Virtual Library

Start Your Search

Xin Yang



Author of

  • +

    MA10 - Emerging Technologies for Lung Cancer Detection (ID 129)

    • Event: WCLC 2019
    • Type: Mini Oral Session
    • Track: Screening and Early Detection
    • Presentations: 1
    • Now Available
    • +

      MA10.02 - Deep Learning with Radiomics May Predict High-Grade Lung Adenocarcinoma Based on Histological Patterns in Ground Glass Opacity Lesions (Now Available) (ID 128)

      15:15 - 16:45  |  Author(s): Xin Yang

      • Abstract
      • Presentation
      • Slides

      Background

      Adenocarcinoma (ADC) is the most c­­­­­­­ommon histological subtype of lung cancers in non small cell lung cancer (NSCLC) in which ground glass opacifications (GGOs) found on computed tomography (CT) scans are the most common lesions. These lesions are usually treated with limited lung resection. However, the presence of a micropapillary or solid component is identified as an independent predictor of prognosis, indicating a more extensive resection. The accurate classification of subtypes still remains difficult in radiology or in frozen pathological analysis, even with the help of classical radiomics. The purpose of our study is to explore imaging phenotyping using a novel method combining radiomics with deep learning (RDL) to predict high-grade patterns within lung ADC.

      Method

      Included in this study were 111 patients differentiated as having GGOs and pathologically confirmed ADC. Four different methods were compared to classify the GGOs for the prediction of the pathological subtypes of high-grade lung ADCs, including classic machine learning, radiomics, deep learning method, and a proposed novel method referred as RDL. A four-fold cross-validation approach was used to evaluate the performance of such methods.

      Result

      We analyzed 32 patients with high-grade patterns and 79 without such patterns. The proposed RDL has achieved an overall accuracy of 0.888, which significantly outperforms classic machine learning, radiomics, and deep learning alone (p< 0.001, paired t-test).


      figure1.pngfigure2.png

      Conclusion

      High-grade lung ADC based on histologic pattern spectrum in GGO lesions might be predicted by a novel framework combining radiomics with deep learning, which reveals a significant advantage over traditional methods.

      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.

  • +

    MA15 - Usage of Computer and Molecular Analysis in Treatment Selection and Disease Prognostication (ID 141)

    • Event: WCLC 2019
    • Type: Mini Oral Session
    • Track: Pathology
    • Presentations: 1
    • Now Available
    • +

      MA15.02 - Deep Learning Approach for Automated Tumor Cells Detection and Estimation of PD-L1 22C3 Assay Expression in Lung Adenocarcinoma (Now Available) (ID 577)

      15:45 - 17:15  |  Author(s): Xin Yang

      • Abstract
      • Presentation
      • Slides

      Background

      It is vital and challenging to assess an accurate PD-L1 expression status on tumor cells for immunotherapy in lung cancer. The purpose of this study was to set up an automated system to detect the tumor cells and estimate the tumor proportion score (TPS) of PD-L1 immunohistochemistry (IHC) expression for lung adenocarcinoma based on deep learning, and provide a potential Artificial Intelligences (AI) assistive diagnostic tool in the quantification of PD-L1 interpretation.

      Method

      Fifty PD-L1 22C3 IHC slides of lung adenocarcinoma samples on digitized whole-slide images (WSI) database was employed. We first designed a model with a fully convolutional neural network (FCNN) based on U-ResNet architecture to obtain the cancer segmentation. Representative regions were selected from each slide, and 100 regions were collected for manual annotations as a training set for cancer detection. Another 50 regions were used to validate the performance of automated cancer detection and TPS estimation as a test set. After the quality control, a whole model of automated cancer cell segmentation and membrane positive estimation was set up on standard PD-L1 22C3 IHC staining. TPS could be automatically predicted by AI tool and then compared with the interpretations of pathologists.

      Result

      The results of automated lung adenocarcinoma cells segmentation on the test set of 22C3 IHC staining showed a moderate sensitivity (71.46%) with a high specificity (95.94%) which was much more crucial for TPS counting. In rest 43 out of 50 regions after a quality control, TPS estimated by the automated PD-L1 analysis based on cancer segmentation showed a significant correlation with the average scores (r=0.9609, p<0.001) and the median scores (r=0.9523, p<0.001) of pathologists' interpretations.

      results.png

      Conclusion

      We provide an automated tumor cells detection and TPS estimation model for lung adenocarcinoma and demonstrate the potential of using machine learning methods to access PD-L1 IHC status conveniently. A further validation of AI tool for automated scoring PD-L1 in diagnostic routine is highly recommended in the future.

      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.

  • +

    P2.09 - Pathology (ID 174)

    • Event: WCLC 2019
    • Type: Poster Viewing in the Exhibit Hall
    • Track: Pathology
    • Presentations: 1
    • Now Available
    • Moderators:
    • Coordinates: 9/09/2019, 10:15 - 18:15, Exhibit Hall
    • +

      P2.09-21 - A Prospective Study of the Concordance for PD-L1 Status in Core Needle Biopsy and Corresponding Resection Specimen in NSCLC  (Now Available) (ID 576)

      10:15 - 18:15  |  Author(s): Xin Yang

      • Abstract
      • Slides

      Background

      Some studies have demonstrated a relatively poor concordance of PD-L1 IHC expression between biopsies and corresponding resection specimens. To address this central and relevant issue having a significant impact on treatment stratification for patients with NSCLC, we evaluated a novel method to compare and evaluate PD-L1 status in biopsies and resection samples.

      Method

      Randomly core needle aspiration biopsy was performed in 170 resected NSCLC samples with 1 to 2 biopsies per centimeter in longest diameter of tumor. Among these 170 cases, a total of 52 cases were selected for the study. 41 cases were characterized as PD-L1 positive as a PD-L1 TPS≥ 1%), 1 case which the TPS<1% resected specimen with obvious stained tumor area and randomly 10 specimens being PD-L1 negative. In total 221 biopsies were available for the 52 resection cases. The PD-L1 expression in resected specimens and corresponding biopsies were evaluated by the PD-L1 IHC 22C3 pharmDx assay (Agilent) on the Dako Autostainer.

      Result

      In our investigation, the concordance of PD-L1 status in biopsy and resection was not influenced by number of tumor cells at 1% and 50% cut-off’s. The length of biopsy improves concordance of PD-L1 status but is not statistically significant (table 1). In figure 1. under 1% cut-off, the PD-L1 status in biopsy and resection is concordant irrespective of biopsies density, whereas the density of biopsies improves the concordance for 50% cut-off. The threshold was 1 per centimeter in longest diameter of tumor.

      table1.jpg

      figure1.jpg

      Conclusion

      The concordance between biopsy and resected specimen is related to the length, density of biopsy and the clinical cutoff. Increasing the biopsy density will improve accuracy of PD-L1 detection.

      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.