Deep learning for automatic detection of cephalometric landmarks on lateral cephalometric radiographs using the Mask Region-based Convolutional Neural Network: a pilot study.

OBJECTIVE: We examined the effectiveness and feasibility of the Mask Region-based Convolutional Neural Network (Mask R-CNN) for automatic detection of cephalometric landmarks on lateral cephalometric radiographs (LCRs). STUDY DESIGN: In total, 400 LCRs, each with 19 manually identified landmarks, were collected. Of this total, 320 images were randomly selected as the training dataset for Mask R-CNN, and the remaining 80 images were used for testing the automatic detection of the 19 cephalometric landmarks, for a total of 1520 landmarks. Detection rate, average error, and detection accuracy rate were calculated to assess Mask R-CNN performance. RESULTS: Of the 1520 landmarks, 1494 were detected, for a detection rate of 98.29%. The average error, or linear deviation distance between the detected points and the originally marked points of each detected landmark, ranged from 0.56 to 9.51 mm, with an average of 2.19 mm. For detection accuracy rate, 649 landmarks (43.44%) had a linear deviation distance less than 1 mm, 1020 (68.27%) less than 2 mm, and 1281 (85.74%) less than 4 mm in deviation from the manually marked point. The average detection time was 1.48 seconds per image. CONCLUSIONS: Deep learning Mask R-CNN shows promise in enhancing cephalometric analysis by automating landmark detection on LCRs, addressing the limitations of manual analysis, and demonstrating effectiveness and feasibility.

The Stability Evaluation of Two Internal Fixation Patterns of Extracapsular Condylar Fracture Based on One Novel Measuring Method.

PURPOSE: The most stable internal fixation pattern for extracapsular condylar fracture (ECF) has been controversial. In this study we aimed to evaluate the stability of 2 common internal fixation patterns using 1 or 2 miniplates separately. One novel measuring method based on the angle of miniplates' localization was introduced. MATERIALS AND METHODS: Twenty-seven patients with 30 sides of extracapsular condylar fracture were enrolled in this retrospective study. All cases were performed open reduction and internal fixation using 1 or 2 titanium plates with both the postoperative immediate CT (T1) and the over 6 months' follow-up CT (T2). Mandibles and implants were segmented and reconstructed respectively using SIMPANT 14.04 software. For 1 miniplate group, the sagittal crossing angle (<1) between miniplate (P1) and posterior border of ipsilateral ramus was measured. For 2 miniplates group, the sagittal crossing angle (<2) between 2 miniplates (P2a, P2b) was measured. Both anteroposterior (∠AP) and mediolateral angle change (∠ML) of each miniplate between T1 and T2 was measured. The stability of 2 internal fixation patterns was measured and evaluated based on the angle change of miniplates. And also the stability of each internal fixation pattern could be investigated based on the relationship between the placement patterns of miniplates and the angle change. RESULTS: Fifty-two miniplates were finally segmented and reconstructed successfully, including 8 of 1 miniplate group and 44 of 2 miniplate group. For 1 miniplate group, the average ∠AP and ∠ML were 6.10° and 8.54°, respectively. For 2 miniplate group, the average ∠AP and ∠ML of P2a were 3.02° and 2.56°, respectively, as well as 3.12° and 3.07° of P2b. CONCLUSIONS: The novel measuring method based on the angle of miniplates' localization showed potential for the stability evaluation of internal fixation of condylar fracture. In summary, the internal fixation patterns using 2 miniplates shows better stability than that of 1 miniplate.

Convolutional Neural Networks for Immediate Surgical Needle Automatic Detection in Craniofacial X-Ray Images.

PURPOSE: Immediate X-ray examination is necessary while the surgical needle falls off during operation. In this study, one convolutional neural network (CNN) model was introduced for automatically surgical needle detection in craniofacial X-ray images. MATERIALS AND METHODS: The craniofacial surgical needle (5-0, ETHICON, USA) was localized in 8 different anatomic regions of 2 pig heads for bilateral X-ray examination separately. Thirty-two images were obtained finally which were cropped into fragmented images and divided into the training dataset and the test dataset. Then, one immediate needle detection CNN model was developed and trained. Its performance was quantitatively evaluated using the precision rate, the recall rate, and the f2-score. One 8-fold cross-validation experiment was performed. The detection rate and the time it took were calculated to quantify the degree of difference between the automatic detection and the manual detection by 3 experienced clinicians. RESULTS: The precision rate, the recall rate, and the f2-score of the CNN model on fragmented images were 98.99%, 92.67%, and 93.85% respectively. For the 8-fold cross-validation experiments, 26 cases of all the 32 X-ray images were automatically marked the right position of the needle (detection rate of 81.25%). The average time of automatically detecting one image was 5.8 seconds. For the 3 clinicians, 65 images of all the 32× 3 images were checked right (detection rate of 67.7%) with the average time-consuming of 33 seconds. CONCLUSION: In summary, after training with a large dataset, the CNN model showed potential for immediate surgical needle automatic detection in craniofacial X-ray images with better detection accuracy and efficiency than the conventional manual method.

Evaluation of the postoperative stability of a counter-clockwise rotation technique for skeletal class II patients by using a novel three- dimensional position-posture method.

The aim of this study is to evaluate the postsurgical stability of skeletal class II patients after performing a counter-clockwise rotational (CCWR) procedure for the maxilla-mandibular complex (MMC) by using a novel Three-dimensional (3D) Position-Posture(P-P) measuring method. Twenty-five patients (5 males and 20 females) were included in this study. The postoperative CT scans of the skull were taken before surgery(T0), 3-7 days (T1), 3 months (T2), and 6 months (T3) after surgery. Specific anatomic landmarks were chosen to determine the position of the segments, while three equally perpendicular planes were created to describe their posture. The results show that the linear relapse of maxillary landmarks during the follow-up were acceptable (≤0.5 mm). The relapse of maxillary pitch plane at 6-months follow-up is 1.52°, which is acceptable. There was a significant pitch plane relapse of the mandibular-body segment with an average of 1.86° between T1 and T2 models, 3.28° between T1 and T3 models. There was no significant difference between roll and yaw planes during the follow-up. We therefore conclude that the P-P method could be used to accurately analyze the postsurgical stability of skeletal class II orthognathic surgery cases. For CCWR procedures, it was also shown that the there is a tendency for recurrence most specially on the body of the mandible.

A Novel System for Navigation-and Robot-Assisted Craniofacial Surgery: Establishment of the Principle Prototype.

PURPOSE: The authors aimed to develop 1 novel navigation-guided robotic system for craniofacial surgery to improve accuracy during operation. MATERIALS AND METHODS: A new 7-DOF (7-degree-of-freedom) robotic arm was designed and manufactured. Based on our self-developed navigation system TBNAVIS-CMFS, the key technique of integration was studied. A phantom skull model was manufactured based on computed tomography image data and used for the preexperimental study. Firstly, virtual planning was achieved through the TBNAVIS-CMFS, where the Le Fort I procedure was executed through simulation. Then, the actual Le Fort 1 osteotomy was expected to perform with the use of the robotic arm following the instructions from the navigation system. RESULTS: The theoretical prototype of navigation-guided robotic system for craniofacial surgery was established successfully, which performed the planned Le Fort I procedure with the whole process visible on the screen. CONCLUSIONS: The technical method of navigation-guided robotics system, allowing the operator to practice the virtual planning procedure through navigation system as well as perform the actual operation thru the robotic arm, could be regarded as a valuable option for benefiting craniofacial surgeons.

Mirroring Tool: The Simplest Computer-Aided Simulation Technology?

PURPOSE: The aims of this study were to present our experience in the use of the mirroring tool in computer-aided simulation (CAS) as well as to describe its diversified uses and indications systematically. MATERIALS AND METHODS: Between 2006 and 2012, 144 patients underwent preoperative CAS with mirroring tool developed by our department called Accu-Navi System. Various mirroring methods were performed and classified systemically. The outcome was evaluated thru its performance intraoperatively and the results postoperatively. RESULTS: The optimal preoperative plans were achieved for all 144 patients using the mirroring tool. This provided the surgeon with accurate surgical information preoperatively and intraoperatively. There were 138 operations performed based on the preoperative plan. The duration of the operations was significantly shortened with improved surgical accuracy and safety. The mean discrepancy was confirmed to be less than 1.0 mm by comparing postoperative computed tomography with the preoperative plan. Postoperative esthetics improved remarkably along with patients' satisfaction without complications thru long-term follow-up evaluations. CONCLUSIONS: The mirroring tool with its many uses can be a valuable CAS procedure for craniofacial surgery.

Features and management of intracranial mandibular condyle dislocation after trauma.

BACKGROUND: Dislocation of the mandibular condyle into the middle cranial fossa is rare in clinics. It often occurs when the mouth is open wide during the injury. It causes restriction of mandibular motion, lower facial asymmetry, pain in the temporomandibular joint (TMJ), etc. OBJECTIVE: To introduce the features of intracranial mandibular condyle dislocation and discuss the management to this kind of trauma. MAJOR FINDINGS: In this paper, the authors present two cases, describing the diagnosis, surgical management, and 1-year follow-up evaluation. The results of the authors' treatment to intracranial mandibular condyle dislocation were satisfactory and stable, and no surgical complications were detected. CONCLUSION: Advanced imaging studies are mandatory for exact diagnosis and successful treatment of intracranial mandibular condyle dislocation, and individualized management is recommended.

Navigation-guided lateral gap arthroplasty as the treatment of temporomandibular joint ankylosis.

PURPOSE: This article presents a novel method of navigation-guided lateral gap arthroplasty (LGA) in the treatment of temporomandibular joint ankylosis (TMJA). MATERIALS AND METHODS: Six patients with unilateral TMJA from 2007 through 2011 were included in this study. Presurgical planning was performed to determine the amount and extent of ankylosed bone to be resected using the simulation platform. Minimum follow-up was 6 months. Patients were monitored for complications and signs of recurrence. Maximum mouth opening (MO) was measured and compared intra- and postoperatively. RESULTS: Preoperative planning was performed at the STN or Accu-Navi workstation. The amount and extent of ankylosed bone to be resected was determined. All 6 LGAs were completed successfully using real-time instrument- and pointer-based navigation. Measurements performed intraoperatively showed that the mean for maximum MO was about 35 to 40 mm and remained the same postoperatively. Follow-up evaluation showed remarkable improvement in function and esthetics, with no signs of recurrence. CONCLUSION: Navigation-guided LGA can be regarded a viable option for performing this delicate and complicated surgical procedure.

Real-time image-guided recontouring in the management of craniofacial fibrous dysplasia.

OBJECTIVE: This report presents our experience using computer-aided recontouring in the surgical management of complex craniofacial fibrous dysplasia with the use of a navigation system developed by the authors. STUDY DESIGN: A total of 21 patients (9 men and 12 women) with craniofacial fibrous dysplasia and mean age of 23 years (range, 17-32 years) were included in this study from 2007 to 2012. By creating a mirror image of the unaffected side on the affected side as a virtual treatment template, we completed the recontouring procedures in real time with the aid of the navigation system that we developed (Accu-Navi). The surgical outcome was assessed by superimposing the postoperative computed tomography (CT) images onto the preoperative CT images. RESULTS: The precise preoperative simulation and intraoperative navigation enabled the surgeon to complete the recontouring procedure visually. Postoperative CT was compared with the preoperative plan, yielding an average discrepancy of <1.0 mm. Postoperative follow-up found that both facial aesthetics and patient satisfaction improved remarkably. CONCLUSIONS: Navigation-guided recontouring shows benefits in improving accuracy and safety for this complicated procedure.

Image-guided surgical navigation for removal of foreign bodies in the deep maxillofacial region.

PURPOSE: Most trauma surgeons encounter numerous penetrating injuries. Some foreign bodies can cause pain, infection, and discomfort to the patient. Serious functional disorders also are likely to occur. Foreign bodies in critical areas must be removed. This report describes the use of image-guided technology for the removal of foreign bodies deep in the maxillofacial region. PATIENTS AND METHODS: From 2008 through 2011, 5 patients with foreign bodies in the maxillofacial area underwent image-guided removal at the authors' department. The STN navigation system (Stryker-Leibinger, Freiburg, Germany) was used for surgical planning and intraoperative navigation. Preoperatively, computerized tomography and digital subtraction angiography were used to create 3-dimensional views of the region to aid surgeons in more accurately defining the spatial location of the foreign object. During surgery, the foreign objects and surgical instruments were visualized on the screen. RESULTS: In all 5 cases, the foreign bodies were removed by minimally invasive access without any complications. Surgical time was approximately 40% shorter compared with the conventional technique of not using image-guided navigation. A 1-year postoperative evaluation showed that the patients' complaints and symptoms had resolved, function was restored, and esthetics were remarkably improved. CONCLUSION: Navigation-guided removal of foreign bodies in the complex, deep maxillofacial region in proximity to vital areas can be regarded an ideal and valuable option for these potentially complicated procedures.

Reconstruction of complex mandibular defect with computer-aided navigation and orthognathic surgery.

Reconstructive surgery for complex mandible defect with condyle is challenging for surgeons. Computer-aided planning and navigation is a useful adjunct for surgeons, which can improve visualization and orientation, insertion, mirroring, creation of planning models, and transformation of a surgical plan accurately. Here, we present a case of complex mandible defect reconstruction with computer-aided navigation and orthognathic surgery.

Simultaneous Le Fort I, II, and III osteotomies for correction of midface deficiency in Apert disease.

Le Fort III osteotomy was usually applied to correct midface hypoplasia in Apert syndrome, and various surgical modifications have been developed in recent years. In this article, we reported the simultaneous Le Fort I, II, and III osteotomies for segmental advancement of midface deficiency involving nasal bones, zygoma, inferior orbital rims, and maxilla in an adult Chinese patient with Apert syndrome. To achieve the ideal advancement of different parts of midface simultaneously, we divided the midface into 4 segments, including nasal bone combined with upper portion of maxilla, lower portion of maxilla, and left and right zygoma, with simultaneous Le Fort I, Le Fort II, and Le Fort III osteotomies, and each segment was repositioned as required respectively to obtain ideal facial aesthetics and favorable occlusion. The long-term stability of bony segment advancements also was observed during 7-year follow-up. Compared with segmental distraction osteogenesis or multiple-stage surgery, which mainly applied to younger patients with more severe midface hypoplasia, this single-stage strategy offered a reliable surgical alternative for treating adult patient with midface hypoplasia that should be corrected in different levels.

A novel device for preoperative registration and automatic tracking in cranio-maxillofacial image guided surgery.

OBJECTIVE: Two key issues in image guided surgery are accurate patient-to-image registration and ongoing tracking of the patient's motion. To address these concerns, a novel device for preoperative registration and automatic tracking was designed, and the accuracy attainable with the device was evaluated in experiments with a skull and in a clinical study. METHODS: The device consists of a system of four titanium screws and four fluorescent spheres fixed to carbon bars which can be easily mounted on the maxillary dentition splint. Before surgery, CT image data of a skull with the device in place was acquired and registered in a navigation system. The rigidity and reproducibility of positioning of the device were measured in 15 repeated CT acquisitions of the skull with the device in place. The registration accuracy was compared to that obtained using micro-screw markers fixed to the maxillary alveolus. To determine the potential of the device in aiding image guided cranio-maxillofacial surgery, registration accuracy and surgical outcome were assessed. RESULTS: Fifteen tests were performed for CT scanning with no loosening of the splint and device. The arithmetic mean of the standard deviation (SD) ranged from 0.47 mm to 0.70 mm. When the device was used for registration, the mean deviations for the eight anatomical structures investigated ranged from 0.56 mm at the left infra-orbital foramen to 0.96 mm at the right temple. Compared with the method in which titanium screws are fixed to the maxillary alveolus, the target registration error (TRE) obtained using the new device was much less. Using this device, clinical reduction of a zygomatic-orbital-maxillary complex fracture was successfully completed with a registration discrepancy of less than 0.5 mm. CONCLUSIONS: By successfully addressing the two key issues of image guided surgery, the device could be considered accurate and potentially useful for assisting in cranio-maxillofacial surgery.

Navigation-guided correction of midfacial post-traumatic deformities (Shanghai experience with 40 cases).

PURPOSE: To evaluate the effectiveness of navigation-guided surgical correction as the treatment of midfacial post-traumatic deformities. PATIENTS AND METHODS: A total of 40 patients with midfacial post-traumatic deformities from 2007 to 2010 were involved in the present study. The preoperative planning and simulation data sets, including the generation of virtual models with the mirror tool, were used as a virtual template to guide surgical correction of those deformities by intraoperative instrument- and/or probe-based navigation. The outcome was checked by both superimposing the postoperative 3-dimensional computed tomography model on the preoperative planning model and clinical examination. RESULTS: Through registration, an accurate match between the intraoperative actual position and 3-dimensional virtual model was achieved with a systematic error of 1 mm difference. With the guidance of the navigation system, the average deviation of the morphologic change after surgical correction was less than 1 mm according to the image comparison between the preplanned and postoperative computed tomography. The 1- to 2-year follow-up evaluation showed that the clinical complaint symptoms were alleviated, and the postoperative function and esthetics improved remarkably. CONCLUSIONS: Navigation-guided correction in the treatment of midfacial post-traumatic deformities can be regarded as an ideal and valuable option for this potentially complicated procedure.