CentroidNet： a light-weight， fast nuclei centroid detection model for breast Ki67 scoring

Wen Ke; Jin Xu; An Hong; He Jie; Wang Jue

doi:10.11834/jig.211207

Medical Image Processing | Views : 0 下载量: 0 CSCD: 1

PDF
Export
Share
Collection
Album

CentroidNet： a light-weight， fast nuclei centroid detection model for breast Ki67 scoring
Vol. 28, Issue 4, Pages: 1119-1133(2023)
Published： 16 April 2023 ，
DOI： 10.11834/jig.211207
稿件说明：

移动端阅览

文可，金旭，安虹，何杰，王珏. 2023. CentroidNet：轻量快速的乳腺癌Ki67细胞核中心点检测模型. 中国图象图形学报， 28(04):1119-1133

Wen Ke， Jin Xu， An Hong， He Jie， Wang Jue. 2023. CentroidNet： a light-weight， fast nuclei centroid detection model for breast Ki67 scoring. Journal of Image and Graphics， 28(04):1119-1133
文可，金旭，安虹，何杰，王珏. 2023. CentroidNet：轻量快速的乳腺癌Ki67细胞核中心点检测模型. 中国图象图形学报， 28(04):1119-1133 DOI： 10.11834/jig.211207.

Wen Ke， Jin Xu， An Hong， He Jie， Wang Jue. 2023. CentroidNet： a light-weight， fast nuclei centroid detection model for breast Ki67 scoring. Journal of Image and Graphics， 28(04):1119-1133 DOI： 10.11834/jig.211207.

摘要

目的

Ki67分数是乳腺癌预后评估的重要指标，计算该分数的关键步骤是检测阴性与阳性癌细胞核。人工检测面临疲劳与主观差异的问题。卷积神经网络有望实现高质量、自动化的细胞核检测，然而需要病理专家为其标注细胞核。为了减轻标注的工作量，不少研究者提出以中心点标注训练卷积神经网络。然而这些方法采用过于复杂的卷积神经网络和后处理流程，未能充分提高易用性和效率、发挥卷积神经网络的质量。对此，提出CentroidNet模型，旨在提高中心点检测的质量、效率和易用性。

方法

CentroidNet模型在图像上放置均匀排布的锚点，为每个锚点预测一个候选点，一部分候选点通过基于阈值的筛选策略成为预测点。本文提出最近锚点匹配策略用于生成训练标签，既保证了端到端推理，又规避了其他一对一标签匹配算法所具有的标签抖动问题。本文建议锚点间距应尽可能接近训练集答案点间最短距离的第一百分位数，并指出这样的锚点间距能够在前景标签数、坐标回归难度与效率之间取得良好的平衡。本文在设计卷积神经网络的结构时，没有采纳广为使用的U-Net或特征金字塔（feature pyramid network， FPN）中的多级上采样与旁路连接，反而提高了质量和效率。

结果

本文在BCData数据集上评估CentroidNet模型的质量与效率。BCData是目前规模最大的、公开的乳腺癌Ki67癌细胞核中心点检测数据集。在质量方面，CentroidNet取得的综合F1分数为0.879 1，媲美当前的最高质量。在效率方面，CentroidNet的推理速度为12.96 ms/幅、显存占用为138.8 MB/幅，达到了当前最高的效率，远低于若干主流或最新的模型。

结论

CentroidNet具有高质量、高效率和高易用性；与现有同类模型相比，进一步提高了乳腺癌Ki67细胞核中心点检测的可行性。

Abstract

Objective

Breast cancer-prognostic Ki67 score can be as a key indicator for the proliferation rate of malignant （invasive） cells. Negative and positive nuclei detection is an essential part of Ki67 scoring. An automated algorithm for nuclei detection can alleviate the negative impact of intra/inter-observer variation and labor-intensive nuclei counting. In recent years， deep learning methods have been developing intensively in relevant to recognition tasks on pathology images in terms of their learning potentials. Deep learning-based model can learn features derived from the raw data and relieve labor-intensive annotation of training images for pathologists. To alleviate the labor cost of annotation， our research is focused on modeling Ki67 nuclei detection as the centroid detection problem. The existing centroid detection models are commonly used to convert the centroid annotation into a probability map with the same size as the input image， the centroid detection problem is thus converted to a semantic segmentation-like problem. However， a semantic segmentation model has restricted by huge computation cost due to its huge amount of convolutional layers-related decoder. Nevertheless， due to the non-deep-learning post processing on the output heatmap， the whole detection process is complicated and inefficient， and the quality of a deep learning model is required to be developed further. Our CentroidNet model is facilitated to optimize nucleus centroid detection.

Method

The CentroidNet consists of a fully convolutional centroid detector initiated by ResNeXt， along with the detector’s training and inference methods. The CentroidNet can place evenly spaced anchor points on the input image. Actually， the anchor points are the places where the center of the detector’s receptive field is sited on. Each anchor point-interconnected detector can predict the classification probabilities and the offset to the anchor point， which is called a candidate point as a whole. In this way， the anchor points are the niches of the candidate points. A candidate point has the highest probability and is higher than 0.9. It is treated as predicted point. To assign labels for anchor/candidate points， we implement the “nearest anchor” strategy. This strategy can assign any annotated point to its nearest anchor point literally. Those anchor points are not assigned via any background label-assigned annotated point， whereas those classified label is “background” and label-coordinated is the anchor’s coordinate. This strategy is mutual-benefited between annotated points and anchor/candidate points. The label jittering problem can be avoided， which is often seen in current one-to-one assignment strategies. The anchors-between spacing is concerned about more. We recommend that the anchor spacing approximate the First Percentile of the minimal distance between each annotated point in the training set to its neighboring annotated points. Such spacing can balance the ratio of foreground labels， regression-coordinated， and optimal efficiency. The CentroidNet’s detector can avoid the commonly adopted paradigm of U-Net or feature pyramid network（FPN）， which involves shortcut connections and multiple upsampling layers. This lightweight detector can improve quality and efficiency.

Result

We evaluate the quality and efficiency of CentroidNet model on BCData， the largest publicly available dataset for centroid detection of Ki67 carcinoma nuclei in breast cancer. For quality analysis， the CentroidNet can achieve an averaged F1 score of 0.879 1 compared to the SOTA（state of the art） score. For efficiency evaluation， the CentroidNet can achieve SOTA with an inference speed of 12.96 ms/image and a GPU memory footprint of 138.8 MB/image.

Conclusion

The CentroidNet is featured of light-weight， efficient and ease of use. It has the potentials for centroid detection of Ki67 carcinoma nuclei in breast cancer.

关键词

乳腺癌Ki67分数中心点检测一对一标签匹配锚点

Keywords

breast cancerKi67 scorecentroid detectionone-to-one label assignmentanchor point

references

Allegretti S， Bolelli F and Grana C. 2020. Optimized block-based algorithms to label connected components on GPUs. IEEE Transactions on Parallel and Distributed Systems， 31（2）： 423-438 ［DOI： 10.1109/tpds.2019.2934683http://dx.doi.org/10.1109/tpds.2019.2934683］

Bian X W and Ping Y F. 2019. Pathology in China： challenges and opportunities. Journal of Third Military Medical University， 41（19）： 1815-1817

卞修武，平轶芳. 2019. 我国病理学科发展面临的挑战和机遇. 第三军医大学学报， 41（19）： 1815-1817 ［DOI： 10.16016/j.1000-5404.201909212http://dx.doi.org/10.16016/j.1000-5404.201909212］

Carion N， Massa F， Synnaeve G， Usunier N， Kirillov A and Zagoruyko S. 2020. End-to-end object detection with transformers//Proceedings of the 16th European Conference on Computer Vision. Glasgow， UK： Springer： 213-229 ［DOI： 10.1007/978-3-030-58452-8_13http://dx.doi.org/10.1007/978-3-030-58452-8_13］

Dowsett M， Nielsen T O， A’Hern R， Bartlett J， Coombes R C， Cuzick J， Ellis M， Henry N L， Hugh J C， Lively T， McShane L， Paik S， Penault-Llorca F， Prudkin L， Regan M， Salter J， Sotiriou C， Smith I E， Viale G， Zujewski J A and Hayes D F. 2011. Assessment of Ki67 in breast cancer： recommendations from the international Ki67 in breast cancer working group. Journal of the National Cancer Institute， 103（22）： 1656-1664 ［DOI： 10.1093/jnci/djr393http://dx.doi.org/10.1093/jnci/djr393］

Elmore J G， Longton G M， Carney P A， Geller B M， Onega T， Tosteson A N A， Nelson H D， Pepe M S， Allison K H， Schnitt S J， O’Malley F P and Weaver D L. 2015. Diagnostic concordance among pathologists interpreting breast biopsy specimens. The Journal of the American Medical Association （JAMA）， 313（11）： 1122-1132 ［DOI： 10.1001/jama.2015.1405http://dx.doi.org/10.1001/jama.2015.1405］

Girshick R. 2015. Fast R-CNN//Proceedings of 2015 IEEE International Conference on Computer Vision. Santiago， Chile： IEEE： 1440-1448 ［DOI： 10.1109/iccv.2015.169http://dx.doi.org/10.1109/iccv.2015.169］

He K M， Zhang X Y， Ren S Q and Sun J. 2016. Deep residual learning for image recognition//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas， USA： IEEE： 770-778 ［DOI： 10.1109/cvpr.2016.90http://dx.doi.org/10.1109/cvpr.2016.90］

Huang Z Y， Ding Y， Song G L， Wang L， Geng R Z， He H L， Du S， Liu X， Tian Y H， Liang Y S， Zhou S K and Chen J. 2020. BCData： a large-scale dataset and benchmark for cell detection and counting//Proceedings of the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention. Lima， Peru： Springer： 289-298 ［DOI： 10.1007/978-3-030-59722-1_28http://dx.doi.org/10.1007/978-3-030-59722-1_28］

Kingma D P and Ba J. 2017. Adam： A method for stochastic optimization ［EB/OL］. ［2022-01-04］. https://arxiv.org/pdf/1412.6980.pdfhttps://arxiv.org/pdf/1412.6980.pdf

Kumar N， Verma R， Sharma S， Bhargava S， Vahadane A and Sethi A. 2017. A dataset and a technique for generalized nuclear segmentation for computational pathology. IEEE Transactions on Medical Imaging， 36（7）： 1550-1560 ［DOI： 10.1109/TMI.2017.2677499http://dx.doi.org/10.1109/TMI.2017.2677499］

LeCun Y， Bengio Y and Hinton G. 2015. Deep learning. Nature， 521（7553）： 436-444 ［DOI： 10.1038/nature14539http://dx.doi.org/10.1038/nature14539］

Lin T Y， Goyal P， Girshick R， He K M and Doll􀅡r P. 2017. Focal loss for dense object detection//Proceedings of 2017 IEEE International Conference on Computer Vision. Venice， Italy： IEEE： 2999-3007 ［DOI： 10.1109/TPAMI.2018.2858826http://dx.doi.org/10.1109/TPAMI.2018.2858826］

Lin Y， Qu Z Y， Chen H， Gao Z K， Li Y X， Xia L L， Ma K， Zheng Y F and Cheng K T. 2022. Label propagation for annotation-efficient nuclei segmentation from pathology images ［EB/OL］. ［2022-02-16］. https://arxiv.org/pdf/2202.08195.pdfhttps://arxiv.org/pdf/2202.08195.pdf

Loshchilov I and Hutter F. 2019. Decoupled weght decay regularization//Proceedings of the 5th International Conference on Learning Representations . Toulon， France：［s.n.］： https://openreview.net/forum？id=Bkg6RiCqY7https://openreview.net/forum?id=Bkg6RiCqY7

Qu H， Wu P X， Huang Q Y， Yi J R， Yan Z N， Li K， Riedlinger G M， De S， Zhang S T and Metaxas D N. 2020. Weakly supervised deep nuclei segmentation using partial points annotation in histopathology images ［EB/OL］. ［2022-01-10］. https://arxiv.org/pdf/2007.05448.pdfhttps://arxiv.org/pdf/2007.05448.pdf

Ribera J， Güera D， Chen Y H and Delp E J. 2019. Locating objects without bounding boxes//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach， USA： IEEE： 6472-6482 ［DOI： 10.1109/cvpr.2019.00664http://dx.doi.org/10.1109/cvpr.2019.00664］

Srinidhi C L， Ciga O and Martel A L. 2021. Deep neural network models for computational histopathology： a survey. Medical Image Analysis， 67： #101813 ［DOI： 10.1016/j.media.2020.101813http://dx.doi.org/10.1016/j.media.2020.101813］

Stewart R， Andriluka M and Ng A Y. 2016. End-to-end people detection in crowded scenes//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas， USA： IEEE： 2325-2333 ［DOI： 10.1109/CVPR.2016.255http://dx.doi.org/10.1109/CVPR.2016.255］

Tian K， Zhang J， Shen H C， Yan K Z， Dong P， Yao J H， Che S， Luo P F and Han X. 2020. Weakly-supervised nucleus segmentation based on point annotations： a coarse-to-fine self-stimulated learning strategy//Proceedings of the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention. Lima， Peru： Springer： 299-308 ［DOI： 10.1007/978-3-030-59722-1_29http://dx.doi.org/10.1007/978-3-030-59722-1_29］

Tian Z， Shen C H， Chen H and He T. 2019. FCOS： fully convolutional one-stage object detection//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision. Seoul， Korea （South）： IEEE： 9626-9635 ［DOI： 10.1109/ICCV.2019.00972http://dx.doi.org/10.1109/ICCV.2019.00972］

Wang J F， Song L， Li Z M， Sun H B， Sun J and Zheng N N. 2021. End-to-end object detection with fully convolutional network//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville， USA： IEEE： 15844-15853 ［DOI： 10.1109/cvpr46437.2021.01559http://dx.doi.org/10.1109/cvpr46437.2021.01559］

Xie S N， Girshick R， Doll􀅡r P， Tu Z W and He K M. 2017. Aggregated residual transformations for deep neural networks//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu， USA： IEEE： 5987-5995 ［DOI： 10.1109/cvpr.2017.634http://dx.doi.org/10.1109/cvpr.2017.634］

Yerushalmi R， Woods R， Ravdin P M， Hayes M M and Gelmon K A. 2010. Ki67 in breast cancer： prognostic and predictive potential. The Lancet Oncology， 11（2）： 174-183 ［DOI： 10.1016/s1470-2045（09）70262-1http://dx.doi.org/10.1016/s1470-2045（09）70262-1］

Zhou X， Gu M and Cheng Z. 2021. Local integral regression network for cell nuclei detection. Entropy， 23（10）： #1336 ［DOI： 10.3390/e23101336http://dx.doi.org/10.3390/e23101336］

Alert me when the article has been cited

提交

TransAS-UNet： regional segmentation of breast cancer Swin Transformer and of UNet algorithm

GZMH： a dataset of breast cancer pathological images for mitosis nuclei detection and segmentation

Classification of whole slide images of breast histopathology based on spatial correlation characteristics

Real-time robust feature-point matching algorithm