Virtual Library

Start Your Search

Hyunjin Park



Author of

  • +

    P1.16 - Treatment of Early Stage/Localized Disease (Not CME Accredited Session) (ID 948)

    • Event: WCLC 2018
    • Type: Poster Viewing in the Exhibit Hall
    • Track:
    • Presentations: 1
    • Moderators:
    • Coordinates: 9/24/2018, 16:45 - 18:00, Exhibit Hall
    • +

      P1.16-10 - Marginal Features Analyses of Lung Adenocarcinoma for Survival Prediction (ID 12587)

      16:45 - 18:00  |  Author(s): Hyunjin Park

      • Abstract
      • Slides

      Background

      Tumor microenvironment is a complex mixture of assorted cells and extra-cellular components which make up an amazingly dynamic area that includes signaling interactions between cancer cells and their surrounding tissue. Tumor microenvironment makes up the peripheral portion of the tumor and major changes in this area has been reported to be associated with a poor prognosis. However, very few studies have investigated the tumor marginal features quantitatively extracted from CT images using a radiomics approach. We aimed to clarify the relationship between tumor marginal features and the micropapillary pattern and correlated with survival.

      a9ded1e5ce5d75814730bb4caaf49419 Method

      We enrolled 334 patients who underwent complete resection for lung adenocarcinoma. Quantitative histologic subtyping was performed for the whole tumor. Using a radiomics approach, quantitative CT analysis was performed and 82 marginal features were extracted. Clinical variables and marginal features were correlated with survival. Using selected clinical variables and marginal features a prognostic model was calculated with subsequent internal and external validation.

      4c3880bb027f159e801041b1021e88e8 Result

      Among various subtypes, solid predominant adenocarcinomas had the lowest proportion (6.9%) of combined micropapillary pattern. At univariate analysis, patient age, tumor size, and multiple marginal features (convexity, surface area, compactness, maximum 3D diameter, sphericity, surface-to-volume ratio, mean pixel value, median pixel value, entropy, uniformity, skewness, kurtosis, roundness factor, solidity, and lacunarity)were predictive of survival. At multivariate cox proportional analysis, convexity (P=0.017), kurtosis (P<0.001), and patient age (P=0.006) were identified as being predictive of survival. Ten-fold cross-validation tests demonstrated that our prediction model significantly classified patients according to survival (P<0.001). Although lower than internal validation, the prediction model also worked at external validation.figure.jpg

      8eea62084ca7e541d918e823422bd82e Conclusion

      Marginal radiomics features of convexity and kurtosis reflect the tumor microenvironment and were predictive of patient survival in lung adenocarcinomas.

      6f8b794f3246b0c1e1780bb4d4d5dc53

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    P2.01 - Advanced NSCLC (Not CME Accredited Session) (ID 950)

    • Event: WCLC 2018
    • Type: Poster Viewing in the Exhibit Hall
    • Track:
    • Presentations: 1
    • Moderators:
    • Coordinates: 9/25/2018, 16:45 - 18:00, Exhibit Hall
    • +

      P2.01-10 - Prognostic Impact of Longitudinal Monitoring of Radiomic Features in Patients with Advanced Non-Small Cell Lung Cancer (ID 12877)

      16:45 - 18:00  |  Author(s): Hyunjin Park

      • Abstract
      • Slides

      Background

      Tumor growth dynamics varies substantially in non-small cell lung cancer (NSCLC). We aimed to develop novel biomarkers reflecting longitudinal change of radiomic features in NSCLC and evaluate prognostic power of those.

      a9ded1e5ce5d75814730bb4caaf49419 Method

      Fifty-three patients with advanced NSCLC included in this retrospective study. Measurable lesions on baseline and follow-up computed tomography (CT) were segmented and 23 radiomic features were extracted. All three variables reflecting patterns of longitudinal change were extracted: the area under the curve (AUC), beta value, and AUC2. We constructed models for predicting survival using multivariate cox regression, and identified the performance of these models.

      4c3880bb027f159e801041b1021e88e8 Result

      In volume, AUC2 showed an excellent correlation with pattern of longitudinal volume change (r = 0.848, p < 0.000), and showed a significant difference in overall survival time (p = 0.035). In multivariate regression analysis, kurtosis of positive pixel values (p < 0.000), and surface area (p = 0.001) on baseline CT, and AUC2 of density (p < 0.000), skewness of positive pixel values (p = 0.003), and entropy at inner (p = 0.001) were found to be associated with overall survival time, and the area under the receiver operating characteristics curves were 0.922, and 0.771 at 1 year, and 3 years of follow-up.

      8eea62084ca7e541d918e823422bd82e Conclusion

      Longitudinal change of radiomic tumor features would be prognostic biomarkers in patients with advanced NSCLC.

      6f8b794f3246b0c1e1780bb4d4d5dc53

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.