End-to-End Cell Recognition by Point Annotation

Shui, Zhongyi; Zhang, Shichuan; Zhu, Chenglu; Wang, Bingchuan; Chen, Pingyi; Zheng, Sunyi; Yang, Lin

doi:10.1007/978-3-031-16440-8_11

Cited by 5 publications

(5 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the following parts, we first compare the capability of DPA-P2PNet with the state-of-the-art PCD and crowd localization competitors, which encompass density map (DM)based approaches containing U-Net (Ronneberger, Fischer, and Brox 2015), DeepLabV3+ (Chen et al 2018;Ryu et al 2023), U-CSRNet (Huang et al 2020), MCSpatNet (Abousamra et al 2021), FIDT and OT-M (Lin and Chan 2023), as well as end-to-end methods including P2PNet (Song et al 2021), CLTR ) and E2E (Shui et al 2022) on the CoNSeP, BCData and PD-L1 datasets. Subsequently, we demonstrate the superiority of mFoV DPA-P2PNet over the precursor studies (Bai, Xu, and Xing 2020;Bai et al 2022) on the OCELOT dataset.…”

Section: Resultsmentioning

confidence: 99%

“…Mainstream PCD methods (Abousamra et al 2021;Cai et al 2021;Zhang et al 2022;Ryu et al 2023) operate similarly with density map-based crowd localization approaches but regress multiple density maps, each corresponding to a distinct cell type. Recently, (Shui et al 2022) introduces the advanced P2PNet to perform PCD in an end-to-end manner. However, the original P2PNet model can only decode from a single-level feature map, which is insufficient to squeeze the most out of the multi-scale information.…”

Section: Point-based Cell Detectionmentioning

confidence: 99%

“…However, the original P2PNet model can only decode from a single-level feature map, which is insufficient to squeeze the most out of the multi-scale information. To alleviate this deficiency, (Shui et al 2022) downsamples the shallow feature maps and executes feature fusion by element-wise summation, resulting in an enhanced feature map to decode. Nevertheless, we argue that this approach would lead to a loss of fine-grained information to some extent and the improved P2PNet still suffers from the biased distribution of pre-defined point proposals.…”

Section: Point-based Cell Detectionmentioning

confidence: 99%

“…However, the heuristic post-processing not only demands tedious hyper-parameter tuning but also results in low throughput. To address these issues, a recent study (Shui et al 2022) introduced Point-to-Point Network (P2PNet) (Song et al 2021) to establish an end-to-end PCD system. Specifically, P2PNet adopts a detection paradigm, where the cell coordinates and categories can be directly obtained by refining and classifying pre-defined point proposals on an input image.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

DPA-P2PNet: Deformable Proposal-Aware P2PNet for Accurate Point-Based Cell Detection

Shui,

Zheng,

Zhu

et al. 2024

AAAI

View full text Add to dashboard Cite

Point-based cell detection (PCD), which pursues high-performance cell sensing under low-cost data annotation, has garnered increased attention in computational pathology community. Unlike mainstream PCD methods that rely on intermediate density map representations, the Point-to-Point network (P2PNet) has recently emerged as an end-to-end solution for PCD, demonstrating impressive cell detection accuracy and efficiency. Nevertheless, P2PNet is limited to decoding from a single-level feature map due to the scale-agnostic property of point proposals, which is insufficient to leverage multi-scale information. Moreover, the spatial distribution of pre-set point proposals is biased from that of cells, leading to inaccurate cell localization. To lift these limitations, we present DPA-P2PNet in this work. The proposed method directly extracts multi-scale features for decoding according to the coordinates of point proposals on hierarchical feature maps. On this basis, we further devise deformable point proposals to mitigate the positional bias between proposals and potential cells to promote cell localization. Inspired by practical pathological diagnosis that usually combines high-level tissue structure and low-level cell morphology for accurate cell classification, we propose a multi-field-of-view (mFoV) variant of DPA-P2PNet to accommodate additional large FoV images with tissue information as model input. Finally, we execute the first self-supervised pre-training on immunohistochemistry histopathology image data and evaluate the suitability of four representative self-supervised methods on the PCD task. Experimental results on three benchmarks and a large-scale and real-world interval dataset demonstrate the superiority of our proposed models over the state-of-the-art counterparts. Codes and pre-trained weights are available at https://github.com/windygoo/DPA-P2PNet.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Point-based Cell Detectionmentioning

confidence: 99%

Section: Point-based Cell Detectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DPA-P2PNet: Deformable Proposal-Aware P2PNet for Accurate Point-Based Cell Detection

Shui,

Zheng,

Zhu

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…The algorithms may also cause inconsistent length due to an inaccurately extracted medial axis or incorrect bending points. Recently, advanced deep learning techniques have shown promising results in medical image analysis [12], [13], [14], [15], [16] because of their strong ability in learning image representations. A new line of work is proposed, aiming to straighten chromosomes by establishing mapping between bent and straight chromosome using deep learning [17].…”

mentioning

confidence: 99%

Masked Conditional Variational Autoencoders for Chromosome Straightening

Zheng

Shui

et al. 2024

IEEE Trans. Med. Imaging

View full text Add to dashboard Cite

Karyotyping is of importance for detecting chromosomal aberrations in human disease. However, chromosomes easily appear curved in microscopic images, which prevents cytogeneticists from analyzing chromosome types. To address this issue, we propose a framework for chromosome straightening, which comprises a preliminary processing algorithm and a generative model called masked conditional variational autoencoders (MC-VAE). The processing method utilizes patch rearrangement to address the difficulty in erasing low degrees of curvature, providing reasonable preliminary results for the MC-VAE. The MC-VAE further straightens the results by leveraging chromosome patches conditioned on their curvatures to learn the mapping between banding patterns and conditions. During model training, we apply a masking strategy with a high masking ratio to train the MC-VAE with eliminated redundancy. This yields a non-trivial reconstruction task, allowing the model to effectively preserve chromosome banding patterns and structure details in the reconstructed results. Extensive experiments on three public datasets with two stain styles show that our framework surpasses the performance of state-of-the-art methods in retaining banding patterns and structure details. Compared to using real-world bent chromosomes, the use of high-quality straightened chromosomes generated by our proposed method can improve the performance of various deep learning models for chromosome classification by a large margin. Such a straightening approach has the potential to be combined with other karyotyping systems to assist cytogeneticists in chromosome analysis.

show abstract