滤波反投影重建中不同卷积核对老年足踝负重锥形束 CT 图像质量的影响*

魏震岳 ,  张立新 ,  王执宇 ,  李向明 ,  娄路馨 ,  杜辉

国际老年医学杂志 ›› 2026, Vol. 47 ›› Issue (6) : 347 -348.

PDF (1679KB)
国际老年医学杂志 ›› 2026, Vol. 47 ›› Issue (6) : 347 -348. DOI: 10.3969/j.issn.1674-7593.2026.06.007
论著

滤波反投影重建中不同卷积核对老年足踝负重锥形束 CT 图像质量的影响*

作者信息 +

Effects of different convolution kernels in filtered back projection reconstruction on image quality of weight-bearing cone-beam CT of the foot and ankle in elderly patients

Author information +
文章历史 +
PDF (1718K)

摘要

目的 比较滤波反投影 (FBP) 重建算法中三种不同卷积核对老年足踝负重锥形束 CT (CBCT) 图像质量的影响。方法 选取 2025 年 1 月-2025 年 12 月首都医科大学附属北京积水潭医院进行足踝负重 CBCT 检查的 100 例老年患者, 分别采用平滑核 (设为 A 组)、 锐利核 (设为 B 组)、 中等核 (设为 C 组) 三种卷积核进行 FBP 重建。 由两名高年资、 具备丰富经验的影像科医师在双盲法下独立阅片, 进行主观图像质量评价, 并客观测量三组图像的客观指标 [ 信噪比 (SNR)、 对比噪声比 (CNR)、 伪影指数 (AI)、 平均梯度 (AG) ]; 对比三组主观、 客观指标的差异, 并评价观察者一致性。结果 两名观察者间的主观评分一致性分析显示, 观察者间的组内相关系数 (ICC) 为 0. 88 (95 % CI: 0. 83 ~ 0. 92), 提示具有良好的一致性; 观察者内的 ICC 为 0. 89 (95 %CI: 0. 84 ~ 0. 93), 提示具有良好的重复性。B 组整体图像质量、 骨小梁显示、 骨赘与关节面显示均高于 A 组、C 组, 且 C 组高于 A 组 (P < 0. 05); B 组综合评分高于 A 组、C 组 (P<0. 05); A 组噪声干扰评分、SNR、CNR 值均高于 B 组、C 组, 且 C 组高于 B 组 (P<0. 05); A 组 AI、AG 值低于 B 组、C 组, 且 C 组低于 B 组 (P<0. 05)。结论 FBP 重建中不同卷积核对老年足踝负重 CBCT 图像的影响存在差异, 各有优缺点, 锐利核对显示骨结构、 关节面等细微结构具有显著优势, 可获得满意的主观图像质量, 但噪声干扰较重; 平滑核与其相反, 对细微结构显示不佳, 但可控制噪声干扰; 中等核可平衡分辨率及噪声, 能够作为折中选择。 临床诊疗需根据患者病变类型、 需求等选择相应卷积核。

Abstract

Objective To compare the effects of three different convolution kernels in the filtered back projection (FBP) reconstruction algorithm on image quality of weight-bearing cone-beam CT (CBCT) of the foot and ankle in elderly patients. Methods One hundred elderly patients who underwent weight-bearing CBCT examination of the foot and ankle at Beijing Jishuitan Hospital, Capital Medical University, from January 2025 to December 2025 were enrolled. FBP reconstruction was performed using three convolution kernels: a smooth kernel (group A), a sharp kernel (group B), and a medium kernel (group C). Two senior radiologists with extensive experience independently reviewed the images under double-blind conditions for subjective image quality assessment, and Objective metrics were measured for each group, including signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), artifact index (AI), and average gradient (AG). Differences in subjective and Objective metrics among the three groups were compared, and interobserver agreement was evaluated. Results Consistency analysis of subjective scores between the two observers showed an interobserver intraclass correlation coefficient (ICC) of 0. 88 (95% CI: 0. 83-0. 92), indicating good interobserver agreement, and an intraobserver ICC of 0. 89 (95% CI: 0. 84-0. 93), indicating good intraobserver reproducibility. Overall image quality, trabecular bone depiction, and visualization of osteophytes and articular surfaces in group B were significantly superior to those in groups A and C, and those in group C were superior to those in group A (P<0. 05); the comprehensive score of group B was higher than that of groups A and C (P<0. 05). The noise interference score, SNR, and CNR in group A were higher than those in groups B and C, and those in group C were higher than those in group B (P<0. 05); the AI and AG values in group A were lower than those in groups B and C, and those in group C were lower than those in group B (P<0. 05). Conclusion Different convolution kernels in FBP reconstruction have distinct effects on weight-bearing CBCT images of the foot and ankle in elderly patients, each with its own advantages and disadvantages. The sharp kernel offers significant advantages in depicting bony structures and articular surfaces and other fine structures, yielding satisfactory subjective image quality, albeit with higher noise interference; the smooth kernel shows the opposite characteristics, with poor depiction of fine structures but effective noise suppression; the medium kernel strikes a balance between resolution and noise and may serve as a compromise option. The appropriate convolution kernel should be selected in clinical practice according to the patient’s lesion type and diagnostic requirements.

关键词

滤波反投影 / 卷积核 / 足踝 / 负重锥形束 CT / 图像质量

Key words

Filtered back projection / Convolution kernel / Foot and ankle / Weight-bearing cone-beam CT / Image quality

引用本文

引用格式 ▾
魏震岳,张立新,王执宇,李向明,娄路馨,杜辉. 滤波反投影重建中不同卷积核对老年足踝负重锥形束 CT 图像质量的影响*[J]. 国际老年医学杂志, 2026, 47(6): 347-348 DOI:10.3969/j.issn.1674-7593.2026.06.007

登录浏览全文

4963

注册一个新账户 忘记密码

参考文献

[1]

Smith J T, Parker E B, Bluman E M, et al. Differences in baseline physical function and mental health PROM scores in patients with foot and ankle conditions[J]. Foot Ankle Int, 2024, 45(6): 621-631.

[2]

冯亮, 张雅飞, 霍洪峰. 足踝复合体本体感觉的测量与评价[J]. 中国组织工程研究, 2024, 28 (20): 3259-3264.

[3]

Feng L, Zhang Y F, Huo H F. Measurement and evaluation of proprioception of foot and ankle complexes [J]. Chin J Tissue Eng Res, 2024, 28(20): 3259-3264.

[4]

Raheman F J, Rojoa D M, Hallet C, et al. Can weight-bearing cone-beam CT reliably differentiate between stable and unstable syndesmotic ankle injuries? A systematic review and meta-analysis [J]. Clin Orthop Relat Res, 2022, 480(8): 1547-1562.

[5]

de Carvalho K A M, Barbachan Mansur N S, de Cesar Netto C. Cone-beam weight-bearing computed tomography of ankle arthritis and total ankle arthroplasty [J]. Foot Ankle Clin, 2023, 28(3): 509-528.

[6]

Wiedeman C, Lorraine P, Wang G, et al. Simulated deep CT characterization of liver metastases with high-resolution filtered back projection reconstruction [J]. Visual Comput Ind Biomed Art, 2024, 7(1): 13.

[7]

Zeng R, LinC Y, Li Q, et al. Performance of a deep learning-based CT image denoising method: generalizability over dose, reconstruction kernel, and slice thickness[J]. Med Phys, 2022, 49(2): 836-853.

[8]

朱江, 辛宇强, 刘英杰, 等. 老年足踝部隐匿性骨折 MSCT 三维重建、磁共振成像特点及联合诊断价值[J]. 临床误诊误治, 2024, 37(8): 69-75.

[9]

Zhu J, Xin Y Q, Liu Y J, et al. MSCT three-dimensional reconstruction, magnetic resonance imaging characteristics and combined diagnostic value of occult fracture in the foot and ankle in the elderly[J]. Clin Misdiagnosis Mistherapy, 2024, 37(8): 69-75.

[10]

Brinch S, Wellenberg R H H, Boesen M P, et al. Weight-bearing cone-beam CT: the need for standardised acquisition protocols and measurements to fulfill high expectations-a review of the literature[J]. Skeletal Radiol, 2023, 52(6): 1073-1088.

[11]

Efrima B, Barbero A, Ovadia J E, et al. Reliability of cone beam weightbearing computed tomography analysis of total ankle arthroplasty positioning and comparison to weightbearing X-ray measurements[J]. Foot Ankle Int, 2023, 44(7): 637-644.

[12]

Lv L, Zeng G L, Zan Y, et al. A back-projection-and-filtering-like (BPF-like) reconstruction method with the deep learning filtration from listmode data in TOF-PET [J]. Med Phys, 2022, 49(4): 2531-2544.

[13]

罗开, 陈亮, 梁巍, 等. 基于稀疏表示的滤波反投影超声层析成像重建算法[J]. 计算机应用, 2023, 43 (3): 903-908.

[14]

Luo K, Chen L, Liang W, et al. Sparse representation-based reconstruction algorithm for filtered back-projection ultrasound tomography [J]. J Comput Appl, 2023, 43 (3): 903-908.

[15]

何泉宇, 王彦玲, 江南, 等. 不同层厚 Sa36 卷积核图像对人工智能冠状动脉钙化定量测量的影响[J]. 影像诊断与介入放射学, 2025, 34(6): 395-403.

[16]

He Q Y, Wang Y L, Jiang N, et al. Effect of slice thickness using artificial intelligence-based Sa36 convolution kernels on quantitative measurement of coronary artery calcification[J]. Diagn Imaging Interv Radiol, 2025, 34 (6): 395-403.

[17]

Krompaß K, GoldbrunnerF A, Hartung V, et al. Combined influence of quantum iterative reconstruction level and kernel sharpness on image quality in photon counting CT angiography of the upper leg[J]. Sci Rep, 2024, 14 (1): 27774.

[18]

Hayashi A, Fukui R, Kamioka S, et al. Task-based assessment of resolution properties of CT images with a new index using deep convolutional neural network[J]. Radiol Phys Technol, 2024, 17(1): 83-92.

[19]

余菡, 曹闻挺, 韦鑫, 等. 重组卷积核联合高级迭代重组在优化 CT 冠状动脉支架成像中的价值[J]. 临床放射学杂志, 2022, 41(5): 892-896.

[20]

Yu H, Cao W T, Wei X, et al. Reconstruction convolution kernel combined with advanced iterative reconstruction in optimizing CT coronary stent imaging[J]. J Clin Radiol, 2022, 41(5): 892-896.

[21]

Liu H, Zhuang Y, Song E, et al. A modality-collaborative convolution and transformer hybrid network for unpaired multi-modal medical image segmentation with limited annotations [ J ]. Med Phys, 2023, 50 (9 ): 5460-5478.

[22]

Huber N R, Ferrero A, Rajendran K, et al. Dedicated convolutional neural network for noise reduction in ultra-high-resolution photon-counting detector computed tomography[J]. Phys Med Biol, 2022, 67(17): 10.

[23]

姜筱璇, 余行, 邓中华, 等. 不同多模型迭代重建算法与卷积核参数对 CT 影像自动分割稳定性的影响 [J]. 中国医疗器械杂志, 2022, 46(2): 219-224.

[24]

Jiang X X, Yu X, Deng Z H, et al. The effects of different adaptive statistical iterative reconstruction-V and convolution kernel parameters on auto-segmentation stability in CT images[J]. Chin J Med Lnstrumentation, 2022, 46(2): 219-224.

[25]

姚广, 李军, 任俊立, 等. 超高分辨率光子计数探测器 CT 对提高神经血管成像图像质量的价值[J]. 中华放射学杂志, 2025, 59(12): 1353-1360.

[26]

Yao G, Li J, Ren J L, et al. Value of ultra-high-resolution photon-counting detector CT in improving neurovascular image quality[J]. Chin J Radiol, 2025, 59(12): 1353-1360.

[27]

王萍, 黄佳珊, 王泽, 等. 双能量 CT 不同卷积函数对足踝关节尿酸单钠晶体沉积检测的影响[J]. 中国医疗设备, 2022, 37(10): 74-78.

[28]

Wang P, Huang J S, Wang Z, et al. Effects of different convolution functions of dual-energy CT on the detection of monosodium urate crystal deposition in ankle joint[J]. China Med Devices, 2022, 37(10): 74-78.

基金资助

*2024 年度贵州省卫生健康高质量发展医学科研联合基金(2024GZYXKYJJXM0118)

AI Summary AI Mindmap
PDF (1679KB)

0

访问

0

被引

详细

导航
相关文章

AI思维导图

/

〈 〉