分析錐狀束電腦斷層拍攝參數與研發新式影像導引技術應用於定位根管系統

Abstract

根管治療時，對於複雜的根管型態或鈣化根管的評估，可使用CBCT (Cone-beam computed tomography，錐狀束電腦斷層)輔助診斷，如何在輻射劑量、病人安全與影像清晰度之間取得平衡，是我們要探討的課題。CBCT是良好的診斷工具，可讓牙醫師得知牙齒內部三維的解剖構造，但對於困難的狹窄、鈣化根管，仍希望有一儀器能夠即時呈現器械與牙齒之間的相對關係，進行影像導引。近年已有使用CBCT的電腦輔助手術系統應用於植牙手術，然而定位程序複雜且只能使用大範圍的CBCT，不適用於需要精細影像的根管治療。 本研究第一部份的目的為比較不同參數拍攝的CBCT對影像品質的影響。此部分共分為四個階段，使用有牙齒的乾燥人頭骨上顎第一大臼齒MB2根管 (Second mesiobuccal canal，近心頰側第二根管)為評估目標。第一階段請牙醫師以絕對目視等級分析法(Absolute visual grading analysis, Absolute VGA) 比較14x5 cm-FOV (Field of view, 照射範圍)、6x6 cm-FOV與4x4 cm-FOV MB2根管的清晰度，評分2分代表MB2根管影像清晰對比良好、1分代表MB2根管影像模糊但仍可辨識、0分代表MB2根管無法辨識。第二階段以6x6 cm-FOV，360度與180度拍攝的影像請受過訓練與未受訓練的牙醫師分別測量牙齒咬合面至目標點的距離。第三階段以廠商根管拍攝預設模式(4x4 cm-FOV /5mA/ 180度/ 慢速)為控制組，實驗組為改變旋轉角度(360度、180度)和原廠設定的三種不同掃描速度(快速、標準、慢速)，請五名受過訓練的牙醫師以相對目視等級分析法 (Relative VGA)評估MB2根管影像品質：評分2分代表影像品質與預設模式相當，MB2根管清晰可見；評分1分代表MB2根管影像模糊，品質較差；評分0分代表影像品質與預設模式相差甚遠， MB2根管無法辨識。第四階段以預設模式為控制組，實驗組為改變旋轉角度(360度、180度)、掃描速度(快速、標準、慢速)與電流(5 mA、2.5 mA)，請五名受過訓練的牙醫師以Relative VGA評估MB2根管影像品質。第二部分的目的為改良應用於定位根管系統的影像導引技術。以口外牙製作模型，基準註冊誤差 (Fiducial registration error, FRE)及定位成功率驗證使用牙齒特徵點(咬頭、凹窩)註冊、線掃描(掃描牙齒咬合面高點連線、掃描牙齒咬合面凹窩連線、掃描牙齒頰舌側表面)註冊的可行性。第三部分篩選出第二部分模型實驗定位成功率佳的組別，進入臨床試驗。 第一部分第一階段結果顯示14x5 cm-FOV不適合用來偵測MB2根管；第二階段6x6 cm-FOV，360度與180度的拍攝參數在偵測MB2根管的能力沒有顯著差異，且須以受過訓練的牙醫師來評估影像品質。第三階段結果顯示5mA/ 180度/ 慢速與預設模式的影像品質評分無差異，慢速模式比起快速、標準模式有較清晰的影像。第四階段結果顯示2.5mA/ 180度/慢速的影像品質評分與預設模式沒有顯著差異，將電流與旋轉角度減半，仍能維持與預設模式相當的影像品質。 第二部分的研究結果顯示使用牙齒的特徵點來定位根管系統是可行的，使用咬頭或凹窩當作特徵點定位的成功率沒有顯著差異；線掃描註冊的部分，路徑沿著牙齒咬合面高點的連線有最高的定位成功率。第三部分臨床試驗，招募15名受試者，排除影像不清晰、金屬散射干擾的三名受試者，共12名受試者納入試驗。結果顯示，在基準註冊誤差方面，臨床試驗與模型實驗使用特徵點註冊沒有顯著差異；臨床試驗與模型實驗使用線掃描註冊沒有顯著差異。在定位成功率方面，臨床使用特徵點註冊的定位成功率較模型實驗差，線掃描註冊的定位成功率則與模型實驗結果相符合。 綜合以上結果，拍攝上顎第一大臼齒MB2根管的CBCT時，欲維持與廠商設模式相同的影像品質，建議使用慢速的掃描模式，但可將預設拍攝電流5 mA降低至2.5 mA，且可將機器旋轉角度360度減半至180度以減少輻射劑量。本研究提出之應用於定位根管系統的影像導引技術，能使用牙齒咬合面高點的連線，以線掃描的方式註冊，且臨床試驗結果與模型實驗相符合，顯示此技術有潛力應用於臨床定位根管系統。

Dentists use CBCT (Cone-beam computed tomography) to evaluate complicated root canal system during root canal treatment. Finding the balance between radiation dose, patient-safety and acquiring a clear image is of crucial importance and is also an issue worth discussing. Although CBCT can provide three-dimensional images for detecting root canals, there is yet to be a synchronization between the instruments and pictures in real-time. Computer-aided navigation system had been developed and used in implant surgery, but it needs presurgical splints and large FOV (Field of view) CBCT images for registration, which are not suitable for endodontic treatment. The first part of this study was to compare the image quality of CBCT in detecting the MB2 (the Second mesiobuccal) canal of upper first molar with different image acquisition parameters. In the first section of this part, absolute VGA (visual grading analysis) was used to compare the image quality of MB2 canal in 14x5 cm-FOV, 6x6 cm-FOV and 4x4 cm-FOV. In the second section, two groups of trained and untrained dentists were asked to measure the distance between the occlusal surface of the target tooth to the target point in 6x6 cm-FOV, 360 and 180 -degree CBCT images. In the third and fourth section, we used the CBCT manufacturer’s default setting for root canal treatment as the control group. The goal was to reduce radiation dose by reducing rotation arcs, increasing scanning speed and lowering the tube current. Five trained dentists were asked to use relative VGA with 3 scales (2 as excellent image quality, the MB2 root canal of the first molar was precisely detected and was in good contrast compared to manufacturer’s default setting; 1 as acceptable image quality, the MB2 root canal of the first molar was hazy but still distinguishable; 0 as unacceptable image quality, the MB2 root canal of the first molar was barely detectable) to assess the image quality. The second part of this study was to improve current image tracking systems and to develop novel medical image tracking techniques for dental root canal navigation treatment. We used extracted human teeth as experiment models. FRE (Fiducial registration error) and the percentage of successfully reaching the targets were adopted to verify the feasibility of this novel image tracking technique. Those methods with a superior success rate of reaching the targets were applied to the third part of this research, which was a clinical trial. Results of the first part showed that the 14x5 cm-FOV was not suitable for locating the MB2 canal. No significant differences between 6x6 cm-FOV, full rotation and half rotation, and that the raters for CBCT image quality should be well-trained. In the third section, 4x4 cm-FOV, 5mA, with two different rotation arcs (360∘and 180∘) and three different scanning modes (fast, standard and slow) combined into six groups. Results showed that images taken with 5mA/ 360∘/ slow speed and 5mA/ 180∘/ slow speed had the same VGA rating, which indicated that slower scanning speed leads to better image quality. Then, we used 4x4 cm-FOV, 5 mA/ 2.5 mA, full rotation/ half rotation and three different scanning speeds combined into 12 groups. Results showed no statically difference between the ratings of 2.5mA/ 180∘/ slow speed group and the manufacturer’s default setting. Results of the second part of this study showed no difference in choosing cusps or fossae during feature point registration. Tracing the cusps and slopes of the teeth has preferred registration success rate during track scan registration. The FRE was smaller in track scan registration. The third part of this research, which was a clinical trial, involved 15 volunteers, three of them were excluded due to artefacts of CBCT images and the unclearness of CBCT images. Results of the clinical trial showed that the FRE of feature point registration was higher than model experiments but without statically significance. The FRE of track scan registration during clinical trial was higher than model experiments but without statically significance. The surface registration success rate of the feature point registration in the clinical trial was inferior to the model experiment, whilst the surface registration success rate of the track scan registration in the clinical corresponds to model experiments. Within the limitations of this study, we can conclude that in need of detecting the MB2 canal of upper first molars, to maintain the same image quality compared to manufacturer’s default mode, one should remain slow scanning speed. However, we could lower the tube current from 5 mA to 2.5 mA and could reduce the rotation arc from 360-degree to 180-degree while retaining the same image quality as the manufacturer default mode. The novel image tracking technique developed in this study allowed dentists to use cusps or fossae as feature point registration or scanning registration. The results of the clinical trial matched the model experiments, which indicated that this novel image tracking technique is feasible in locating root canal systems.