Biological target volume overlapping segmentation system method for avoiding false-positive PET findings in assessing...

FDG PET/CT has a recognized high predictive power to assess the response to neoadjuvant chemoradiation therapy (CRT) in patients affected by locally advanced rectal cancer (LARC), but a relatively high number of false-positive findings decrease its specificity: with the aim to solve this problem, a new method of imaging analysis is here proposed.

The new method here described, named Biological target volume (BTV) Overlapping Segmentation System (BOSS), has been applied on 24 consecutive patients with LARC that were all previously classified as nonresponders to CRT by means of the response index criterion that is adopted in our center. The BOSS method is based on the quantification of the amount of superimposition between pretreatment and posttreatment BTV. All BTVpre was down using a threshold of 60% of SUVmax in the tumor (BTV60). The results (overlap volumes and percentage of overlap volumes) were then matched up with postoperative pathology classified by the Mandard's tumor regression grade (TRG) system.

Eleven patients were classified as responders (TRG1-2) and 13 as nonresponders (TRG 3-5). Among all the results obtained by BOSS method, only the percentage of overlap volume data between BTV60 and BTVpost (%Over_60) was able to correctly distinguish between responders and nonresponders. In our experience, a cutoff of 56% on the %Over_60 provided the best results in terms of true negative (11 cases), true positive (12 cases), false negative (1 case), and false positive (none).

This new method, we developed, appears able to unmask the false-positive cases, improving the specificity of FDG PET/CT to predict the response to CRT patients with LARC.