"Apathetic look" is a valuable indicator of intraoperative supplementary motor area syndrome during awake craniotomy.

Resection of a glioma from the dorsomedial frontal lobe, including the supplementary motor area (SMA), can result in postoperative SMA syndrome. SMA syndrome may occur during awake craniotomies. However, it is often difficult to intraoperatively distinguish between motor dysfunction due to pyramidal tract damage from that due to SMA syndrome. Patients with suspected intraoperative SMA syndrome are indifferent to their surroundings, have stiff facial muscles, and maintain a fixed gaze. We defined this condition as "apathetic look." The present study aimed to investigate whether intraoperative "apathetic look" is useful for identifying intraoperative SMA syndrome in patients with glioma close to motor-related areas, including the SMA, during awake craniotomy. This study included 33 consecutive patients with glioma included in the SMA. We excluded patients whose tumors extended to motor-related areas. We also assessed whether intraoperative SMA syndrome occurred in each patient. We evaluated the correlation between the occurrence of intraoperative SMA syndrome and various clinical factors, including intraoperative "apathetic look." Of the 33 patients, 12 had intraoperative SMA syndrome. Intraoperative "apathetic look" showed strong correlation with intraoperative SMA syndrome (p < 0.0001). Additionally, higher extent of resection (EOR) and resection of the corpus callosum showed a significantly higher incidence of intraoperative "apathetic look." All 12 patients with intraoperative SMA syndrome showed intraoperative "apathetic look" and recovered from SMA syndrome with high EOR. In conclusion, intraoperative "apathetic look" shows strong correlation with intraoperative SMA syndrome. Therefore, "apathetic look" may be a valuable indicator of intraoperative SMA syndrome during awake craniotomy.

Relationship between characteristics of glioma treatment and surgical site infections.

PURPOSE: Surgical site infections (SSIs) after neurosurgery are common in daily practice. Although numerous reports have described SSIs in neurosurgery, reports specific to gliomas are limited. This study aimed to investigate the relationship between SSIs and glioma treatment characteristics, such as reoperations, radiation therapy, and chemotherapy. METHODS: We examined 1012 consecutive patients who underwent craniotomy for glioma between November 2013 and March 2022. SSIs were defined as infections requiring reoperation during the observation period, regardless of their location. We retrospectively analyzed SSIs and patient factors. RESULTS: During the observation period, SSIs occurred in 3.1% (31/1012). In the univariate analysis, three or more surgeries (P = 0.007) and radiation therapy (P = 0.03) were associated with SSIs, whereas intraoperative magnetic resonance imaging (MRI) was not significantly associated (P = 0.35). Three or more surgeries and radiation therapy were significantly correlated with each other (P < .0001); therefore, they were analyzed separately in the multivariate analysis. Three or more surgeries were an independent factor triggering SSIs (P = 0.02); in contrast, radiation therapy was not an independent factor for SSIs (P = 0.07). Several SSIs localized in the skin occurred more than 1 year after surgery. CONCLUSIONS: Undergoing three or more surgeries for glioma is an independent risk factor for SSIs. Glioma SSIs can occur long after surgery. These results are considered characteristic of gliomas. We recommend careful long-term observation of patients at a high risk of SSIs.

Video-based neonatal state assessment method for timing of procedures.

BACKGROUND: Procedures should be performed when an infant is most receptive to disruptions in order to reduce the stress on the infant. However, frequent direct observations place a heavy burden on medical staff. There is therefore a need for a method for quantitatively and automatically evaluating the neonatal state. METHODS: Ten infants in our hospital were enrolled in this study. The states of the infants were assessed by medical staff using the Brazelton Neonatal Behavioral Assessment Scale and were recorded on video at the same time. The recorded states were reclassified as activity levels, a new state classification method that includes middle activity, which is the appropriate time for a procedure. Using image analysis, motions of the infant were quantified as two indices: activity and pause time. Activity and pause time were compared for each activity level. The cutoff values of the indices were calculated, and the sensitivity and specificity of the middle activity were calculated. RESULTS: There was a significant difference between all groups of activity level (P < 0.01). The maximum sensitivity and specificity of middle activity were 71.7% and 51.2%, respectively. CONCLUSIONS: The neonatal state of infants can be quantitatively and automatically evaluated using video cameras, and the activity level can be used to determine an appropriate time for procedures in infants. This will reduce the burden on medical staff and lead to less stressful procedures for infants.

Development of a Semiautomatic Dura Mater Suturing Device for Preventing Cerebrospinal Fluid Leakage in Transsphenoidal Surgery.

Dural suturing in transsphenoidal surgery requires well-honed technical skills. We have developed a semiautomatic dural suturing device and confirmed its effectiveness by comparing it with the conventional method. This device significantly shortens the suturing time compared with the conventional method. The dural suturing time in transsphenoidal surgery could be decreased significantly by use of this novel device.

Prediction of lower-grade glioma molecular subtypes using deep learning.

INTRODUCTION: It is useful to know the molecular subtype of lower-grade gliomas (LGG) when deciding on a treatment strategy. This study aims to diagnose this preoperatively. METHODS: A deep learning model was developed to predict the 3-group molecular subtype using multimodal data including magnetic resonance imaging (MRI), positron emission tomography (PET), and computed tomography (CT). The performance was evaluated using leave-one-out cross validation with a dataset containing information from 217 LGG patients. RESULTS: The model performed best when the dataset contained MRI, PET, and CT data. The model could predict the molecular subtype with an accuracy of 96.6% for the training dataset and 68.7% for the test dataset. The model achieved test accuracies of 58.5%, 60.4%, and 59.4% when the dataset contained only MRI, MRI and PET, and MRI and CT data, respectively. The conventional method used to predict mutations in the isocitrate dehydrogenase (IDH) gene and the codeletion of chromosome arms 1p and 19q (1p/19q) sequentially had an overall accuracy of 65.9%. This is 2.8 percent point lower than the proposed method, which predicts the 3-group molecular subtype directly. CONCLUSIONS: A deep learning model was developed to diagnose the molecular subtype preoperatively based on multi-modality data in order to predict the 3-group classification directly. Cross-validation showed that the proposed model had an overall accuracy of 68.7% for the test dataset. This is the first model to double the expected value for a 3-group classification problem, when predicting the LGG molecular subtype.

Laser Speckle Contrast Imaging for Intraoperative Quantitative Assessment of Intestinal Blood Perfusion During Colorectal Surgery: A Prospective Pilot Study.

BACKGROUND: The main limitation of perfusion assessment with indocyanine green fluorescence angiography during colorectal surgery is that the surgeon assesses the quality of perfusion subjectively. The ideal intestinal viability test must be minimally invasive, objective, and reproducible. We evaluated the quantitativity and reproducibility of laser speckle contrast imaging for perfusion assessment during colorectal surgery. METHODS: This was a prospective, nonrandomized, pilot study of 8 consecutive patients who underwent elective left-sided colorectal resection. Laser speckle perfusion images at the site of proximal transection of the bowel were obtained intraoperatively. We tested the hypothesis that laser speckle contrast imaging was able to quantitatively identify areas of diminished intestinal perfusion after devascularization and assessed the reproducibility of this method. RESULTS: All surgical procedures were uneventful and blood flow measurements were successfully made in all patients. None of the patients developed postoperative complications related to the anastomosis and stoma. Data analyses were successfully optimized to perform quantitative regional perfusion assessments in all cases. The bowel tissue blood flows of the anal side region adjacent to the transection line were significantly lower than those of the oral side region adjacent to the transection line after ligation of marginal vessels ( P = .012). Interrater reliability was high (intraclass correlation coefficients = 0.989), and a Bland-Altman plot showed few differences of mean flux data between 2 investigators. CONCLUSION: Laser speckle contrast imaging is feasible for real-time assessment of bowel perfusion with quantitativity and excellent reproducibility during colorectal surgery without administration of any contrast agents.

Evaluation of Surgical Instruments With Radiofrequency Identification Tags in the Operating Room.

BACKGROUND: Surgical instrument retention and instrument breakage compromise surgery quality and lead to medical malpractice. We developed an instrument tracking system that could alert surgeons to instrument retention during surgery and monitor instrument use to reduce the risk of breakage. METHODS: This prospective, experimental clinical trial included 15 patients undergoing inguinal hernia surgery or lumpectomy under general anesthesia at Saiseikai Kurihashi Hospital. Radiofrequency identification (RFID)-tagged surgical instruments were used, and a detection antenna was placed on a mayo stand during the operation. We analyzed the 1-loop detection ratio (OLDR)-that is, the capability of the antenna to detect devices in a single reading-and the total detection rate (TDR)-that is, the data accumulated for the duration of the operation-of the RFID-tagged instruments. RESULTS: Data analysis revealed that the OLDR was 95% accurate, whereas the TDR was 100% accurate. The antenna could not detect the RFID tag when there was interference from electrocautery noise radiation, and 6% of instrument movement was undetected by the antenna; however, the TDR and instrument use were detected at all times. CONCLUSIONS: Surgical instruments can be tracked during surgery, and this tracking can clarify the usage rate of each instrument and serve as a backup method of instrument counting. However, this study was conducted on a small scale, and RFID tags cannot be attached to small surgical instruments used in complex operations such as neurosurgery. Further efforts to develop a tracking system for these instruments are warranted.

Flexible coaxial laser endoscope with arbitrarily selected spots in endoscopic view for photodynamic tumor therapy.

Photodynamic therapy (PDT), which aims to directly destruct tumor cells without any damage to proximal healthy tissue, is widely used in clinics. However, the devices most often employed in hospitals do not readily enable exact control of the irradiated target location. As a result, the laser often irradiates not only the tumor, but also neighboring healthy tissues or blood vessels, with more serious consequences resulting from necrosis. In this paper, we propose a novel flexible coaxial laser endoscope, which localizes the laser illumination only to the selected tumor target, with minimal illumination of the surrounding tissue. In this system, visible light is first transmitted into an imaging-fiber bundle and then reflected by a polarizing beam splitter, which permits initial imaging of the tumor. Once the tumor target is confirmed, an automated stage moves the laser fiber head and focusing lens system to the appropriate position. The laser light is then turned on, passed through a beam splitter, and focused onto the imaging-fiber bundle, ultimately irradiating only the specified target. To evaluate the effectiveness of the device, we first examined the endoscope image quality for MTF50, MTF20, and chromatic aberration, finding high contrast and low aberration. We then measured the laser power at four locations within this system, from the laser fiber head to the endoscope tip. Although some power loss is observed, the power density at the endoscope tip satisfies therapy requirements. Finally, the laser positioning accuracy of our system was measured at 21 positions throughout the endoscope image at distances from 20 mm to 50 mm. We find that the maximum error is less than 1.2 mm, well within clinical requirement. Therefore, we have developed an optimal system for PDT that effectively transmits laser light to the desired target with unprecedented precision, which we anticipate will find wide use in the clinic.

New simple image overlay system using a tablet PC for pinpoint identification of the appropriate site for anastomosis in peripheral arterial reconstruction.

PURPOSE: To evaluate the accuracy and utility of a new image overlay system using a tablet PC for patients undergoing peripheral arterial reconstruction. METHODS: Eleven limbs treated with distal bypass surgery were studied. Three-dimensional images obtained by processing a preoperative contrast-enhanced computed tomography scan were superimposed onto the back-camera images of a tablet PC. We used this system to pinpoint a planned distal anastomotic site preoperatively and to make a precise incision directly above it during surgery. We used a branch artery near the distal anastomotic site as a reference point and the accuracy of the system was validated by comparing its results with the intraoperative findings. The precision of the system was also compared with that of a preoperative ultrasonographic examination. RESULTS: Both the image overlay system and ultrasonography (US) accurately identified the target branch artery in all except one limb. In that limb, which had a very small reference branch artery, preoperative US wrongly identified another branch, whereas the image overlay system located the target branch with an error of 10 mm. CONCLUSIONS: Our image overlay system was easy to use and allowed us to precisely identify a target artery preoperatively. Therefore, this system could be helpful for pinpointing the most accurate incision site during surgery.

Vision-based markerless registration using stereo vision and an augmented reality surgical navigation system: a pilot study.

BACKGROUND: This study evaluated the use of an augmented reality navigation system that provides a markerless registration system using stereo vision in oral and maxillofacial surgery. METHOD: A feasibility study was performed on a subject, wherein a stereo camera was used for tracking and markerless registration. The computed tomography data obtained from the volunteer was used to create an integral videography image and a 3-dimensional rapid prototype model of the jaw. The overlay of the subject's anatomic site and its 3D-IV image were displayed in real space using a 3D-AR display. Extraction of characteristic points and teeth matching were done using parallax images from two stereo cameras for patient-image registration. RESULTS: Accurate registration of the volunteer's anatomy with IV stereoscopic images via image matching was done using the fully automated markerless system, which recognized the incisal edges of the teeth and captured information pertaining to their position with an average target registration error of < 1 mm. These 3D-CT images were then displayed in real space with high accuracy using AR. Even when the viewing position was changed, the 3D images could be observed as if they were floating in real space without using special glasses. CONCLUSION: Teeth were successfully used for registration via 3D image (contour) matching. This system, without using references or fiducial markers, displayed 3D-CT images in real space with high accuracy. The system provided real-time markerless registration and 3D image matching via stereo vision, which, combined with AR, could have significant clinical applications.

3D evaluation model of facial aesthetics based on multi-input 3D convolution neural networks for orthognathic surgery.

BACKGROUND: Quantitative evaluation of facial aesthetics is an important but also time-consuming procedure in orthognathic surgery, while existing 2D beauty-scoring models are mainly used for entertainment with less clinical impact. METHODS: A deep-learning-based 3D evaluation model DeepBeauty3D was designed and trained using 133 patients' CT images. The customised image preprocessing module extracted the skeleton, soft tissue, and personal physical information from raw DICOM data, and the predicting network module employed 3-input-2-output convolution neural networks (CNN) to receive the aforementioned data and output aesthetic scores automatically. RESULTS: Experiment results showed that this model predicted the skeleton and soft tissue score with 0.231 ± 0.218 (4.62%) and 0.100 ± 0.344 (2.00%) accuracy in 11.203 ± 2.824 s from raw CT images. CONCLUSION: This study provided an end-to-end solution using real clinical data based on 3D CNN to quantitatively evaluate facial aesthetics by considering three anatomical factors simultaneously, showing promising potential in reducing workload and bridging the surgeon-patient aesthetics perspective gap.

3-D endoscope using a single CCD camera and pneumatic vibration mechanism.

BACKGROUND: During endoscopic surgical procedures, surgeons must manipulate an endoscope inside the body cavity to observe a surgical area while estimating the distance between that area and the surgical instruments by reference to a monitor on which the movement and size of the surgical instruments are displayed in 2-D endoscopic images. Therefore, there is a risk of the endoscope or instruments physically damaging body tissues. To overcome this problem, we developed a Ø5-mm, 3-D endoscope using a single 1/10-inch CCD camera and pneumatic vibration mechanism. METHODS: The 3-D endoscope proposed in this paper consists of an outer and inner sleeve, a 1/10-inch CCD camera attached to its distal end, and a pneumatic vibration mechanism attached to its proximal end. This endoscope can acquire left and right endoscopic images for stereovision in synchrony with the periodical motion generated by the vibration mechanism. We measured the displacement at the proximal and distal end of the 3-D endoscope simultaneously, and evaluated the feasibility of its use in vivo. RESULTS: The displacement at the distal end of the endoscope to which the CCD camera is attached was approximately ±0.25 mm. The timing when the displacement of the CCD camera was at maximal amplitude coincided with the timing when the displacement of its proximal end was at maximal amplitude. In the in vivo experiment, this 3-D endoscope can provide clear 3-D images of the surgical area. CONCLUSIONS: The developed 3-D endoscope that uses a single CCD camera and pneumatic vibration mechanism can successfully visualize internal organs inside the body even though the CCD camera is moved by the vibration. Therefore, the risk of damage to fragile body tissues can be significantly decreased.

Precise Brain-shift Prediction by New Combination of W-Net Deep Learning for Neurosurgical Navigation.

Brain tissue deformation during surgery significantly reduces the accuracy of image-guided neurosurgeries. We generated updated magnetic resonance images (uMR) in this study to compensate for brain shifts after dural opening using a convolutional neural network (CNN). This study included 248 consecutive patients who underwent craniotomy for initial intra-axial brain tumor removal and correspondingly underwent preoperative MR (pMR) and intraoperative MR (iMR) imaging. Deep learning using CNN to compensate for brain shift was performed using the pMR as input data, and iMR obtained after dural opening as the ground truth. For the tumor center (TC) and the maximum shift position (MSP), statistical analysis using the Wilcoxon signed-rank test was performed between the target registration error (TRE) for the pMR and iMR (i.e., the actual amount of brain shift) and the TRE for the uMR and iMR (i.e., residual error after compensation). The TRE at the TC decreased from 4.14 ± 2.31 mm to 2.31 ± 1.15 mm, and the TRE at the MSP decreased from 9.61 ± 3.16 mm to 3.71 ± 1.98 mm. The Wilcoxon signed-rank test of the pMR TRE and uMR TRE yielded a p-value less than 0.0001 for both the TC and MSP. Using a CNN model, we designed and implemented a new system that compensated for brain shifts after dural opening. Learning pMR and iMR with a CNN demonstrated the possibility of correcting the brain shift after dural opening.

Construction of brain area risk map for decision making using surgical navigation and motor evoked potential monitoring information.

PURPOSE: Surgical devices or systems typically operate in a stand-alone manner, making it difficult to perform integration analysis of both intraoperative anatomical and functional information. To address this issue, the intraoperative information integration system OPeLiNK® was developed. The objective of this study is to generate information for decision making using surgical navigation and intraoperative monitoring information accumulated in the OPeLiNK® database and to analyze its utility. METHODS: We accumulated intraoperative information from 27 brain tumor patients who underwent resection surgery. First, the risk rank for postoperative paralysis was set according to the attenuation rate and amplitude width of the motor evoked potential (MEP). Then, the MEP and navigation log data were combined and plotted on an intraoperative magnetic resonance image of the individual brain. Finally, statistical parametric mapping (SPM) transformation was performed to generate a standard brain risk map of postoperative paralysis. Additionally, we determined the anatomical high-risk areas using atlases and analyzed the relationship with each set risk rank. RESULTS: The average distance between the navigation log corresponding to each MEP risk rank and the anatomical high-risk area differed significantly between the with postoperatively paralyzed and without postoperatively paralyzed groups, except for "safe." Furthermore, no excessive deformation was observed resulting from SPM conversion to create the standard brain risk map. There were cases in which no postoperative paralysis occurred even when MEP decreased intraoperatively, and vice versa. CONCLUSION: The time synchronization reliability of the study data is very high. Therefore, our created risk map can be reported as being functional at indicating the risk areas. Our results suggest that the statistical risks of postoperative complications can be presented for each area where brain surgery is to be performed. In the future, it will be possible to provide surgical navigation with intraoperative support that reflects the risk maps created.

Surgery assistance system for continuous resection of brain tumors-proposal of continuous tumor resection forceps, tumor cell separation, dehydration, and isolation mechanism.

The tumor resection ratio must be improved due the increased possibility of recurrence or malignancy. The purpose of this study was to develop a system that includes forceps with a continuous suction function and flow cytometry to diagnose the malignancy of the tumor for safe, accurate, and effective surgery. A newly developed continuous tumor resection forceps consists of a triple pipe structure, which enables continuous suction of the tumor by integrating the reflux water and suction system. The forceps includes tip opening/closure detection switch to control the adsorption and suction strength when tip is opened and closed. To perform accurate tumor diagnosis using flow cytometry, a filtering mechanism was developed for dehydrating reflux water from continuous suction forceps. In addition, a cell isolation mechanism comprising a roller pump and shear force loading mechanism was also newly developed. By using a triple pipe structure, a significantly larger tumor collection ratio was observed compared to the previous double-pipe structure. By performing suction pressure control with the opening/closure detection switch, inaccurate suction can be prevented. By widening the filter area of dehydration mechanism, it was possible to improve the reflux water dehydration ratio. The most appropriate filter area was 85 mm2. By using a newly developed cell isolation mechanism, the processing time can be reduced to less than 1/10 of the original time, keeping the same cell isolation ratio, when compared to the existing pipetting method. Neurosurgery assistance system with continuous tumor resection forceps and a cell separation, dehydration and isolation mechanism was developed. An effective and safe tumor resection, accurate and fast diagnosis of malignancy can be achieved by using the current system.

Magnetic resonance imaging-compatible tactile sensing device based on a piezoelectric array.

Minimally invasive surgery is a widely used medical technique, one of the drawbacks of which is the loss of direct sense of touch during the operation. Palpation is the use of fingertips to explore and make fast assessments of tissue morphology. Although technologies are developed to equip minimally invasive surgery tools with haptic feedback capabilities, the majority focus on tissue stiffness profiling and tool-tissue interaction force measurement. For greatly increased diagnostic capability, a magnetic resonance imaging-compatible tactile sensor design is proposed, which allows minimally invasive surgery to be performed under image guidance, combining the strong capability of magnetic resonance imaging soft tissue and intuitive palpation. The sensing unit is based on a piezoelectric sensor methodology, which conforms to the stringent mechanical and electrical design requirements imposed by the magnetic resonance environment The sensor mechanical design and the device integration to a 0.2 Tesla open magnetic resonance imaging scanner are described, together with the device's magnetic resonance compatibility testing. Its design limitations and potential future improvements are also discussed. A tactile sensing unit based on a piezoelectric sensor principle is proposed, which is designed for magnetic resonance imaging guided interventions.

The actual status of drug prices and adjustment factors for drug price calculation: an analysis of ultra-orphan drug development in Japan.

BACKGROUND: Extremely high prices facilitate drug development for ultra-rare diseases (ultra-orphan drugs). However, various problems arise in terms of healthcare financing and fairness, and the status of ultra-orphan drug pricing remains ambiguous. In this study, we investigated ultra-orphan drug prices in Japan relative to that of other drugs. We examined the relationship between annual expected drug prices and expected sales, and the expected number of patients, for 393 drugs containing new active ingredients for therapeutic use that were listed on the National Health Insurance drug price list in Japan between April 16, 2010 and August 26, 2020. In addition, we compared prices, the drug price calculation method, and price calculation adjustment factors for ultra-orphan and other drugs. RESULTS: Drug prices tended to increase as the expected number of patients to whom the drug was administered decreased; however, this trend diminished when the expected number of patients was less than 1000. On the other hand, the expected sales tended to decrease as the number of expected patients decreased, and this tendency was reinforced when the expected number of patients was less than 1000. The cost accounting method tended to be used for the price calculation of ultra-orphan drugs, but there were no price differences based on the drug price calculation method. Regarding the price calculation adjustment factors, the premium for usefulness tended to be higher for ultra-orphan drugs. The premium for marketability was higher for non-orphan drugs but did not differ from that for orphan drugs, except for ultra-orphan drugs. CONCLUSIONS: The status of drug prices and expected sales differed beyond a threshold of 1000 expected patients, indicating that recovering the development cost for ultra-orphan drugs is difficult. In addition, the higher premium for usefulness for ultra-orphan drugs reflects the largely unmet need of the associated diseases. Scarcity among orphan drugs is not considered for marketability, highlighting the need for a new framework to promote the development of ultra-orphan drugs.

A trial to visualize perforators images from CTA with a tablet device: experience of operating on minipigs.

A reliable method for precise perforator mapping can be extremely valuable in perforator flap surgery. In this study, we attempted to map perforator location using 3-dimensional computed tomography angiography (CTA), a newly developed application, and a tablet device. Preliminary examinations to test the device were conducted in mini-pigs. We used 5 female mini-pigs. Preoperative imaging of the vasculature was undertaken with CTA in the prone position, following Iopamidol (200 ml) injection via the internal jugular vein. Prior to the examination, we placed round markers on the backs of the mini-pigs. To assess accuracy, we compared the perforator positions acquired with an optical position measurement device with the perforator positions acquired with the tablet device. Furthermore, we compared the perforator positions with the tablet navigation device, which we measured directly. We measured 12 perforators with the optical position measurement device. The mean difference was 10 mm (minimum, 2 mm; maximum, 20 mm). We measured these perforators with the tablet navigation device. The mean difference was 5.4 mm (minimum, 0 mm; maximum, 20 mm). The perforator flaps were elevated safely. The perforator flaps could be elevated safely using our device, as the mean difference was only 10 mm, which is acceptable for navigating perforator flap operations. Pig backs are triangular in shape; therefore, we were unable to place markers on the contralateral side. Thus, for clinical applications of the device, we should determine the ideal marker locations.

Machine-learning-based approach for predicting postoperative skeletal changes for orthognathic surgical planning.

BACKGROUND: Manually surgical planning becomes an increasing workload of surgeons because of the fast-growing patient population. This study introduced a machine-learning-based approach to assist surgical planning in orthognathic surgery. METHODS: Both preoperative and one-year-later postoperative computerised tomography images of 56 patients were collected. A 12-layers cascaded deep neural network structure with two successive models was proposed to yield an end-to-end solution, where the first model extracts landmarks from 2D patches of 3D volume and the second model predicts postoperative skeletal changes. RESULTS: The experimental results showed that the model obtained a prediction accuracy of 5.4 mm at the landmark level in 42.9 s. It also represented 74.4% of 3D regions at volume level when compared with the ground truth of human surgeons. CONCLUSIONS: This study demonstrated the feasibility of predicting postoperative skeletal changes for orthognathic surgical planning by using machine learning, showing great potential for reducing the workload of surgeons.