U-Net神经网络分割锥形束CT影像中下颌磨牙牙体与牙髓腔及其准确性验证

目的: 使用U-net卷积神经网络实现锥形束CT（cone-beam CT, CBCT）影像中下颌磨牙的牙体和牙髓腔的自动分割,采用基于显微CT（Micro-CT）扫描结果构建的三维模型作为金标准,评估分割准确性。方法: 从同济大学附属口腔医院放射科收集20组包含完整单侧下颌磨牙的口腔小视野CBCT数据,预处理后,由牙体牙髓病学专家使用MITK Workbench软件手动标注牙体与牙髓腔,作为U-net神经网络分割算法的训练集。另收集5颗下颌磨牙和相应的小视野CBCT数据,5组数据经相同预处理后作为测试集。随后由完成训练的神经网络和同一专家对测试集数据进行牙体和牙髓腔分割和三维重建。离体牙预处理后行Micro-CT扫描,将三维重建后获得的模型作为金标准。分别比较测试集数据中,专家的手动标注、神经网络分割结果与金标准两两之间的差异。采用Dice相似性系数（Dice similarity coefficient, DSC）、平均对称表面距离（average symmetric surface distance, ASSD）、Hausdorff距离（Hausdorff distance, HD）和形态差异分析对结果进行评估。采用SPSS 20.0软件包对数据进行统计学分析。结果: 神经网络分割结果与金标准相比,其牙体组的DSC为(95.30±1.01)%、ASSD为(0.11±0.02) mm、HD为(1.05±0.31) mm,牙髓腔组的DSC为(81.21±2.27)%、ASSD为(0.15±0.05) mm、HD为(3.29±1.85) mm,结合形态差异分析结果显示,神经网络的分割结果与金标准的牙体与髓室部分基本相似,但在根管部分,能分割出较粗的根管,对于根管下段和侧支根管等较细的根管分割能力有限。结论: 在现有实验条件下,以专家手动标注作为训练样本的U-net神经网络,实现了在CBCT影像上对下颌磨牙牙体与髓室的自动化精准分割。但对根管部分,其分割结果有待进一步提升。     

CBCT和根尖放射线片诊断根尖周炎的对比性研究

目的：比较应用CBCT和根尖放射线片对慢性根尖周炎的严重程度及病变范围进行评估的差异性,评价CBCT在慢性根尖周病变的诊断、治疗方案的确定及预后等方面的临床价值。方法：选取临床上有根尖周炎症状的53例患者（96颗牙）,均拍摄根尖放射线片和CBCT,分别由2位专家采用单盲法进行放射线影像学诊断,比较两种影像学方法对根尖周炎的存在及病变范围评估的差异性。结果：根尖放射线片发现82.3%,CBCT发现100%的患牙存在根尖周病变。并且在两者都诊断出根尖炎时,CBCT的PAI值明显高于根尖片。结论：在诊断慢性根尖周炎方面,CBCT扫描与根尖放射线片相比具有更高的敏感性和精确度。和根尖放射线片比较,CBCT对临床上根尖周炎的诊断、严重程度及预后判断都更有优势,从而能够更好的指导临床治疗。     

基于深度学习在曲面体层图像中人工智能辅助诊断系统初步研究

目的:基于深度学习对口腔曲面体层图像分析,开展人工智能在口腔常见疾病辅助诊断系统的研发,挖掘人工智能对曲面体层图像分割及辅助诊断价值.方法:回顾性纳入2000张口腔曲面体层片建立数据集(训练集1400张、测试集600张,累计标注82042例).运用基于卷积神经网络的深度学习算法,通过算法设计、模型训练和验证,构建口腔常见疾病智能影像诊断模型PanoNet,利用6个子网络模型分别执行不同口腔疾病的分割及识别.结果:PanoNet在恒牙列识别及龋病、根尖周炎、阻生牙、种植体、牙体修复术后等疾病识别中准确率、敏感度和特异度均高于85％(kappa>0.81);在牙槽骨吸收分级识别中准确率、敏感度、特异度分别为76.50％、75.25％、79.00％(kappa=0.44).结论:基于卷积神经网络的深度学习算法建立的口腔曲面体层图像诊断模型PanoNet能有效识别上述口腔常见疾病,体现人工智能在曲面体层片上对口腔常见疾病的影像辅助诊断的应用价值.     

基于MobileNetV3网络的龋病和根尖周炎根尖片的诊断

目的 研究深度学习技术智能诊断龋齿和根尖周炎的效果，初步探讨深度学习在口腔疾病诊断中的应用价值。方法 以2 298张包含健康牙齿、龋病、根尖周炎的根尖片数据集为研究对象，随机划分为1 573张训练集图像，233张验证集图像以及492张测试集图像。通过多种神经网络对比验证，选择性能较好的MobileNetV3网络模型应用于牙病诊断，并通过调整网络超参数优化模型。采用精确率、准确率、召回率和F1分数评估模型识别龋齿和根尖周炎的能力，并使用类激活热力图对网络模型性能进行可视化分析。结果基于MobileNetV3网络模型的牙齿病变检测算法对健康牙齿、龋病和根尖周炎进行分类的精确率、召回率和准确率分别为99.42%、99.73%和99.60%,F1分数为99.57%，达到了较为理想的智能诊断效果。可视化类激活热力图也显示出网络模型能够较为准确地提取牙科病变的特征。结论 基于MobileNetV3网络模型的牙齿病变检测算法能够排除图像质量和人为因素的干扰，具有较高的诊断准确率，可满足口腔医学教学和临床应用需求。     

人工智能在根尖周囊肿病理诊断中的初步应用

目的 研究人工智能应用于根尖周囊肿病理诊断的效果,初步探索人工智能在口腔病理学领域中的应用。方法 以87例根尖周囊肿的病理图像作为研究对象,构建U-net型结构的神经网络,将87幅根尖周囊肿的HE图像和标注图像分为训练集72幅图和测试集15幅图,分别用于训练模型和测试模型,最后利用目标级指标F1分数和像素级指标Dice系数以及受试者工作特征（receiver operating characteristic,ROC）曲线评价U-net网络模型在根尖周囊肿上皮识别中的能力。结果 U-net网络模型识别根尖周囊肿上皮的性能：F1分数为0.75,Dice系数为0.685,ROC曲线下面积为0.878。结论 通过人工智能构建的U-net网络模型在识别根尖周囊肿上皮时具有较好的分割结果,能够初步应用于根尖周囊肿的病理诊断,有望进一步大样本验证后逐步推广于临床。     

纳米银溶液用于慢性根尖周炎根管内封药的实验动物研究

目的体内研究0.1%纳米银溶液、氢氧化钙糊剂封药在Beagle犬慢性根尖周炎动物模型根管内的抑菌效果及对根尖周组织炎症刺激性的影响。 方法选取3只成年雄性Beagle犬的30颗双尖牙,拍摄锥形束CT（CBCT）确认构建慢性根尖周炎模型后按照冠向下预备法预备30颗双尖牙。将实验牙采用随机数字表法分为3组,每组10颗,分别封药0.1%纳米银溶液、氢氧化钙糊剂、空白对照（只冲洗,不封药）。根管封药4周后再次拍摄CBCT,使用CBCT机自带软件DCTViewer 2.0测量术前、术后冠状面、矢状面视图中根尖透射影像面积;处死Beagle犬,取部分牙根及根尖周组织染色,在光学显微镜下测量根管壁的抑菌深度并观察根尖周组织的炎症程度。使用SPSS 24.0统计软件对实验数据进行统计分析。 结果0.1%纳米银溶液组、氢氧化钙糊剂组、空白对照组犬牙根尖周透射影像面积在冠状位分别减少（4.8 ± 3.4）、（1.6 ± 1.6）和（2.2 ± 2.6）mm²,组间差异具有统计学意义（F = 5.607,P = 0.002）,在矢状位分别减少（6.3 ± 3.9）、（4.1 ± 4.6）和（1.3 ± 2.8）mm²,组间差异也有统计学意义（F = 6.869,P<0.001）;根管壁的抑菌深度分别为243（159,372）、123.5（90,134）和104.5（81,135）μm,组间差异有统计学意义（χ² = 18.519,P<0.001）;0.1%纳米银溶液组根尖周组织的炎症程度3例为0级、4例为1级、3例为2级,氢氧化钙糊剂组1例0级、5例1级、3例2级、1例3级,空白对照组1例0级、3例1级、4例2级、2例3级,组间差异无统计学意义（χ² = 3.955,P = 0.052）。 结论0.1%纳米银溶液能够促进Beagle犬根尖周组织病变的愈合、抑制根管内细菌的生长,且对根尖周组织无明显炎症刺激性。     

锥形束CT对后牙根尖周炎骨病损的评价作用

目的:以锥形束CT(cone-beam computed tomography,CBCT)为标准,评价X线片在诊断后牙根尖周炎骨病损中的作用。方法:收集门诊同时拍摄X线片和CBCT图像的病例80例,共106颗后牙,包括前磨牙和磨牙各53颗,其中健康牙58颗,临床诊断为慢性牙髓炎11颗,诊断为慢性根尖周炎34颗(含8颗根管治疗后的患牙),根管治疗后表现正常的牙3颗。由2名有经验的医师对CBCT图像及X线片进行判读,确定根尖周指数(periapical index,PAI)分级。采用SPSS13.0软件包对所得数据进行χ2检验。结果:分别对106颗疑似患牙的CBCT图像与X线片进行判读,根尖周炎的检出率分别为59.4%和39.6%,差异有显著性(χ2=8.32,P<0.01)。X线片为二维影像,其结构重叠产生伪影,使病变范围界限不清,而CBCT三维图像则对病损范围有明确的显示,有利于疾病的诊断与治疗。另外,X线片不能表现CBCT显示的骨皮质破坏情况。结论:CBCT图像诊断根尖周炎比X线片更有临床价值,可展现X线片无法显示的细节,对疾病的破坏范围和相关结构毗邻显示更清楚,从而准确划分根尖周炎的分级,为临床正确诊断以及科学制定治疗计划提供有效的依据。     

应用锥形束CT诊断颞下颌关节骨关节病的探讨

目的探讨锥形束CT（cone beam CT，CBCT）在颞下颌关节骨关节病诊断中的应用前景。方法临床诊断为颞下颌关节骨关节病（炎）、不可复（可复）性盘前移位伴骨关节病患者共48例（96侧关节）。48例同时拍摄经咽侧位x线平片和CBCT，比较两种x线检查方法的病变检出率、医师判断的重复性和一致性。结果颞下颌关节骨关节病x线表现分为6型：髁突表面皮质骨模糊消失型（I型）、表面缺损破坏型（Ⅱ型）、髁突磨平型（Ⅲ型）、骨质硬化型（IV型）、骨质增生型（V型）、囊样变型（Ⅵ型）。CBCT的检出率分别为65．63％、37．50％、27．08％、31．25％、28．13％、1．04％；经咽侧位x线平片的检出率分别为52．08％、19．79％、32．29％、23．96％、12．50％、2．08％。对每一型病变的程度和范围，同一医师两次判断或不同医师之间，对I、Ⅱ型病变的评判，经咽侧位x线平片和CBCT均有高度的一致性，Kappa值大于0．60。结论除Ⅲ型外，CBCT对每一类型的病变均有很高的检出率，所显示的病变及其部位清晰、明确。CBCT清晰的病变影像、明确的病变部位和显示多层面病变的优势，使其有望成为颞下颌关节骨关节病判定病变程度、预后以及药物治疗后效果的定量评价手段。     

锥形束CT与根尖片对beagle犬显微根尖手术的预后评估

目的:采用根尖片和锥形束CT(cone beam computed tomography, CBCT)对beagle犬实验性根尖周炎显微根尖手术的预后进行评估。方法:将3条beagle犬18颗前磨牙髓腔暴露于口腔环境中8周, 拍摄根尖片及CBCT显示36个牙根均形成实验性根尖周炎。根管治疗后行显微根尖外科手术,术后即刻、6月分别拍摄根尖片和CBCT,根据根尖片和CBCT的根尖透射影像面积,比较二者在识别根尖骨质缺损的差别。结果:显微根尖手术后即刻CBCT矢状面、冠状面根尖透射影像面积均大于根尖片,差别具有统计学意义(P=0.000,P=0.026);CBCT矢状面、冠状面比较两组间差别无统计学意义(P=0.070)。显微根尖手术后6月复查CBCT矢状面、冠状面根尖透射影像面积均大于根尖片,差别具有统计学意义(P=0.000,P=0.012);CBCT矢状面根尖透射影像面积大于冠状面,两组间差别有统计学意义(P=0.001)。结论:CBCT在根尖周骨质缺损的识别优于根尖片,CBCT在根尖手术骨缺损预后评估中是一项有效的评价手段。     

成釉细胞瘤的锥形束CT表现分析

目的 对成釉细胞瘤的锥形束CT（CBCT）表现进行总结分析,为其临床诊断提供依据。方法 对病理确诊的有完整CBCT影像资料的37例成釉细胞瘤病例进行回顾性分析,观察其不同CBCT表现。结果 37例资料中,31例为原发病例,6例为复发病例。CBCT表现： 37例病变中,有 36例（97.3%）为骨内型,其中多房型 17例,单房型 17例,蜂窝型 2例; 1例（2.7%）为骨外型,即软组织型。88.2%（15/17）多房型病变内见舌形嵴;34例（94.4%）颌骨呈唇颊侧或（和）腭舌侧膨隆,颌骨骨密质局部不连续。结论 CBCT检查能精确地反映成釉细胞瘤病变形态和内部结构,对其术前诊断及手术计划具有重要的指导意义。     

Clinical Factors Associated with Apical Periodontitis Visible on Cone-beam Computed Tomography but Missed with Periapical Radiographs: A Retrospective Clinical Study

IntroductionCone-beam computed tomographic (CBCT) imaging is useful in detecting apical periodontitis, which is often missed in periapical (PA) radiographs. This study aimed to identify preoperative predictors correlated with the presence of apical periodontitis visible only in CBCT images and to investigate the important characteristics of such lesions.MethodsIn total, 332 consecutive patients with both PA radiographs and CBCT images were enrolled in this study. The patients’ clinical charts were reviewed retrospectively to collect information regarding their symptoms and diagnoses. Periapical lesions were assessed using a modified CBCT PA index by 2 endodontists. Patient-related factors (age, sex, and symptoms) and tooth-related factors (tooth type, location, pulp status, and pulpal diagnosis) were assessed to determine their relationships with the presence of apical periodontitis visible only in CBCT images.ResultsApical periodontitis was detected in 24.6% and 35.5% of untreated teeth by PA radiographs and CBCT images, respectively. In a multivariate logistic regression analysis, pulp necrosis was significantly correlated with the presence of apical periodontitis visible only in CBCT images (odds ratio = 5.401; 95% confidence interval, 1.911–15.265; P = .001); the involvement of molars showed borderline nonsignificant correlation (odds ratio = 2.843; 95% confidence interval, 0.990–8.164; P = .052). Lesion sizes smaller than 2 mm in diameter and the involvement of molars were significant factors of lesions visible only in CBCT images (P < .05).ConclusionsPulp necrosis was a preoperative predictor of apical periodontitis visible only in CBCT images. This research could provide a proper indication for CBCT imaging at diagnostic stages.     

Artificial Intelligence for Fast and Accurate 3-Dimensional Tooth Segmentation on Cone-beam Computed Tomography

IntroductionTooth segmentation on cone-beam computed tomographic (CBCT) imaging is a labor-intensive task considering the limited contrast resolution and potential disturbance by various artifacts. Fully automated tooth segmentation cannot be achieved by merely relying on CBCT intensity variations. This study aimed to develop and validate an artificial intelligence (AI)-driven tool for automated tooth segmentation on CBCT imaging.MethodsA total of 433 Digital Imaging and Communications in Medicine images of single- and double-rooted teeth randomly selected from 314 anonymized CBCT scans were imported and manually segmented. An AI-driven tooth segmentation algorithm based on a feature pyramid network was developed to automatically detect and segment teeth, replacing manual user contour placement. The AI-driven tool was evaluated based on volume comparison, intersection over union, the Dice score coefficient, morphologic surface deviation, and total segmentation time.ResultsOverall, AI-driven and clinical reference segmentations resulted in very similar segmentation volumes. The mean intersection over union for full-tooth segmentation was 0.87 (±0.03) and 0.88 (±0.03) for semiautomated (SA) (clinical reference) versus fully automated AI-driven (F-AI) and refined AI-driven (R-AI) tooth segmentation, respectively. R-AI and F-AI segmentation showed an average median surface deviation from SA segmentation of 9.96 μm (±59.33 μm) and 7.85 μm (±69.55 μm), respectively. SA segmentations of single- and double-rooted teeth had a mean total time of 6.6 minutes (±76.15 seconds), F-AI segmentation of 0.5 minutes (±8.64 seconds, 12 times faster), and R-AI segmentation of 1.2 minutes (±33.02 seconds, 6 times faster).ConclusionsThis study showed a unique fast and accurate approach for AI-driven automated tooth segmentation on CBCT imaging. These results may open doors for AI-driven applications in surgical and treatment planning in oral health care.     

Evaluating the Periapical Status of Teeth with Irreversible Pulpitis by Using Cone-beam Computed Tomography Scanning and Periapical Radiographs

Introduction

The purpose of this study was to compare the prevalence of apical periodontitis (AP) on individual roots of teeth with irreversible pulpitis viewed with periapical (PA) radiographs and cone-beam computed tomography (CBCT) scans.

Methods

PA radiographs and CBCT scans were taken of 138 teeth in 130 patients diagnosed with irreversible pulpitis (symptomatic and asymptomatic). Two calibrated examiners assessed the presence or absence of AP lesions by analyzing the PA and CBCT images. A consensus was reached in the event of any disagreement. The data were analyzed using the hypothesis test, and significance was set at P ≤ .05.

Results

Three hundred seven paired roots were assessed with both PA and CBCT images. A comparison of the 307 paired roots revealed that AP lesions were present in 10 (3.3%) and absent in 297 (96.7%) pairs of roots when assessed with PA radiography. When the same 307 sets of roots were assessed with CBCT scans, AP lesions were present in 42 (13.7%) and absent in 265 (86.3%) paired roots. The prevalence of AP lesions detected with CBCT was significantly higher in the symptomatic group compared with the asymptomatic group (P < .05). An additional 22 roots were identified with CBCT alone.

Conclusions

The present study highlights the advantages of using CBCT for detecting AP lesions, especially in teeth with symptomatic irreversible pulpitis.     

Micro–Computed Tomography–Guided Artificial Intelligence for Pulp Cavity and Tooth Segmentation on Cone-beam Computed Tomography

IntroductionThis study proposes a novel data pipeline based on micro–computed tomographic (micro-CT) data for training the U-Net network to realize the automatic and accurate segmentation of the pulp cavity and tooth on cone-beam computed tomographic (CBCT) images.MethodsWe collected CBCT data and micro-CT data of 30 teeth. CBCT data were processed and transformed into small field of view and high-resolution CBCT images of each tooth. Twenty-five sets were randomly assigned to the training set and the remaining 5 sets to the test set. We used 2 data pipelines for U-Net network training: one manually labeled by an endodontic specialist as the control group and one processed from the micro-CT data as the experimental group. The 3-dimensional models constructed using micro-CT data in the test set were taken as the ground truth. The Dice similarity coefficient, precision rate, recall rate, average symmetric surface distance, Hausdorff distance, and morphologic analysis were used for performance evaluation.ResultsThe segmentation accuracy of the experimental group measured by the Dice similarity coefficient, precision rate, recall rate, average symmetric surface distance, and Hausdorff distance were 96.20% ± 0.58%, 97.31% ± 0.38%, 95.11% ± 0.97%, 0.09 ± 0.01 mm, and 1.54 ± 0.51 mm in the tooth and 86.75% ± 2.42%, 84.45% ± 7.77%, 89.94% ± 4.56%, 0.08 ± 0.02 mm, and 1.99 ± 0.67 mm in the pulp cavity, respectively, which were better than the control group. Morphologic analysis suggested the segmentation results of the experimental group were better than those of the control group.ConclusionsThis study proposed an automatic and accurate approach for tooth and pulp cavity segmentation on CBCT images, which can be applied in research and clinical tasks.     

Comparison of intraoral radiography and limited cone beam computed tomography for the assessment of root-fractured permanent teeth

Abstract –  Aim : To compare intraoral occlusal (OC) and periapical (PA) radiographs vs. limited cone beam computed tomography (CBCT) in diagnosing root-fractured permanent teeth. Material and methods :  In 38 patients (mean age 24 years, range 8–52 years) with 44 permanent teeth with horizontal root fractures, intraoral radiographs (PA and OC) and limited CBCT were used to evaluate the location (apical, middle, cervical third of the root) and angulation of the fracture line. Furthermore, the conventional radiographs and CBCT images were compared for concordance of fracture location. Results :  In the PA and OC radiographs, 28 fractures (63.6%) were located in the middle third of the root, 11 (25.0%) in the apical third and 5 (11.4%) in the cervical third. The PA/OC radiographs and the sagittal CBCT images (facial aspect) yielded the same level of root fracture in 70.5% of cases (31 teeth; 95% CI: 54.1–82.7%). The PA/OC radiographs and sagittal CBCT images (palatal aspect) showed the same level of root fracture in 31.8% of cases. There was a statistically significant association between the angle at which the root fracture line intersected the axis of the tooth and the level of root fracture in the facial aspect of the sagittal CBCT images. Conclusions :  The diagnosis of the location and angulation of root fractures based on limited CBCT imaging differs significantly from diagnostic procedures based on intraoral radiographs (PA/OC) alone. The clinical significance for treatment strategies and for the prognosis of root-fractured teeth has to be addressed in future studies.