic lesions can be created provided that a large dataset is available. It was found that the dosimetry software overestimated the washout rate. Further studies should aim to test SPECT-based dosimetry and use a larger retrospective dataset.