Author of

• 11890

  +

  P2.19 - Poster Session 2 - Imaging (ID 180)

  • Event: WCLC 2013
  • Type: Poster Session
  • Track: Imaging, Staging & Screening
  • Presentations: 1
  • Moderators:
  • Coordinates: 10/29/2013, 09:30 - 16:30, Exhibit Hall, Ground Level

  • 11893

    +

    P2.19-003 - A texture analysis approach to assess the severity of acute normal tissue changes in thoracic CT scans following radiation therapy for lung cancer (ID 970)

    09:30 - 09:30  |  Author(s): S.G. Armato
    • Abstract

    Background
    Quantitative analysis of thoracic computed tomography (CT) scans may be used to assess response to radiation therapy (RT). The purpose was to (1) identify a set of mathematical descriptors of image texture that correlate with the severity of radiologic normal lung tissue changes following RT and (2) quantitatively measure the extent of severity-dependent change in the values of these texture features.

    Methods
    Twenty-four patients who underwent definitive RT (median dose: 66 Gy, 2 Gy/fraction) for lung cancer were retrospectively identified. For each patient, three CT scans were collected: pre-treatment scan, post-treatment (median: 34 days, range: 8-82 days) scan, and RT planning scan with an associated dose map. Four dose regions (0-10 Gy, 10-30 Gy, 30-50 Gy, and >50 Gy) in the RT dose map were identified and mapped to the post-treatment scan through demons deformable registration. Sixty regions of interest (ROIs) distributed evenly among the four dose regions were automatically identified in the normal lung tissue of each post-treatment scan. An experienced radiologist compared the severity of change in each ROI from pre-treatment to post-treatment and categorized each as containing: 1) no abnormality, 2) mild abnormality, 3) moderate abnormality, or 4) severe abnormality. Twenty texture features that characterized gray-level intensity, region morphology, and gray-level distribution were calculated in all post-treatment ROIs and compared with anatomically matched ROIs mapped to the pre-treatment scan using demons registration. Fitted coefficients for the percent feature value change (ΔFV) between pre-treatment and post-treatment ROIs at each category of visible radiation damage were calculated using linear regression.

    Results
    Most ROIs contained no abnormality (66%) or mild abnormality (30%). ROIs with moderate (3%) or severe (1%) abnormalities were observed in 9 patients. For 19 of 20 features, significant differences (p<0.05) in ΔFV existed among ROIs assigned to various severity levels. For 12 features, significant differences in ΔFV existed among all four categories of radiation damage (see figure). Compared with regions with no abnormality, ΔFV for these 12 features increased, on average, by 1.6% (95% CI: 1.3-2%), 12% (95% CI: 9-15%), and 30% (95% CI: 19-41%), respectively, for the mild, moderate, and severe abnormality ROI categories.Figure 1

    Conclusion
    For 12 of the 20 features, ΔFV was significantly different among all categories of visible radiation damage. As severity increased, the mean feature value change from the pre-treatment scan also increased. In future studies, this approach may be used as a quantitative indicator of the severity of acute normal lung tissue damage following RT.

• 12736

  +

  P3.14 - Poster Session 3 - Mesothelioma (ID 197)

  • Event: WCLC 2013
  • Type: Poster Session
  • Track: Mesothelioma
  • Presentations: 1
  • Moderators:
  • Coordinates: 10/30/2013, 09:30 - 16:30, Exhibit Hall, Ground Level

  • 12746

    +

    P3.14-010 - Evaluation of Mesothelioma Tumor Thickness Measurement Variability (ID 2522)

    09:30 - 09:30  |  Author(s): S.G. Armato
    • Abstract

    Background
    Single time-point unidimensional tumor thickness measurements define measurable disease for clinical trial inclusion and also constitute a field in the IASLC prospective mesothelioma staging database. The modified RECIST guidelines for mesothelioma did not alter the 10mm minimum tumor measurement recommendation. It is unclear at what minimum thickness interobserver variability exceeds 20% of tumor thickness and how morphology, and location affect interobserver variability in a single baseline measurement.

    Methods
    105 thoracic CT scans were collected retrospectively from 50 mesothelioma patients. Each scan was reviewed by a medical oncologist, who identified 170 discrete sites of mesothelioma tumor across all scans that represented a range of thickness. Lesion morphology (concave rind, convex rind, convex mass, fusiform mass), and anatomic location (chest wall, mediastinum, anterior angle, or posterior angle; craniocaudal location; bone/soft tissue) were categorized. Using a custom computer interface (Abras), reference tumor thickness measurements were obtained by creating a line segment that spanned the tumor from the parietal tumor margin along the chest wall or mediastinal structures to the inner tumor margin. Measurements were selected to capture a range of tumor thicknesses and morphologies, rather than the most clinically relevant measurement sites. An observer study was conducted in which each of five other physicians was presented with the individual CT sections and the same digitally fixed location of the outer tumor margin at each of the 170 pre-defined tumor measurement sites. Each observer then independently created a line segment to capture tumor thickness at all measurement sites. Relative differences among the tumor thickness measurements of observers were estimated using a random-effects analysis of variance model (ANOVA) to identify the smallest tumor thickness at which linear measurements could be made reliably. Comparisons were made with the RECIST tumor response criterion of 20% for progression.

    Results
    The median mean measurement across all sites was 9.68mm, reflecting the study’s aim to investigate measurement variability as a function of tumor thickness, given that the current standard minimum is set at 10mm. The mean range of observer measurements across all sites was 15.1% of the mean per-site measurement (SD 9.1%). Measurements acquired at tumor sites with reference thickness less than 7.5 mm demonstrated inter-observer variability (as defined by the difference between the maximum and minimum measurements of the observers at each site) with a 75th percentile that included 20% of the tumor thickness, while measurements acquired at sites with mean tumor thickness >7.5mm showed inter-observer variability with a 75[th] percentile <20% of tumor thickness. Inter-observer variability tended to be lower for convex mass lesions relative to that associated with the other three tumor morphologies.

    Conclusion
    The results of this study have implications for the definition of minimum measurable tumor adopted by clinical trial and staging protocols.