Virtual Library

 x 
My Library Virtual Library FAQ

Start Your Search

Takashi Hayashi



Author of

  • +

    MA 14 - Diagnostic Radiology, Staging and Screening for Lung Cancer I (ID 672)

    • Event: WCLC 2017
    • Type: Mini Oral
    • Track: Radiology/Staging/Screening
    • Presentations: 1
    • +

      MA 14.02 - Simulation of the Four Rounds of NELSON Lung Cancer Screening Triage Algorithm (ID 9284)

      15:50 - 15:55  |  Presenting Author(s): Takashi Hayashi

      • Abstract
      • Presentation
      • Slides

      Background:
      Imaging screening programs are designed for specific eligibility criteria and technologies. Effectiveness of these programs is precisely known only after monitoring large populations over a long period. Ensuring generalizability of screening programs is required when targeting a population which is different from the original one tested or when modifying involved technologies. The NELSON screening program is a lung cancer triage process featuring four rounds of variable intervals (1, 2 and 2.5 years). Each screening round classifies the nodule’s malignancy according to the nodule’s volume, growth and volume doubling time, using CT. The aim of our study was to assess by simulation the influence of variable precision of measurement on the robustness of NELSON’s diagnostic algorithm.

      Method:
      We simulated 10[6] nodules using a Chi[2] distribution for nodule size [3mm; 20mm], an inverse Chi[ 2] distribution of growing. 2.1% of nodules were malignant (true positive). We tested several distributions of measurement error using a zero-mean Gaussian distribution and a standard deviation (SD) ranging [0%; 20%]. We reported positive and negative predictive values (PPV, NPV) at each round.

      Result:
      After round 4, we found that NPV decreased with increasing measurement error from 100% to 99.89%, PPV decreased from 100% to 29.6%. Figure 1 Figure 1: Detection performances of NELSON’s triage algorithm depending on measurement error. As shown in this graph, an increase of SD leads to a decrease of PPV (gray curve) and has almost no impact on NPV (yellow curve).



      Conclusion:
      Increasing measurement error of nodules significantly degrades the positive predictive value of NELSON’s diagnostic algorithm in identifying malignant pulmonary nodules, whereas the negative predictive value remained stable. We confirmed the efficacy of the successive rounds when measurement error is larger than 5%; however, the algorithm could be improved for larger measurement errors. Simulations could help us to assess better strategies lung in screening studies.

      Only 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, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.

      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.