Cartilage Segmentation in High-Resolution 3D Micro-CT Images via Uncertainty-Guided Self-training with Very Sparse Annotation

Zheng, Hao; Perrine, Susan M. Motch; Pitirri, M. Kathleen; Kawasaki, Kazuhiko; Wang, Chaoli; Richtsmeier, Joan T.; Chen, Danny Z.

doi:10.1007/978-3-030-59710-8_78

Cited by 24 publications

(23 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The two-phase approach involves automatically segmenting the chondrocranium with very sparse annotation to bridge the performance gap relative to full annotation and integrating limited human corrections to finetune the model. Our two-phase approach (supplemental materials) built on an automatic segmentation procedure (35) and produced fully 3D reconstructions of the chondrocranium at each embryonic age for Fgfr2c C342Y/+ mice and their Fgfr2c +/+ littermates, which were first evaluated for cartilage thickness using Figure 1. Morphology of the mouse embryonic chondrocranium and cranial skeleton.…”

Section: Resultsmentioning

confidence: 99%

“…We assess the effects of deviations in FGF/FGFR signaling on embryonic development of the chondrocranium and dermatocranium in Fgfr2c C342Y/+ mice whose skull phenotype parallels that of humans with Crouzon/Pfeiffer syndrome (21,30,32,33). The impact of this research is twofold: 1) the samples and novel methods for embryonic cartilage visualization (34) and deep learning based segmentation using uncertainty-guided self-training with very sparse annotation (35) allow us to address questions inaccessible in the study of humans but inform us about human craniofacial development and disease process; and 2) our 3D morphological analyses provide a unique opportunity for innovative evaluation of the effect of a mutation on embryonic cranial cartilage formation and on the relationship between chondrocranial cartilage and dermal bone formation. We anticipate one of three outcomes: i) chondrocranial morphology of Fgfr2c C342Y/+ mice and that of their controls, Fgfr2c +/+ (unaffected) littermates, are similar indicating that the mutation affects the cranial osteoblast lineage but not the chondrocyte series; ii) chondrocranial morphology separates Fgfr2c C342Y/+ mice from Fgfr2c +/+ littermates indicating that the mutation affects the chondrocyte series and the osteoblast lineage but the morphological effects of the mutation on the dermatocranium and the chondrocranium are dissociated; or iii) chondrocranial morphology separates Fgfr2c C342Y/+ mice from Fgfr2c +/+ littermates and the morphological effects of the mutation on chondrocranial cartilages and dermatocranial bone show correspondence, emphasizing integration of chondrocranial and dermatocranial development.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A dysmorphic mouse model reveals developmental interactions of chondrocranium and dermatocranium

Perrine

Pitirri²,

Kawasaki³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

The cranial endo- and dermal skeletons, which comprise the vertebrate skull, evolved independently and form separately during embryogenesis. In mammals, the mostly cartilaginous cranial endoskeleton forms prior to the bony dermatocranium. Many features of the chondrocranium are transient, undergoing endochondral ossification or disappearing, so its role in skull morphogenesis is not understood The fibroblast growth factor (FGF) and receptor (FGFR) signaling pathway contributes significantly to the regulation of osteochondroprogenitor cell function. Mutations in FGFR genes are associated with diseases that impact the skull including dwarfing chondrodyplasia and craniosynostosis syndromes. We investigate the developing chondrocranium and dermatocranium using a mouse model for craniosynostosis carrying a gain of function mutation in Fgfr2 to assess development of these cranial skeleton systems. Dermatocrania and chondrocrania of Fgfr2cC342Y/+ mice and their Fgfr2c+/+ littermates were quantified in 3D from microcomputed tomography images of mouse embryos. Chondrocrania of embryonic mice carrying the Fgfr2 mutation are larger than their Fgfr2c+/+ littermates and include novel extensions of cartilage over the lateral and dorsal aspect of the brain. Like the forming chondrocranium, the embryonic dermatocranium is larger in Fgfr2cC342Y/+ mice throughout embryogenesis but after disappearance of much of the chondrocranium, the dermatocranium becomes progressively smaller relative to Fgfr2c+/+ littermates during postnatal growth. Results reveal the direct effects of this Fgfr2c mutation on embryonic cranial cartilage, the impact of chondrocranial structure on developing dermatocranial elements, the importance of the chondrocranium in decoding the impact of specific genetic variants on head morphogenesis, and the potential for harnessing these effects as therapeutic targets.Significance StatementWe present the first fully complete three-dimensional (3D) reconstructions of the mouse embryonic chondrocranium using a novel methodology of uncertainty guided segmentation of microcomputed tomography images with sparse annotation. We provide 3D reconstructions of chondrocrania of the Fgfr2cC342Y/+ Crouzon syndrome mouse and typically developing littermates for embryonic days 13.5, 14.5, 15.5, 16.5, and 17.5. This is the first study of the effects of an FGFR2 mutation on embryonic chondrocranial cartilage. 3D reconstructions of embryonic dermal bones reveal that the dermatocranium develops outside of, and in shapes that conform to the chondrocranium. Our findings have implications for the study and treatment of human craniofacial disease and for understanding the impact of chondrocranial morphology on the evolution of skull morphology.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A dysmorphic mouse model reveals developmental interactions of chondrocranium and dermatocranium

Perrine

Pitirri²,

Kawasaki³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Volume segmentation with sparsely annotated slices. For 3D medical image segmentation, uniformly sampled slices with annotations were used in [220]- [224] to train a 3D deep network model by assigning a zero weight to unannotated voxels in the loss function. Bai et al [221] performed label propagation from annotated slices to unannotated slices based on non-rigid registration and introduce an exponentially weighted loss function for model training.…”

Section: Partially-supervised Segmentationmentioning

confidence: 99%

“…Specifically, they utilized the registration of consecutive 2D slices to propagate labels to unlabeled voxels. To segment 3D medical volumes with sparsely annotated 2D slices, Zheng et al [224] utilized uncertainty-guided self-training to gradually boost the segmentation accuracy. Before training segmentation models with sparsely annotated slices, Zheng et al [225] first identified the most influential and diverse slices for manual annotation with a deep network.…”

Section: Partially-supervised Segmentationmentioning

confidence: 99%

Medical Image Segmentation With Limited Supervision: A Review of Deep Network Models

Wang

2021

IEEE Access

View full text Add to dashboard Cite

Despite the remarkable performance of deep learning methods on various tasks, most cutting-edge models rely heavily on large-scale annotated training examples, which are often unavailable for clinical and health care tasks. The labeling costs for medical images are very high, especially in medical image segmentation, which typically requires intensive pixel/voxel-wise labeling. Therefore, the strong capability of learning and generalizing from limited supervision, including a limited amount of annotations, sparse annotations, and inaccurate annotations, is crucial for the successful application of deep learning models in medical image segmentation. However, due to its intrinsic difficulty, segmentation with limited supervision is challenging and specific model design and/or learning strategies are needed. In this paper, we provide a systematic and up-to-date review of the solutions above, with summaries and comments about the methodologies. We also highlight several problems in this field, discussed future directions observing further investigations.

show abstract

“…In healthcare, the authors in [15] point out issues related to the segmentation of craniofacial cartilage images. Labelling such images is very challenging since only experts can differentiate cartilages.…”

Section: Introductionmentioning

confidence: 99%

Convolutional Neural Networks Refitting by Bootstrapping for Tracking People in a Mobile Robot

Álvarez-Aparicio¹,

Guerrero‐Higueras²,

Calderita³

et al. 2021

Applied Sciences

View full text Add to dashboard Cite

Convolutional Neural Networks are usually fitted with manually labelled data. The labelling process is very time-consuming since large datasets are required. The use of external hardware may help in some cases, but it also introduces noise to the labelled data. In this paper, we pose a new data labelling approach by using bootstrapping to increase the accuracy of the PeTra tool. PeTra allows a mobile robot to estimate people’s location in its environment by using a LIDAR sensor and a Convolutional Neural Network. PeTra has some limitations in specific situations, such as scenarios where there are not any people. We propose to use the actual PeTra release to label the LIDAR data used to fit the Convolutional Neural Network. We have evaluated the resulting system by comparing it with the previous one—where LIDAR data were labelled with a Real Time Location System. The new release increases the MCC-score by 65.97%.

show abstract

Cartilage Segmentation in High-Resolution 3D Micro-CT Images via Uncertainty-Guided Self-training with Very Sparse Annotation

Cited by 24 publications

References 16 publications

A dysmorphic mouse model reveals developmental interactions of chondrocranium and dermatocranium

A dysmorphic mouse model reveals developmental interactions of chondrocranium and dermatocranium

Medical Image Segmentation With Limited Supervision: A Review of Deep Network Models

Convolutional Neural Networks Refitting by Bootstrapping for Tracking People in a Mobile Robot

Contact Info

Product

Resources

About