Accelerated muscle mass estimation from CT images through transfer learning

Background The cost of labeling to collect training data sets using deep learning is especially high in medical applications compared to other fields. Furthermore, due to variances in images depending on the computed tomography (CT) devices, a deep learning based segmentation model trained with a ce...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:BMC medical imaging Jg. 24; H. 1; S. 271 - 18
Hauptverfasser: Yoon, Seunghan, Kim, Tae Hyung, Jung, Young Kul, Kim, Younghoon
Format: Journal Article
Sprache:Englisch
Veröffentlicht: London BioMed Central 09.10.2024
BioMed Central Ltd
Springer Nature B.V
BMC
Schlagworte:
Abdomen
Analysis
Artificial intelligence
Comparative analysis
Computed tomography
Convolutional neural network
CT image segmentation
CT imaging
Datasets
Deep Learning
Diagnostic imaging
Humans
Image processing
Image Processing, Computer-Assisted - methods
Image segmentation
Imaging
Knowledge transfer
Labeling
Liver
Machine learning
Medical image segmentation
Medical imaging
Medical imaging equipment
Medicine
Medicine & Public Health
Methods
Muscle, Skeletal - diagnostic imaging
Muscles
Neural networks
Performance degradation
Performance evaluation
Radiology
Technology application
Tomography, X-Ray Computed - methods
Transfer learning
South Korea
Deep learning
Convolutional neural network
Medical image segmentation
CT image segmentation
ISSN:1471-2342, 1471-2342
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Background The cost of labeling to collect training data sets using deep learning is especially high in medical applications compared to other fields. Furthermore, due to variances in images depending on the computed tomography (CT) devices, a deep learning based segmentation model trained with a certain device often does not work with images from a different device. Methods In this study, we propose an efficient learning strategy for deep learning models in medical image segmentation. We aim to overcome the difficulties of segmentation in CT images by training a VNet segmentation model which enables rapid labeling of organs in CT images with the model obtained by transfer learning using a small number of manually labeled images, called SEED images. We established a process for generating SEED images and conducting transfer learning a model. We evaluate the performance of various segmentation models such as vanilla UNet, UNETR, Swin-UNETR and VNet. Furthermore, assuming a scenario that a model is repeatedly trained with CT images collected from multiple devices, in which is catastrophic forgetting often occurs, we examine if the performance of our model degrades. Results We show that transfer learning can train a model that does a good job of segmenting muscles with a small number of images. In addition, it was confirmed that VNet shows better performance when comparing the performance of existing semi-automated segmentation tools and other deep learning networks to muscle and liver segmentation tasks. Additionally, we confirmed that VNet is the most robust model to deal with catastrophic forgetting problems. Conclusion In the 2D CT image segmentation task, we confirmed that the CNN-based network shows better performance than the existing semi-automatic segmentation tool or latest transformer-based networks.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:1471-2342
1471-2342
DOI:10.1186/s12880-024-01449-4