Co-heterogeneous and Adaptive Segmentation from Multi-source and Multi-phase CT Imaging Data: A Study on Pathological Liver and Lesion Segmentation

Raju, Ashwin; Cheng, Chi‐Tung; Huo, Yuankai; Cai, Jinzheng; Huang, Junzhou; Xiao, Jing; Lü, Le; Liao, Chien‐Huang; Harrison, Adam P.

doi:10.1007/978-3-030-58592-1_27

Cited by 31 publications

(19 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This dual-PHNN baseline is very strong on our public data, achieving a DSC score of 96.9% on the D s test set. For discriminator, we use a 3D version of a popular architecture [16] using atrous convolution, which has proved a useful discriminator for liver masks [21]. We do not use the multi-level discriminator variant proposed by Tsai et al [26], as its added complexity does not seem to result in noticeable improvements for liver-based PADA [21].…”

Section: User-guided Domain Adaptationmentioning

confidence: 99%

“…For discriminator, we use a 3D version of a popular architecture [16] using atrous convolution, which has proved a useful discriminator for liver masks [21]. We do not use the multi-level discriminator variant proposed by Tsai et al [26], as its added complexity does not seem to result in noticeable improvements for liver-based PADA [21]. Specific details on the choice of hyper parameters are listed in the supplementary material.…”

Section: User-guided Domain Adaptationmentioning

confidence: 99%

See 1 more Smart Citation

User-Guided Domain Adaptation for Rapid Annotation from User Interactions: A Study on Pathological Liver Segmentation

Raju¹,

Ji²,

Cheng³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Mask-based annotation of medical images, especially for 3D data, is a bottleneck in developing reliable machine learning models. Using minimal-labor user interactions (UIs) to guide the annotation is promising, but challenges remain on best harmonizing the mask prediction with the UIs. To address this, we propose the user-guided domain adaptation (UGDA) framework, which uses prediction-based adversarial domain adaptation (PADA) to model the combined distribution of UIs and mask predictions. The UIs are then used as anchors to guide and align the mask prediction. Importantly, UGDA can both learn from unlabelled data and also model the high-level semantic meaning behind different UIs. We test UGDA on annotating pathological livers using a clinically comprehensive dataset of 927 patient studies. Using only extreme-point UIs, we achieve a mean (worst-case) performance of 96.1% (94.9%), compared to 93.0% (87.0%) for deep extreme points (DEXTR). Furthermore, we also show UGDA can retain this state-of-the-art performance even when only seeing a fraction of available UIs, demonstrating an ability for robust and reliable UI-guided segmentation with extremely minimal labor demands.

show abstract

Section: User-guided Domain Adaptationmentioning

confidence: 99%

Section: User-guided Domain Adaptationmentioning

confidence: 99%

User-Guided Domain Adaptation for Rapid Annotation from User Interactions: A Study on Pathological Liver Segmentation

Raju¹,

Ji²,

Cheng³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…While there are efforts towards SSL for landmark localization [5,18], they are based off of convolutional neural network (CNN)-only approaches, rather than the stateof-the-art GCN-based DAG. Some other prominent SSL successes in medical imaging have also been reported, such as for segmentation [2,15] and abnormality detection [22], but these are also CNN-based. Most other SSL methods are mainly developed for natural image classification tasks [7,8,1,19], which likewise do not address GCN-based landmark localization.…”

Section: Introductionmentioning

confidence: 98%

Scalable Semi-supervised Landmark Localization for X-ray Images using Few-shot Deep Adaptive Graph

Zhou¹,

Lai²,

Li³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Landmark localization plays an important role in medical image analysis. Learning based methods, including convolutional neural network (CNN) and graph convolutional network (GCN), have demonstrated the state-of-the-art performance. However, most of these methods are fully-supervised and heavily rely on manual labeling of a large training dataset. In this paper, based on a fully-supervised graph-based method, deep adaptive graph (DAG), we proposed a semi-supervised extension of it, termed few-shot DAG, i.e., five-shot DAG. It first trains a DAG model on the labeled data and then fine-tunes the pre-trained model on the unlabeled data with a teacher-student semi-supervised learning (SSL) mechanism. In addition to the semi-supervised loss, we propose another loss using Jensen-Shannon (JS) divergence to regulate the consistency of the intermediate feature maps. We extensively evaluated our method on pelvis, hand and chest landmark detection tasks. Our experiment results demonstrate consistent and significant improvements over previous methods.

show abstract

“…Many efforts have been developed for automating lesion size measurement. Specifically, deep convolutional neural networks are successfully applied to segment tumors in brain [12], lung [31,16,39], pancreas [41,37], liver [8,19,7,22,28,36], enlarged lymph node [20,40], etc. Most of these approaches are specifically designed for a certain lesion type, however, an effective and efficient lesion size measurement tool should be able to handle a variety of lesions in practice [3,26,1,29].…”

Section: Introductionmentioning

confidence: 99%

Weakly-Supervised Universal Lesion Segmentation with Regional Level Set Loss

Tang¹,

Cai²,

Yan³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Accurately segmenting a variety of clinically significant lesions from whole body computed tomography (CT) scans is a critical task on precision oncology imaging, denoted as universal lesion segmentation (ULS). Manual annotation is the current clinical practice, being highly time-consuming and inconsistent on tumor's longitudinal assessment. Effectively training an automatic segmentation model is desirable but relies heavily on a large number of pixel-wise labelled data. Existing weaklysupervised segmentation approaches often struggle with regions nearby the lesion boundaries. In this paper, we present a novel weakly-supervised universal lesion segmentation method by building an attention enhanced model based on the High-Resolution Network (HRNet), named AHR-Net, and propose a regional level set (RLS) loss for optimizing lesion boundary delineation. AHRNet provides advanced high-resolution deep image features by involving a decoder, dual-attention and scale attention mechanisms, which are crucial to performing accurate lesion segmentation. RLS can optimize the model reliably and effectively in a weaklysupervised fashion, forcing the segmentation close to lesion boundary. Extensive experimental results demonstrate that our method achieves the best performance on the publicly large-scale DeepLesion dataset and a hold-out test set.

show abstract

Co-heterogeneous and Adaptive Segmentation from Multi-source and Multi-phase CT Imaging Data: A Study on Pathological Liver and Lesion Segmentation

Cited by 31 publications

References 37 publications

User-Guided Domain Adaptation for Rapid Annotation from User Interactions: A Study on Pathological Liver Segmentation

User-Guided Domain Adaptation for Rapid Annotation from User Interactions: A Study on Pathological Liver Segmentation

Scalable Semi-supervised Landmark Localization for X-ray Images using Few-shot Deep Adaptive Graph

Weakly-Supervised Universal Lesion Segmentation with Regional Level Set Loss

Contact Info

Product

Resources

About