ESTRO 2020 Abstract book

S198 ESTRO 2020

OC-0345 Fully automated AI-based Non-Small Cell Lung Cancer detection and segmentation pipeline S. Primakov 1 , A. Ibrahim 1 , G. Wu 1 , S. Sanduleanu 1 , H. Gietema 2 , L. Hendriks 3 , O. Morin 4 , H. Woodruff 1 , P. Lambin 1 1 Maastricht University Medical Centre+, Precision Medicine, Maastricht, The Netherlands ; 2 Maastricht University Medical Centre+, Department of Radiology and Nuclear Medicine, Maastricht, The Netherlands ; 3 Maastricht University Medical Centre+, Department of Pulmonary Diseases, Maastricht, The Netherlands ; 4 University of California San Francisco, Department of Radiation Oncology, San Francisco, USA Purpose or Objective Localizing and delineating tumors is essential for radiotherapy planning and quantitative imaging workflows. However, manual contouring is highly laborious and time-consuming and prone to variability and poor reproducibility. To address these issues we created a fully automated pipeline for detecting and segmenting lung tumors on CT images to facilitate adaptive re-planning and automated RECIST. Material and Methods Multi-centric CT images from 1043 NSCLC patients with expert delineations of the gross tumor volume were used to train, test, and validate our detection and segmentation method. A three-step approach was developed, consisting of data pre-processing, lung isolation and tumor segmentation (Figure 1). A pre-processing algorithm was developed to standardize images within a heterogeneous dataset with regard to hardware and acquisition. A 2D U- net type convolutional neural network architecture with test time augmentation and volumetric post-processing was trained on 936 CT scans. The quantitative performance was evaluated on the remaining 107 patients using the Dice similarity coefficient (DSC), Jaccard index (J), and 95th Hausdorff distance (H95th). In addition to the quantitative analysis, we developed a qualitative assessment tool, which allows for the scoring of the generated segmentations while recording the assessor’s expertise level. For the qualitative assessment, we enrolled 22 participants (9 computer scientists, 8 medical doctors working in the field of medical imaging and 5 radiologists). A pilot in-silico clinical trial was performed in order to evaluate the time needed for contouring and resulting reproducibility of the contours.

Results The DSC and root-mean-square of the change in median ADC were on average 0.84 and 3.0%, respectively, for the RT-only patients (Table 1). These numbers were similar for the different treatment time points (1 and 2 weeks into RT). The inter-observer variability had a DSC of 0.86 whereas the network achieved a DSC of 0.89 in the same 5 patients. The network performed poorer on the patients receiving induction chemo, potentially due to chemo affecting the clarity of the tumour borders. The network performed well on the independent MR-Linac dataset (Fig 1) with an average DSC of 0.81 and a change in median ADC of 2.0%. This demonstrates its potential to generalise between systems. Conclusion We enabled automatic and accurate contouring of metastasized lymph nodes in the head and neck region on diagnostic and MR-Linac DWI images. This will help enable a feasible MR-Linac workflow with daily DWI imaging.

Results On the external validation dataset the proposed method achieved a median slice-wise detection accuracy of 0.99 (IQR=0.01), specificity of 0.99, and sensitivity of 0.90. The segmentation achieved an average DSC = 0.75 (median = 0.78, IQR =0.22), average Jaccard index J =0.62 (median