Semiautomatic volume of interest drawing for (18)F-FDG image analysis-method and preliminary results

Functional imaging of cancer adds important information to the conventional measurements in monitoring response. Serial (18)F-fluorodeoxyglucose (FDG) positron emission tomography (PET), which indicates changes in glucose metabolism in tumours, shows great promise for this. However, there is a need for a method to quantitate alterations in uptake of FDG, which accounts for changes in tumour volume and intensity of FDG uptake. Selection of regions or volumes [ROI or volumes of interest (VOI)] by hand drawing, or simple thresholding, suffers from operator-dependent drawbacks.

We present a simple, robust VOI growing method for this application. The method requires a single seed point within the visualised tumour and another in relevant normal tissue. The drawn tumour VOI is insensitive to the operator inconsistency and is, thus, a suitable basis for comparative measurements. The method is validated using a software phantom. We demonstrate the use of the method in the assessment of tumour response in 31 patients receiving chemotherapy for various carcinomas.

Valid assessment of tumour response could be made 2-4 weeks after starting chemotherapy, giving information for clinical decision making which would otherwise have taken 9-12 weeks. Survival was predicted from FDG-PET 2-4 weeks after starting chemotherapy (p = 0.04) and after 9-12 weeks FDG-PET gave a better prediction of survival (p = 0.002) than CT or MRI (p = 0.015).

FDG-PET using this method of analysis has potential as a routine tool for optimising use of chemotherapy and improving its cost effectiveness. It also has potential for increasing the accuracy of response assessment in clinical trials of novel therapies.