Les Roger Folio DO, MPH, Presenter: Nothing to Disclose

Michael Choi, Abstract Co-Author: Nothing to Disclose

Jeff Solomon PhD, Abstract Co-Author: Nothing to Disclose

Nicholas Schaub, Abstract Co-Author: Nothing to Disclose

Serial CT imaging of metastatic tumors is vital in evaluating the efficacy of cancer treatment. The current system to evaluate tumor burden (RECIST) is subjective, time consuming and not routinely included in radiology reports.

Patients treated at the National Cancer Institute in IRB approved studies for metastatic melanoma were retrospectively reviewed. Seventy-one lung, liver, and subcutaneous lesions in 19 patients were manually measured using RECIST criteria prior to therapy (baseline CT) and within three months after therapy (follow-up CT). Semi-automated registration, segmentation, and RECIST measurements at both time points were performed using a new lesion application in PACS (Carestream Health, Rochester, NY); the software automatically segments and measures lesions on the follow-up CT using the baseline CT as reference. We compared manual and software-generated RECIST measurements using Bland-Altman plots.

The median manually-measured RECIST diameter for all lesions at baseline was 2.1 (1.0-6.2) cm. The refine registration function identified 69/71 lesions on the follow-up CT. On the baseline CT, all 18 liver, 27/32 (84%) lung, and 10/18 (55%) subcutaneous lesions segmented correctly. On the follow-up CT, 16/18 (80%) liver, 21/27 (78%) lung, and 8/10 (80%) subcutaneous lesions segmented correctly. The Bland-Altman plot demonstrated that software-generated RECIST measurements showed no bias with a 95% confidence interval of +/- 0.7cm, with minor differences based on anatomic location (liver +/- 0.9cm, lung +/-0.6cm, subcutaneous +/-0.3cm. There was no measurement bias between the baseline and follow-up CT.

Semi-automated PACS system accurately measured lesions within +/-0.7cm. Sixty-six percent of lesions segmented correctly on the follow-up CT using the baseline CT as reference.

Semi-automated metastatic lesion measurement should improve efficiency and expand availability of effective tumor assessment.

Folio, L, Choi, M, Solomon, J, Schaub, N, Semi-automatic Target Lesion Localization, Segmentation and RECIST Measurements on Serial CT Studies.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12021009.html