基于先验知识优化的医学图像候选区域生成方法

doi:10.13229/j.cnki.jdxbgxb.20230508

摘要/Abstract

摘要：

针对钙化斑块区域小、与非斑块区域易混淆的特点，本文提出一种医学先验知识引导的候选区域生成优化方法。该方法基于目标检测网络Faster R-CNN，通过锚框的位置与形状两方面优化候选区域生成。利用钙化斑块定义生成指导掩码图像，筛选候选区域生成位置。采用锚框形状预测分支生成候选区域形状。针对特征金字塔网络中不同尺度的特征图，提出多尺度的指导掩码金字塔。在CCTA图像中检测钙化斑块的实验结果表明：与集成标准RPN的Faster R-CNN模型相比，本文方法的AP与Recall分别提高12.8%与25.7%；在平均每张图像假阳性数量为2的情况下，本文方法Recall值达到86.05%。

关键词: 计算机应用, 钙化斑块检测, 先验知识, 深度学习, 医学图像

Abstract:

Considering the small size of calcified plaques and the difficulty in distinguishing them from non-plaque areas， a medical prior knowledge-guided optimization method for proposal generation was proposed. This method is based on the object detection network Faster R-CNN. Proposal generation is optimized through the location and shape of anchor. Calcified plaque definition is used to generate a guidance mask image to direct the location generation of proposal. Anchor shape prediction branch is employed to optimize the shape generation of proposal. A multi-scale guidance mask pyramid architecture is proposed for feature maps of different scales in FPN. The experimental results of detecting calcified plaques in CCTA images show that the proposed method improved AP and Recall by 12.8% and 25.7% respectively compared with Faster R-CNN with standard RPN； When the average number of false positives per image is 2， Recall of the proposed method is 86.05%.

Key words: computer application, calcified plaque detection, prior knowledge, deep learning, medical image

中图分类号:

TP391

赵孟雪,车翔玖,徐欢,刘全乐. 基于先验知识优化的医学图像候选区域生成方法[J]. 吉林大学学报(工学版), 2025, 55(2): 722-730.

Meng-xue ZHAO,Xiang-jiu CHE,Huan XU,Quan-le LIU. A method for generating proposals of medical image based on prior knowledge optimization[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(2): 722-730.

图/表 10

图1

图2

图3

指导掩码金字塔获取过程伪代码"

Algorithm 1 Mask pyramid acquisition process

1： function get_mask_pyramid（img $I$ ， int $n$ ）：

2： # input： $I$ is the CT image file $(S * S)$

3： # input： $n$ is the serial number of the scale $n$ >2

4： # output： $M_p r ?$ is a mask image set $n$

5： # getting mask image $M$

6： $M$ = torch.zeros（ $[S, S]$ ）

7： while CT value at $I (i, j)$ is greater than or equal to 130 do

8： $M (i, j)$ = 1

9： end while

10： # getting mask pyramid $M_p r$

11： conv1 = MaxPool（kernel_size=4， stride=4， padding=0）

12： conv2 = MaxPool（kernel_size=2， stride=2， padding=0）

13： $M 1$ = conv1（ $M$ ）

14： $M_p r$ = $[M 1]$

15： for $t$ = 2 to $n$ do

16： $M t$ = conv2（ $M (t - 1)$ ）

17： $M_p r$ = $M_p r + [M t]$

18： end for

19： return $M_p r$

图3

图4

图5

表1

图6

图7

图8

表2

参考文献 40

1	Naghavi M, Wang H, Lozano R, et al. Global, regional, and national age-sex specific all-cause and cause-specific mortality for 240 causes of death, 1990-2013: a systematic analysis for the global burden of disease study 2013[J]. The Lancet, 2015, 385(9963): 117-171.
2	Hutcheson J D, Goettsch C, Bertazzo S, et al. Genesis and growth of extracellular-vesicle-derived microcalcification in atherosclerotic plaques[J]. Nature Materials, 2016, 15(3): 335-343.
3	Išgum I, Van G B, Viergever M A. Automatic detection of calcifications in the aorta from abdominal CT scans[J]. International Congress Series, 2003, 1256(C): 1037-1042.
4	Isgum I, Prokop M, Niemeijer M, et al. Automatic coronary calcium scoring in low-dose chest computed tomography[J]. IEEE Transactions on Medical Imaging, 2012, 31(12): 2322-2334.
5	Wolterink J M, Leiner T, Takx R A P, et al. Automatic coronary calcium scoring in non-contrast-enhanced ECG-triggered cardiac CT with ambiguity detection[J]. IEEE Transactions on Medical Imaging, 2015, 34(9): 1867-1878.
6	Wolterink J M, Leiner T, Devos B D, et al. Automatic coronary artery calcium scoring in cardiac CT angiography using paired convolutional neural networks[J]. Medical Image Analysis, 2016, 34: 123-136.
7	Tourassi G D, Armato S G, Lessmann N, et al. Deep convolutional neural networks for automatic coronary calcium scoring in a screening study with low-dose chest CT[J].Medical Imaging 2016:Computer-Aided Diagnosis,2016, 9785: 255-260.
8	Liu J, Lu L, Yao J, et al. Pelvic artery calcification detection on CT scans using convolutional neural networks[J]. Conference on Computer-Aided Diagnosis,2017, 10134: 319-325.
9	Mohri M, Rostamizadeh A, Talwalkar A. Foundations of Machine Learning[M]. London: MIT Press, 2018.
10	Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks[J]. Advances in Neural Information Processing Systems, 2012, 25: 1097-1105.
11	Lin T Y, Maire M, Belongie S, et al. Microsoft coco: common objects in context[J]. Computer Vision, 2014, 8693: 740-755.
12	Acharya U R, Meiburger K M, Koh J E W, et al. Automated detection of calcified plaque using higher-order spectra cumulant technique in computer tomography angiography images[J]. International Journal of Imaging Systems and Technology, 2020, 30: 285-297.
13	Li Y C, Shen T Y, Chen C C, et al. Automatic detection of atherosclerotic plaque and calcification from intravascular ultrasound images by using deep convolutional neural networks[J]. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 2021, 68(5): 1762-1772.
14	Chen C, Biffi C, Tarroni G, et al. Learning shape priors for robust cardiac mr segmentation from multi-view images[J]. Lecture Notes in Computer Science, 2019, 11765: 523-531.
15	Yue Q, Luo X, Ye Q, et al. Cardiac segmentation from LGE MRI using deep neural network incorporating shape and spatial priors[J]. Lecture Notes in Computer Science, 2019, 11765: 559-567.
16	Mirikharaji Z, Hamarneh G. Star shape prior in fully convolutional networks for skin lesion segmentation[C]∥Medical Image Computing and Computer Assisted Intervention, Granada, Spain, 201: 737-745.
17	Liao W, Rosenhahn B, Shuai L, et al. Natural language guided visual relationship detection[J]. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, 2019, 6: 444-453.
18	Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]∥Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Columbus, USA, 2014: 580-587.
19	Yang G, Chen Y, Ning X, et al. Automatic coronary calcium scoring using noncontrast and contrast CT images[J]. Medical Physics, 2016, 43(5): 2174-2186.
20	Durlak F, Wels M, Schwemmer C, et al. Growing a random forest with fuzzy spatial features for fully automatic artery-specific coronary calcium scoring[J]. Lecture Notes in Computer Science, 2017, 10541: 27-35.
21	Zhang Y, Wang S, Zhao H, et al. CT image classification based on convolutional neural network[J]. Neural Computing and Applications, 2020, 33(14): 8191-8200.
22	Gao K, Su J, Jiang Z, et al. Dual-branch combination network (DCN): towards accurate diagnosis and lesion segmentation of COVID-19 using CT images[J]. Medical Image Analysis, 2021, 67: 101836.
23	Liang X, Li N, Zhang Z, et al. Incorporating the hybrid deformable model for improving the performance of abdominal CT segmentation via multi-scale feature fusion network[J]. Medical Image Analysis, 2021, 73: 102156.
24	Fan J, Cao X, Yap P T, et al. BIRNet: brain image registration using dual-supervised fully convolutional networks[J]. Medical Image Analysis, 2019, 54: 193-206.
25	Prasoon A, Petersen K, Igel C, et al. Deep feature learning for knee cartilage segmentation using a triplanar convolutional neural network[J]. Lecture Notes in Computer Science, 2013, 8150: 246-253.
26	Chellamuthu K, Liu J, Yao J, et al. Atherosclerotic vascular calcification detection and segmentation on low dose computed tomography scans using convolutional neural networks[C]∥IEEE 14th International Symposium on Biomedical Imaging, Melbourne, Australia, 2017: 388-391.
27	Gernaat S A M, Van S G M, Koh V, et al. Automatic quantification of calcifications in the coronary arteries and thoracic aorta on radiotherapy planning CT scans of Western and Asian breast cancer patients[J]. Radiotherapy and Oncology, 2018, 127(3): 487-492.
28	Lessmann N, Van Ginneken B, Zreik M, et al. Automatic calcium scoring in low-dose chest CT using deep neural networks with dilated convolutions[J]. IEEE Transactions on Medical Imaging, 2018, 37(2): 615-625.
29	Zreik M, Van H R W, Wolterink J M, et al. A recurrent CNN for automatic detection and classification of coronary artery plaque and stenosis in coronary CT angiography[J]. IEEE Transactions on Medical Imaging, 2019, 38(7): 1588-1598.
30	Fischer A M, Eid M, Dececco C N, et al. Accuracy of an artificial intelligence deep learning algorithm implementing a recurrent neural network with long short-term memory for the automated detection of calcified plaques from coronary computed tomography angiography[J]. Journal of Thoracic Imaging, 2020, 35: 49-57.
31	Ren S, He K, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. Advances in Neural Information Processing Systems, 2015, 28: 91-99.
32	Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]∥IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 936-944.
33	Agatston A S, Janowitz W R, Hildner F J, et al. Quantification of coronary artery calcium using ultrafast computed tomography[J]. Journal of the American College of Cardiology, 1990, 15(4): 827-832.
34	Wang J, Chen K, Yang S, et al. Region proposal by guided anchoring[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 2960-2969.
35	Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]∥IEEE International Conference on Computer Vision, Venice, Italy, 2017: 2999-3007.
36	Wolterink J M, Leiner T, Devos B D, et al. An evaluation of automatic coronary artery calcium scoring methods with cardiac CT using the orCaScore framework[J]. Medical Physics, 2016, 43(5): 2361-2373.
37	Zhao M, Che X, Liu H, et al. Medical prior knowledge guided automatic detection of coronary arteries calcified plaque with cardiac CT[J]. Electronics (Switzerland), 2020, 9(12): 1-14.
38	Chen K, Wang J, Pang J, et al. MMdetection: open mmlab detection toolbox and benchmark[J]. Arxiv Preprint, 2019, 6: 190607155.
39	Sun P, Zhang R, Jiang Y, et al. Sparse R-CNN: end-to-end object detection with learnable proposals[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 14454-14463.
40	Ge Z, Liu S, Wang F, et al. Yolox: exceeding yolo series in 2021[J]. Arxiv Preprint, 2017, 7: 210708430.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

方法	Recall	AP	参数量/MB	FLOPs/GB	推理时间/ms
Faster R-CNN^［31］	0.744	0.554	41.14	63.67	34.56
文献［26］	0.756	0.567	82.28	127.34	—
GA-Faster-RCNN^［34］	0.814	0.724	41.71	63.19	51.49
Sparse R-CNN^［39］	0.680	0.500	105.94	44.16	55.44
YOLOX^［40］	0.797	0.655	8.94	8.52	28.32
本文	0.872	0.811	41.71	63.18	65.63

方法	位置优化	形状优化	Recall	AP
Faster R-CNN^［31］			0.744	0.554
模型 A	√		0.773	0.663
模型 B		√	0.837	0.792
本文	√	√	0.872	0.811

[1]	车翔玖,武宇宁,刘全乐. 基于因果特征学习的有权同构图分类算法[J]. 吉林大学学报(工学版), 2025, 55(2): 681-686.
[2]	蔡晓东,周青松,张言言,雪韵. 基于动静态和关系特征全局捕获的社交推荐模型[J]. 吉林大学学报(工学版), 2025, 55(2): 700-708.
[3]	徐慧智,蒋时森,王秀青,陈爽. 基于深度学习的车载图像车辆目标检测和测距[J]. 吉林大学学报(工学版), 2025, 55(1): 185-197.
[4]	郭晓然,王铁君,闫悦. 基于局部注意力和本地远程监督的实体关系抽取方法[J]. 吉林大学学报(工学版), 2025, 55(1): 307-315.
[5]	汪豪,赵彬,刘国华. 基于时间和运动增强的视频动作识别[J]. 吉林大学学报(工学版), 2025, 55(1): 339-346.
[6]	刘元宁,臧子楠,张浩,刘震. 基于深度学习的核糖核酸二级结构预测方法[J]. 吉林大学学报(工学版), 2025, 55(1): 297-306.
[7]	张磊,焦晶,李勃昕,周延杰. 融合机器学习和深度学习的大容量半结构化数据抽取算法[J]. 吉林大学学报(工学版), 2024, 54(9): 2631-2637.
[8]	李路,宋均琦,朱明,谭鹤群,周玉凡,孙超奇,周铖钰. 基于RGHS图像增强和改进YOLOv5网络的黄颡鱼目标提取[J]. 吉林大学学报(工学版), 2024, 54(9): 2638-2645.
[9]	赵宏伟,武鸿,马克,李海. 基于知识蒸馏的图像分类框架[J]. 吉林大学学报(工学版), 2024, 54(8): 2307-2312.
[10]	张云佐,郑宇鑫,武存宇,张天. 基于双特征提取网络的复杂环境车道线精准检测[J]. 吉林大学学报(工学版), 2024, 54(7): 1894-1902.
[11]	乔百友,武彤,杨璐,蒋有文. 一种基于BiGRU和胶囊网络的文本情感分析方法[J]. 吉林大学学报(工学版), 2024, 54(7): 2026-2037.
[12]	郭昕刚,何颖晨,程超. 抗噪声的分步式图像超分辨率重构算法[J]. 吉林大学学报(工学版), 2024, 54(7): 2063-2071.
[13]	张丽平,刘斌毓,李松,郝忠孝. 基于稀疏多头自注意力的轨迹kNN查询方法[J]. 吉林大学学报(工学版), 2024, 54(6): 1756-1766.
[14]	孙铭会,薛浩,金玉波,曲卫东,秦贵和. 联合时空注意力的视频显著性预测[J]. 吉林大学学报(工学版), 2024, 54(6): 1767-1776.
[15]	李延风,刘名扬,胡嘉明,孙华栋,孟婕妤,王奥颖,张涵玥,杨华民,韩开旭. 基于梯度转移和自编码器的红外与可见光图像融合[J]. 吉林大学学报(工学版), 2024, 54(6): 1777-1787.