Transfer learning for data-efficient abdominal muscle segmentation with convolutional neural networks
AuthorsMcSweeney, D. M.
Henderson, E. G.
Van Herk, Marcel
Weaver, Jamie M
Bromiley, P. A.
AffiliationDivision of Cancer Sciences, University of Manchester, Manchester, UK. Radiotherapy Related Research, The Christie Foundation Trust, Manchester, UK. Department of Medical Oncology, The Christie Hospital NHS Foundation Trust, Manchester, UK. Division of Informatics, Imaging and Data Sciences, University of Manchester, Manchester, UK.
MetadataShow full item record
AbstractBackground Skeletal muscle segmentation is an important procedure for assessing sarcopenia, an emerging imaging biomarker of patient frailty. Data annotation remains the bottleneck for training deep learning auto-segmentation models. Purpose There is a need to define methodologies for applying models to different domains (e.g., anatomical regions or imaging modalities) without dramatically increasing data annotation. Methods To address this problem, we empirically evaluate the generalizability of various source tasks for transfer learning: natural image classification, natural image segmentation, unsupervised image reconstruction, and self-supervised jigsaw solving. Axial CT slices at L3 were extracted from PET-CT scans for 204 oesophago-gastric cancer patients and the skeletal muscle manually delineated by an expert. Features were transferred and segmentation models trained on subsets ( 𝑛=5,10,25,50,75,100,125 ) of the manually annotated training set. Four-fold cross-validation was performed to evaluate model generalizability. Human-level performance was established by performing an inter-observer study consisting of ten trained radiographers. Results We find that accurate segmentation models can be trained on a fraction of the data required by current approaches. The Dice similarity coefficient and root mean square distance-to-agreement were calculated for each prediction and used to assess model performance. Models pre-trained on a segmentation task and fine-tuned on 10 images produce delineations that are comparable to those from trained observers and extract reliable measures of muscle health. Conclusions Appropriate transfer learning can generate convolutional neural networks for abdominal muscle segmentation that achieve human-level performance while decreasing the required data by an order of magnitude, compared to previous methods ( 𝑛=160→10 ). This work enables the development of future models for assessing skeletal muscle at other anatomical sites where large annotated data sets are scarce and clinical needs are yet to be addressed.
CitationMcSweeney DM, Henderson EG, van Herk M, Weaver J, Bromiley PA, Green A, et al. Transfer learning for data‐efficient abdominal muscle segmentation with convolutional neural networks [Internet]. Medical Physics. Wiley; 2022.
- A Deep Learning Model to Automate Skeletal Muscle Area Measurement on Computed Tomography Images.
- Authors: Amarasinghe KC, Lopes J, Beraldo J, Kiss N, Bucknell N, Everitt S, Jackson P, Litchfield C, Denehy L, Blyth BJ, Siva S, MacManus M, Ball D, Li J, Hardcastle N
- Issue date: 2021
- Learning fuzzy clustering for SPECT/CT segmentation via convolutional neural networks.
- Authors: Chen J, Li Y, Luna LP, Chung HW, Rowe SP, Du Y, Solnes LB, Frey EC
- Issue date: 2021 Jul
- Deep neural network for automatic volumetric segmentation of whole-body CT images for body composition assessment.
- Authors: Lee YS, Hong N, Witanto JN, Choi YR, Park J, Decazes P, Eude F, Kim CO, Chang Kim H, Goo JM, Rhee Y, Yoon SH
- Issue date: 2021 Aug
- Semantic segmentation of cerebrospinal fluid and brain volume with a convolutional neural network in pediatric hydrocephalus-transfer learning from existing algorithms.
- Authors: Grimm F, Edl F, Kerscher SR, Nieselt K, Gugel I, Schuhmann MU
- Issue date: 2020 Oct
- Fully automatic multi-organ segmentation for head and neck cancer radiotherapy using shape representation model constrained fully convolutional neural networks.
- Authors: Tong N, Gou S, Yang S, Ruan D, Sheng K
- Issue date: 2018 Oct