Identification of pulsatile tinnitus and visualization of high pathogenic regions based on CT images
-
摘要:
搏动性耳鸣(PT)的病因诊断依赖于影像学检测，但病因众多，缺乏普适性强、机制明确的诊断标准。基于搏动性耳鸣患者和无耳鸣人群的计算机断层扫描(CT)影像横截面图，提出一种高精度的耳鸣识别神经网络模型，并自动标示高致病区域，辅助临床诊断。使用迁移学习Resnet-v1-50模型，取骨窗颞骨中部水平截面样本进行分类学习，并以梯度加权类激活映射(grad-CAM)方法对分类高权重区域自动标注；统计CT截面大图(全颅)、中图(双侧颞骨)、小图(右侧颞骨)3种数据集的耳鸣分类高权重区域涉及的解剖结构，逐步细化感兴趣区域，提高分类高权重区域标注分辨率。实验结果显示：包含双侧颞骨的中图数据集分类精度最好，测试集精度达到100%。搏动性耳鸣分类高权重区域集中于双侧或单侧颞骨部位，主要包括颞骨蜂房、鼓窦、乙状窦骨板、上鼓室等部位。搏动性耳鸣与颞骨及附近骨质结构有密切关系；搏动性耳鸣患者在双侧颞骨或耳鸣对侧颞骨均有较大概率存在区别于无耳鸣人群的结构异常；颞骨蜂房、鼓窦、乙状窦骨板、鼓室等结构均有较高概率包含搏动性耳鸣的高致病区域。以上影像分析结论与搏动性耳鸣生物力学研究结论实现了相互佐证。
Abstract:The diagnosis of pulsatile tinnitus (PT) typically relies on medical imaging tests. However, due to the wide range of possible causes, there is still a lack of universally accepted diagnostic criteria with a clearly defined mechanism. This study aims to propose a neural network model for high-accuracy PT identification based on CT images of PT patients and non-PT individuals, as well as automatically label the high pathogenic regions to assist in diagnosis. Transfer learning based on the ResNet-v1-50 model was employed to identify PT using horizontal cross-sections of the middle temporal bone in the bone window. The high-weight regions for identification were labeled using the grad-CAM method. These regions, along with related anatomical structures, were statistically analyzed across three databases: large sections (entire cranium), medium sections (bilateral temporal bones), and small sections (right-side temporal bone), allowing for the gradual refinement of the area of interest and increased resolution of high-weight regions in classification. The best identification, which achieved 100% accuracy in the test set, came from the medium area that included both temporal bones. The high-weight regions identified in PT were concentrated in either bilateral or unilateral temporal bones, primarily involving the temporal bone air cells, tympanic antrum, sigmoid sinus cortical plate, and superior tympanum. The occurrence of PT is closely associated with temporal bone and nearby bone structures. PT patients have a probability of structural abnormalities in either bilateral or contralateral temporal bones, which are different from those without tinnitus. Specifically, bone structures including temporal bone air cells, tympanic antrum, sigmoid sinus cortical plate and tympanic cavity have a high probability of containing the primary pathogenic factors for PT. These imaging-based conclusions align with previous biomechanical findings, further corroborating the understanding of PT etiology.
-
图 2 大图数据集搏动性耳鸣分类的典型高权重区域分布
Figure 2. Typical high-weight region distribution of pulsatile tinnitus classification in whole graph dataset
图 3 中图数据集搏动性耳鸣分类的典型高权重区域
Figure 3. Typical high-weight region distribution of pulsatile tinnitus classification in middle graph dataset
图 4 小图数据集中搏动性耳鸣分类的典型高权重区域分布
Figure 4. Typical high-weight region distribution of pulsatile tinnitus classification in small graph dataset
表 1 迁移学习Resnet各数据集残差矩阵
Table 1. Transfer learning Resnet residual matrix for each dataset
实际情况 预测情况 大图数据集 中图数据集 小图数据集 耳鸣 无耳鸣 耳鸣 无耳鸣 耳鸣 无耳鸣 耳鸣 38 6 31 0 31 0 无耳鸣 5 27 0 26 3 34 
下载: 导出CSV
表 2 大图数据集高权重区域统计结果
Table 2. Statistical results of high-weight regions in the whole graph dataset
区域划分 个数 占比/% 中部 249 99.203 前部 2 0.797 后部 0 0
下载: 导出CSV
表 3 中图高权重区域统计结果
Table 3. Statistical results of high-weight regions in middle graph dataset
区域划分 个数 占比/% 双侧 101 34.948 仅颅内右侧 101 34.948 仅颅内左侧 85 29.412 后头骨 26 9.000 颅中央 18 6.228
下载: 导出CSV
表 4 右侧小图高亮区域统计结果
Table 4. Statistical results of high-weight regions in small graph dataset
区域划分 个数 占比/% 颞骨蜂房 156 56.522 鼓窦 133 48.188 乙状窦骨板 84 30.435 上鼓室 71 25.725 颞骨岩部 65 23.551
下载: 导出CSV
-
[1] 吴敏曼, 周家璇, 尹文艳, 等. 耳鸣的治疗进展[J]. 现代中西医结合杂志, 2009, 18(18): 2215-2217. doi: 10.3969/j.issn.1008-8849.2009.18.088WU M M, ZHOU J X, YIN W Y, et al. Progress in treatment of tinnitus[J]. Modern Journal of Integrated Traditional Chinese and Western Medicine, 2009, 18(18): 2215-2217(in Chinese). doi: 10.3969/j.issn.1008-8849.2009.18.088 [2] KRISHNAN A, MATTOX D E, FOUNTAIN A J, et al. CT arteriography and venography in pulsatile tinnitus: Preliminary results[J]. AJNR American Journal of Neuroradiology, 2006, 27(8): 1635-1638. [3] WEISSMAN J L, HIRSCH B E. Imaging of tinnitus: A review[J]. Radiology, 2000, 216(2): 342-349. doi: 10.1148/radiology.216.2.r00au45342 [4] PEGGE S A H, STEENS S C A, KUNST H P M, et al. Pulsatile tinnitus: differential diagnosis and radiological work-up[J]. Current Radiology Reports, 2017, 5(1): 5. doi: 10.1007/s40134-017-0199-7 [5] KIM S H, AN G S, CHOI I, et al. Pre-treatment objective diagnosis and post-treatment outcome evaluation in patients with vascular pulsatile tinnitus using transcanal recording and spectro-temporal analysis[J]. PLoS One, 2016, 11(6): e0157722. doi: 10.1371/journal.pone.0157722 [6] SONG J J, AN G S, CHOI I, et al. Objectification and differential diagnosis of vascular pulsatile tinnitus by transcanal sound recording and spectrotemporal analysis: a preliminary study[J]. Otology & Neurotology, 2016, 37(6): 613-620. [7] EISENMAN D J. Sinus wall reconstruction for sigmoid sinus diverticulum and dehiscence: a standardized surgical procedure for a range of radiographic findings[J]. Otology and Neurotology, 2011, 32(7): 1116-1119. doi: 10.1097/MAO.0b013e31822a1c7d [8] LIU W J, LIU Z H, ZHENG N, et al. Temporal bone pneumatization and pulsatile tinnitus caused by sigmoid sinus diverticulum and/or dehiscence[J]. BioMed Research International, 2015, 2015(1): 970613. [9] TOPAL O, ERBEK S S, ERBEK S, et al. Subjective pulsatile tinnitus associated with extensive pneumatization of temporal bone[J]. European Archives of Oto-Rhino-Laryngology, 2008, 265(1): 123-125. [10] TÜZ M, DOĞRU H, YEŞILDAĞ A. Subjective pulsatile tinnitus associated with extensive pneumatization of temporal bone[J]. Auris Nasus Larynx, 2003, 30(2): 183-185. doi: 10.1016/S0385-8146(03)00002-6 [11] WARD B K, CAREY J P, MINOR L B. Superior canal dehiscence syndrome: lessons from the first 20 years[J]. Frontiers in Neurology, 2017, 8: 177. doi: 10.3389/fneur.2017.00177 [12] LIU X, DU Z H, HAN T, et al. Quantitative analysis of blood flow in cerebral venous sinus with stenosis by patient-specific CFD modeling[J]. IEEE Access, 2018, 7: 3848-3854. [13] IBRAHIM A U, OZSOZ M, SERTE S, et al. Pneumonia classification using deep learning from chest X-ray images during COVID-19[J]. Cognitive Computation, 2021, 16(4): 1-13. [14] EL ASNAOUI K, CHAWKI Y, IDRI A. Automated methods for detection and classification pneumonia based on X-ray images using deep learning[M]//Yassine M, Youssef B, Mamoun A, et al. Artificial Intelligence and Blockchain for Future Cybersecurity Applications. Berlin: Springer, 2021: 257-284. [15] 刘飞, 张俊然, 杨豪. 基于深度学习的医学图像识别研究进展[J]. 中国生物医学工程学报, 2018, 37(1): 86-94. doi: 10.3969/j.issn.0258-8021.2018.01.012LIU F, ZHANG J R, YANG H. Research progress of medical image recognition based on deep learning[J]. Chinese Journal of Biomedical Engineering, 2018, 37(1): 86-94(in Chinese). doi: 10.3969/j.issn.0258-8021.2018.01.012 [16] BHATTACHARYYA A, BHAIK D, KUMAR S, et al. A deep learning based approach for automatic detection of COVID-19 cases using chest X-ray images[J]. Biomedical Signal Processing and Control, 2022, 71: 103182. doi: 10.1016/j.bspc.2021.103182 [17] JIANG H Y, XU J, SHI R J, et al. A multi-label deep learning model with interpretable grad-CAM for diabetic retinopathy classification[C]// IEEE Engineering in Medicine and Biology Society Annual International Conference. Pisscataway: IEEE Press, 2020: 1560-1563. [18] TIAN S, FAN X Y, WANG Y W, et al. An in vitro experimental study on the relationship between pulsatile tinnitus and the dehiscence/thinness of sigmoid sinus cortical plate[J]. Journal of Biomechanics, 2019, 84: 197-203. doi: 10.1016/j.jbiomech.2018.12.049 [19] TIAN S, WANG L Z, YANG J M, et al. Sigmoid sinus cortical plate dehiscence induces pulsatile tinnitus through amplifying sigmoid sinus venous sound[J]. Journal of Biomechanics, 2017, 52: 68-73. doi: 10.1016/j.jbiomech.2016.12.012 [20] TIAN S, FAN X Y, WANG Y W, et al. A study on relationship between pulsatile tinnitus and temporal bone pneumatization grade[J]. Computer Methods in Biomechanics and Biomedical Engineering, 2019, 22(7): 788-796. doi: 10.1080/10255842.2019.1593386 [21] HOFMANN E, BEHR R, NEUMANN-HAEFELIN T, et al. Pulsatile tinnitus: imaging and differential diagnosis[J]. Deutsches Arzteblatt International, 2013, 110(26): 451-458. -
下载:
点击查看大图
计量
- 文章访问数: 206
- HTML全文浏览量: 82
- PDF下载量: 6
- 被引次数: 0
