Longitudinal Quantitative Assessment of COVID-19 Infection Progression from Chest CTs

Kim, Seong-Tae; Goli, Leili; Paschali, Magdalini; Khakzar, Ashkan; Keicher, Matthias; Czempiel, Tobias; Burian, Egon; Braren, Rickmer; Navab, Nassir; Wendler, Thomas

doi:10.1007/978-3-030-87234-2_26

Cited by 7 publications

(8 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Longitudinal segmentation, which uses two concatenated scans as input to exploit both spatial and temporal cues (spatio-temporal) cues, has shown very promising results for COVID-19 analysis [9,14]. The major limitation of longitudinal segmentation is the limited availability of annotated training data to sufficiently learn the underlying distribution, resulting in performance and generalization drops.…”

Section: Longitudinal Self-supervised Learningmentioning

confidence: 99%

“…We, therefore, propose a new longitudinal self-supervised pretraining model based on the image restoration task. In this study, a baseline longitudinal network [9] is defined based on a fully convolutional DenseNet [15] consisting of a downsampling path with five transition down blocks and an upsampling path with five transition up blocks, respectively.…”

Section: Longitudinal Self-supervisionmentioning

confidence: 99%

“…Longitudinal segmentation. For the segmentation task, we follow Kim et al [9] as a baseline model, an early fusion network that using as input slices of two concatenated registered scans of a patient from two different time points. This allows the network to capture temporal changes between the two scans.…”

Section: Longitudinal Covid-19 Pathology Quantificationmentioning

confidence: 99%

“…Several medical studies have investigated the use of longitudinal data to improve segmentation performance in [7,8]. Kim et al [9] introduced a longitudinal model to analyze COVID-19 progression over time and monitor the recovery process and response to various medical treatments. However, deep learning approaches require a large amount of labeled data to learn the complex texture and boundaries of infections.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Longitudinal Self-Supervision for COVID-19 Pathology Quantification

Czempiel¹,

Rogers²,

Keicher³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Quantifying COVID-19 infection over time is an important task to manage the hospitalization of patients during a global pandemic. Recently, deep learning-based approaches have been proposed to help radiologists automatically quantify COVID-19 pathologies on longitudinal CT scans. However, the learning process of deep learning methods demands extensive training data to learn the complex characteristics of infected regions over longitudinal scans. It is challenging to collect a large-scale dataset, especially for longitudinal training. In this study, we want to address this problem by proposing a new self-supervised learning method to effectively train longitudinal networks for the quantification of COVID-19 infections. For this purpose, longitudinal self-supervision schemes are explored on clinical longitudinal COVID-19 CT scans. Experimental results show that the proposed method is effective, helping the model better exploit the semantics of longitudinal data and improve two COVID-19 quantification tasks.

show abstract

Section: Longitudinal Self-supervised Learningmentioning

confidence: 99%

Section: Longitudinal Self-supervisionmentioning

confidence: 99%

Section: Longitudinal Covid-19 Pathology Quantificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Longitudinal Self-Supervision for COVID-19 Pathology Quantification

Czempiel¹,

Rogers²,

Keicher³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Longitudinal measurements can provide temporal information that comprehensively reflects dynamic changes of COVID-19 deterioration. So far, only a few studies explored temporal information from longitudinal CT or clinical measurements using deep learning (DL)-based methods (Fang et al, 2021;Kim et al, 2021;Pu et al, 2021;Wang et al, 2021a;Shamout et al, 2021). Zhou et al, (Zhou et al, 2021) used clinical features only to predict COVID-19 severity.…”

Section: Introductionmentioning

confidence: 99%