Back to the list
Congress: ECR25
Poster Number: C-14717
Type: Poster: EPOS Radiologist (scientific)
Authorblock: A. Olesinski, R. Lederman, J. Sosna, L. Joskowicz; Jerusalem/IL
Disclosures:
Alon Olesinski: Consultant: HighRAD
Richard Lederman: Consultant: HighRAD
Jacob Sosna: Consultant: HighRAD
Leo Joskowicz: Consultant: HighRAD and Ezra
Keywords: Artificial Intelligence, Lymph nodes, Oncology, CT, CAD, Computer Applications-Detection, diagnosis, Cancer
Purpose Accurate assessment of mediastinal lymph nodes (LNs) in contrast-enhanced CT (ceCT) scans is essential for cancer staging and treatment planning. Current guidelines require measurement of enlarged lymph nodes whose short axis length (SAL) is >10mm. Manual detection and measurement of LNs is, however, time-consuming and subject to observer variability. We have developed a novel annotation-efficient semi-supervised deep learning method for automatic detection and segmentation of mediastinal lymph nodes in ceCT scans.The aim of ths study is to evaluate the performance of a novel...
Read more Methods and materials The method consists of a 3D nnU-Net voxel classification model for LN detection (Fig. 1). It is a semi-supervised method combines the precision of expert annotations with the quantity advantages of pseudo-labeled data. It uses an ensemble of 3D nnU-Net models trained on a few expert-annotated scans to generate pseudo-labels on a large dataset of unannotated scans. The pseudo-labels are then filtered to remove false positive LNs by excluding LNs outside the mediastinum and LNs overlapping with other anatomical structures. Finally, a single 3D...
Read more Results The enhanced 3D nnU-Net yielded a detection precision and recall (std) of 0.85 (0.26) and 0.89 (0.24) for LNs > 10mm and 0.73 (0.29) and 0.72 (0.28) for LNs 5-10mm. It significantly (p<0.01) improved the recall(std) by 6 (14)% and 15 (30)% with respect to the initial model, with similar precision. The mean short axis difference(std) was 4.6 (4.7)mm and 2.3(3.2)mm respectively.
Read more Conclusion Our method requires x4-10 less annotated data to achieve a performance similar to supervised models trained on the same dataset. Label-efficient automatic detection and measurement of mediastinum lymph nodes in chest CECT yields acurrate results and may help in the evaluation of patients with enlarged mediastinal LNs.
Read more References 1. Mountain CF, Dresler CM. 1997. Regional lymph node classification for lung cancer staging. Chest 111(6):1718-23.2. Schwartz LH, Bogaerts J, Ford R, Shankar L, Therasse P, Gwyther S, Eisenhauer EA. 2009. Evaluation of lymph nodes with RECIST 1.1. European J. Cancer 45(2):261-7.3. Isensee F, Jaeger PF, Kohl SA, Petersen J, Maier-Hein KH. 2021. nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nature Methods 18(2):203-11.4. Wasserthal J, Breit HC, Meyer MT, Pradella M, Hinck D, Sauter AW, Heye T, Boll...
Read more
GALLERY