Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation
Medical imaging data annotation is expensive and time-consuming. Supervised deep learning approaches may encounter overfitting if trained with limited medical data, and further affect the robustness of computer-aided diagnosis (CAD) on CT scans collected by various scanner vendors. Additionally, the...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Frontiers Media S.A.
2022-12-01
|
Series: | Frontiers in Radiology |
Subjects: | |
Online Access: | https://www.frontiersin.org/articles/10.3389/fradi.2022.1041518/full |
_version_ | 1811197644701171712 |
---|---|
author | Jingya Liu Liangliang Cao Oguz Akin Yingli Tian |
author_facet | Jingya Liu Liangliang Cao Oguz Akin Yingli Tian |
author_sort | Jingya Liu |
collection | DOAJ |
description | Medical imaging data annotation is expensive and time-consuming. Supervised deep learning approaches may encounter overfitting if trained with limited medical data, and further affect the robustness of computer-aided diagnosis (CAD) on CT scans collected by various scanner vendors. Additionally, the high false-positive rate in automatic lung nodule detection methods prevents their applications in daily clinical routine diagnosis. To tackle these issues, we first introduce a novel self-learning schema to train a pre-trained model by learning rich feature representatives from large-scale unlabeled data without extra annotation, which guarantees a consistent detection performance over novel datasets. Then, a 3D feature pyramid network (3DFPN) is proposed for high-sensitivity nodule detection by extracting multi-scale features, where the weights of the backbone network are initialized by the pre-trained model and then fine-tuned in a supervised manner. Further, a High Sensitivity and Specificity (HS2) network is proposed to reduce false positives by tracking the appearance changes among continuous CT slices on Location History Images (LHI) for the detected nodule candidates. The proposed method’s performance and robustness are evaluated on several publicly available datasets, including LUNA16, SPIE-AAPM, LungTIME, and HMS. Our proposed detector achieves the state-of-the-art result of 90.6% sensitivity at 1/8 false positive per scan on the LUNA16 dataset. The proposed framework’s generalizability has been evaluated on three additional datasets (i.e., SPIE-AAPM, LungTIME, and HMS) captured by different types of CT scanners. |
first_indexed | 2024-04-12T01:17:09Z |
format | Article |
id | doaj.art-f6efb7cfd31e470d93b5310e2c435098 |
institution | Directory Open Access Journal |
issn | 2673-8740 |
language | English |
last_indexed | 2024-04-12T01:17:09Z |
publishDate | 2022-12-01 |
publisher | Frontiers Media S.A. |
record_format | Article |
series | Frontiers in Radiology |
spelling | doaj.art-f6efb7cfd31e470d93b5310e2c4350982022-12-22T03:53:55ZengFrontiers Media S.A.Frontiers in Radiology2673-87402022-12-01210.3389/fradi.2022.10415181041518Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptationJingya Liu0Liangliang Cao1Oguz Akin2Yingli Tian3The City College of New York, New York, NY, USAUMass CI, Amherst, MA, USAMemorial Sloan Kettering Cancer Center, New York, NY, USAThe City College of New York, New York, NY, USAMedical imaging data annotation is expensive and time-consuming. Supervised deep learning approaches may encounter overfitting if trained with limited medical data, and further affect the robustness of computer-aided diagnosis (CAD) on CT scans collected by various scanner vendors. Additionally, the high false-positive rate in automatic lung nodule detection methods prevents their applications in daily clinical routine diagnosis. To tackle these issues, we first introduce a novel self-learning schema to train a pre-trained model by learning rich feature representatives from large-scale unlabeled data without extra annotation, which guarantees a consistent detection performance over novel datasets. Then, a 3D feature pyramid network (3DFPN) is proposed for high-sensitivity nodule detection by extracting multi-scale features, where the weights of the backbone network are initialized by the pre-trained model and then fine-tuned in a supervised manner. Further, a High Sensitivity and Specificity (HS2) network is proposed to reduce false positives by tracking the appearance changes among continuous CT slices on Location History Images (LHI) for the detected nodule candidates. The proposed method’s performance and robustness are evaluated on several publicly available datasets, including LUNA16, SPIE-AAPM, LungTIME, and HMS. Our proposed detector achieves the state-of-the-art result of 90.6% sensitivity at 1/8 false positive per scan on the LUNA16 dataset. The proposed framework’s generalizability has been evaluated on three additional datasets (i.e., SPIE-AAPM, LungTIME, and HMS) captured by different types of CT scanners.https://www.frontiersin.org/articles/10.3389/fradi.2022.1041518/fullself-supervised learninglung nodule detectionfalse positive reductionfeature pyramid networkmedical image analysisdeep learning |
spellingShingle | Jingya Liu Liangliang Cao Oguz Akin Yingli Tian Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation Frontiers in Radiology self-supervised learning lung nodule detection false positive reduction feature pyramid network medical image analysis deep learning |
title | Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation |
title_full | Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation |
title_fullStr | Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation |
title_full_unstemmed | Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation |
title_short | Robust and accurate pulmonary nodule detection with self-supervised feature learning on domain adaptation |
title_sort | robust and accurate pulmonary nodule detection with self supervised feature learning on domain adaptation |
topic | self-supervised learning lung nodule detection false positive reduction feature pyramid network medical image analysis deep learning |
url | https://www.frontiersin.org/articles/10.3389/fradi.2022.1041518/full |
work_keys_str_mv | AT jingyaliu robustandaccuratepulmonarynoduledetectionwithselfsupervisedfeaturelearningondomainadaptation AT liangliangcao robustandaccuratepulmonarynoduledetectionwithselfsupervisedfeaturelearningondomainadaptation AT oguzakin robustandaccuratepulmonarynoduledetectionwithselfsupervisedfeaturelearningondomainadaptation AT yinglitian robustandaccuratepulmonarynoduledetectionwithselfsupervisedfeaturelearningondomainadaptation |