Prediction of tumor spread through air spaces with an automatic segmentation deep learning model in peripheral stage I lung adenocarcinoma

Abstract

Background: To evaluate the clinical applicability of deep learning (DL) models based on automatic segmentation in preoperatively predicting tumor spread through air spaces (STAS) in peripheral stage I lung adenocarcinoma (LUAD).

Methods: This retrospective study analyzed data from patients who underwent surgical treatment for lung tumors from January 2022 to December 2023. An external validation set was introduced to assess the model's generalizability. The study utilized conventional radiomic features and DL models for comparison. ROI segmentation was performed using the VNet architecture, and DL models were developed with transfer learning and optimization techniques. We assessed the diagnostic accuracy of our models via calibration curves, decision curve analysis, and ROC curves.

Results: The DL model based on automatic segmentation achieved an AUC of 0.880 (95% CI 0.780-0.979), outperforming the conventional radiomics model with an AUC of 0.833 (95% CI 0.707-0.960). The DL model demonstrated superior performance in both internal validation and external testing cohorts. Calibration curves, decision curve analysis, and ROC curves confirmed the enhanced diagnostic accuracy and clinical utility of the DL approach.

Conclusion: The DL model based on automatic segmentation technology shows significant promise in preoperatively predicting STAS in peripheral stage I LUAD, surpassing traditional radiomics models in diagnostic accuracy and clinical applicability. Clinical trial number The clinical trial was registered on April 22, 2024, with the registration number researchregistry10213 ( www.researchregistry.com ).

Keywords: Deep Learning; Lung Adenocarcinoma; Radiomics; Tumor Spread through Air Spaces.

Fig. 1

Screening flowchart for enrolled patients

Fig. 2

Workflow of this study

Fig. 3

Number and ratio of handcrafted features

Fig. 4

Coefficients of tenfold cross validation (A), MSE of tenfold cross validation (B), The histogram of the Rad-score based on the selected features (C)

Fig. 5

Metric results for Machine Learning Radiomics Signature. (A LR model; B SVM model; C RF model; D ExtraTrees model; E XGBoost model; F LightGBM model)

Fig. 6

ROC results for Radiomics Signature of different Machine Learning model. (A Cohort train AUC; B Cohort test AUC; C Cohort validation AUC)

Fig. 7

Metric results for Deep Learning Radiomics Signature (A densenet121 model; B resnet50 model; C resnet101 model)

Fig. 8

ROC results for Deep Learning Signature of different model (A Cohort train AUC; B Cohort test AUC; C Cohort validation AUC)

Fig. 9

Illustrates the ROC for different signatures across various cohorts, offering a visual comparison of their diagnostic abilities (A Cohort train AUC; B Cohort test AUC; C Cohort validation AUC)

Fig. 10

Displays the calibration curves for different signatures in the various cohorts. These curves are instrumental in understanding how well the predicted probabilities of the models match the actual outcomes (A Cohort train calibration curve; B Cohort test calibration curve; C Cohort validation calibration curve)

Fig. 11

Different signatures' DCA on various cohorts (A Cohort train DCA; B Cohort test DCA; C Cohort validation DCA)