Segmental Instrumentation Research Articles

Fast instruments and tissues segmentation of micro-neurosurgical scene using high correlative non-local network

Surgical scene segmentation provides critical information for guidance in micro-neurosurgery. Segmentation of instruments and critical tissues contributes further to robot assisted surgery and surgical evaluation. However, due to the lack of relevant scene segmentation dataset, scale variation and local similarity, micro-neurosurgical segmentation faces many challenges. To address these issues, a high correlative non-local network (HCNNet), is proposed to aggregate multi-scale feature by optimized non-local mechanism. HCNNet adopts two-branch design to generate features of different scale efficiently, while the two branches share common weights in shallow layers. Several short-term dense concatenate (STDC) modules are combined as the backbone to capture both semantic and spatial information. Besides, a high correlative non-local module (HCNM) is designed to guide the upsampling process of the high-level feature by modeling global context generated from the low-level feature. It filters out confused pixels of different classes in the non-local correlation map. Meanwhile, a large segmentation dataset named NeuroSeg is constructed, which contains 15 types of instruments and 3 types of tissues that appear in meningioma resection surgery. The proposed HCNNet achieves the state-of-the-art performance on NeuroSeg, it reaches an inference speed of 54.85 FPS with the highest accuracy of 59.62% mIoU, 74.7% Dice, 70.55% mAcc and 87.12% aAcc.

Adding sacral anchors through an S1 alar screw and multirod construct as a strategy for lumbosacral junction augmentation: an in vitro comparison to S1 pedicle screws alone with sacroiliac fixation.

Achieving solid fusion of the lumbosacral junction continues to be a challenge in long-segment instrumentation to the sacrum. The purpose of this study was to test the condition of adding sacral anchors through an S1 alar screw (S1AS) and multirod construct relative to using S1 pedicle screws (S1PSs) alone with sacroiliac fixation in lumbosacral junction augmentation. Seven fresh-frozen human lumbar-pelvic spine cadaveric specimens were tested under nondestructive moments (7.5 Nm). The ranges of motion (ROMs) in extension, flexion, left and right lateral bending (LB), and axial rotation (AR) of instrumented segments (L3-S1); the lumbosacral region (L5-S1); and the adjacent segment (L2-3) were measured, and the axial construct stiffness (ACS) was recorded. The testing conditions were 1) intact; 2) bilateral pedicle screw (BPS) fixation at L3-S1 (S1PS alone); 3) BPS and unilateral S2 alar iliac screw (U-S2AIS) fixation; 4) BPS and unilateral S1AS (U-S1AS) fixation; 5) BPS and bilateral S2AIS (B-S2AIS) fixation; and 6) BPS and bilateral S1AS (B-S1AS) fixation. Accessory rods were used in testing conditions 3-6. In all directions, the ROMs of L5-S1 and L3-S1 were significantly reduced in B-S1AS and B-S2AIS conditions, compared with intact and S1PS alone. There was no significant difference in reduction of the ROMs of L5-S1 between B-S1ASs and B-S2AISs. Greater decreased ROMs of L3-S1 in extension and AR were detected with B-S2AISs than with B-S1ASs. Both B-S1ASs and B-S2AISs significantly increased the ACS compared with S1PSs alone. The ACS of B-S2AISs was significantly greater than that of B-S1ASs, but with greater increased ROMs of L2-3 in extension. Adding sacral anchors through S1ASs and a multirod construct was as effective as sacropelvic fixation in lumbosacral junction augmentation. The ACS was less than the sacropelvic fixation but with lower ROMs of the adjacent segment. The biomechanical effects of using S1ASs in the control of long-instrumented segments were moderate (better than S1PSs alone but worse than sacropelvic fixation). This strategy is appropriate for patients requiring advanced lumbosacral fixation, and the risk of sacroiliac joint violation can be avoided.

TMF-Net: A Transformer-Based Multiscale Fusion Network for Surgical Instrument Segmentation From Endoscopic Images

Automatic surgical instrument segmentation is a necessary step for the steady operation of surgical robots, and the segmentation accuracy directly affects the surgical effect. Nevertheless, accurate surgical instrument segmentation from endoscopic images remains a challenging task due to the complex environment and instrument motion during surgery. Based on the encoder–decoder structure, a transformer-based multiscale fusion network, named TMF-Net, is proposed to address the difficulties in the area of surgical instrument segmentation. To realize the effective feature representation based on the pretrained ResNet34 and transformer and to strengthen both advantages of different encoder units, a dual-encoder unit is proposed to simultaneously learn the semantic relationship between adjacent pixels and distant pixels and comprehensively capture the global context information. Meanwhile, to retain more contextual information, a trapezoid atrous spatial pyramid pooling (trapezoid ASPP) block is proposed for feature enhancement of local features with different receptive fields to enrich feature information. Furthermore, in addition to multiscale surgical instruments in endoscopic images, a multiscale attention fusion (MAF) block is proposed to fuse multiscale feature maps to make the segmentation network direct more attention to the efficient channels so that it can improve the segmentation accuracy. Two typical datasets are used for performance analysis and verification, including Kvasir-Instrument and Endovis2017. Experimental results indicate that the proposed TMF-Net could effectively improve segmentation accuracy on surgical instruments, and it could also yield a competitive segmentation result in comparison with advanced detection methods.

A dual-encoder feature attention network for surgical instrument segmentation

A dual-encoder feature attention network for surgical instrument segmentation

Multistage surgical treatment of early-onset scoliosis in patients with Ehlers-Danlos syndrome: A series of observations

BACKGROUND: EhlersDanlos syndrome (EDS) is a group of hereditary pathological conditions caused by various disorders of collagen biosynthesis. The study analyzed the results of multistage surgical treatment of early scoliosis in patients with severe spinal deformities due to EDS. No similar observations have been found in the literature.
 CLINICAL CASES: Four patients with a verified diagnosis of EDS and progressive spinal deformities were subjected to multistage surgical treatment using the VEPTRII instrumentation, which included periodic distractions and final spinal fusion with segmental instrumentation. Stage-by-stage surgical treatment was initiated from the age of 3 to 6 years. In 3 of 4 cases, the kyphotic component prevailed over the scoliotic one (86140 vs. 21110). The number of staged distractions ranged from 6 to 10. The age of the final stage (correction and dorsal fusion) was 914 years (surgery was performed in three of four cases). The primary correction was 3056, the loss of correction before the final stage was 1435, and the correction during the final stage was 2240. A significant correction of the frontal and sagittal imbalances of the spine was noted. Blood loss during the final fusion was 540750 mL, and the operation time was 310350 min. Ten complications occurred, of which 9 were associated with implants and disappeared during staged distractions. No neurological and vascular complications occurred.
 DISCUSSION: Scoliosis occurring in the first decade of life in patients with EDS is characterized by early-onset, rapid progression, and tendency to form a significant kyphotic component of spinal deformity.
 CONCLUSIONS: Multistage treatment of early scoliosis in patients with EDS using VEPTRII tools allows for obtaining quite satisfactory results and has not severe complications. The final fusion gives a significant corrective effect; however, new research and accumulation of data are needed to optimize the treatment process.

Perioperative complications after posterior spinal fusion versus minimally invasive fusionless surgery in neuromuscular scoliosis: a comparative study.

Early-onset scoliosis is a common deformity in neuromuscular disease. When conservative treatment becomes ineffective, several surgical options can be proposed. The most common technique is posterior spinal fusion (PSF) consisting of performing a multiple segmental instrumentation with pedicular screws on the full spine associated with decortication and bone graft. Minimally invasive fusionless surgery (MIFS) is an alternative to correct and fix definitively the spine without graft. The objective of this study was to compare early surgical inpatient period between PSF and MIFS in neuromuscular scoliosis. 140 NMS operated by PSF or MIFS between 2012 and 2017 was retrospectively reviewed. The following data were compared between groups: general characteristics (age, sex, etiology), preoperative preparation (halo traction, noninvasive ventilation or tracheostomy), Cobb angle and pelvic obliquity correction, use of drugs (vasopressor and/or inotropes, expansion fluids, transfusion and volumes), postoperative complications, and need of noninvasive ventilation. 75 patients were managed by PSF with a mean age of 14.3 ± 2.3y and 65 by MIFS with a mean age of 11.8 ± 3y. Average pelvic obliquity and major curve correction were similar postoperatively. Intraoperative blood transfusion was significantly more common in PSF group (OR, 14; 95% CI [6.3-33.0]). Vasopressors were used non-significantly more often in the PSF group and expansion fluids similar in the two groups. PSF group had more overall complications (OR, 4.6; 95% CI [2.3-9.8]), more infections (OR, 3.6; 95% CI [1.5-9.3]) and more hemodynamic complications (OR, 4.1; 95% CI [1.4-15.1]). Average intubation duration was 5days in the PSF and 4days in MIFS (p = 0.05). In this series of neuromuscular patients, the complication rate was reduced in MIFS comparatively to PSF, with lower blood transfusion and less infections.

SSIS-Seg: Simulation-Supervised Image Synthesis for Surgical Instrument Segmentation.

Surgical instrument segmentation can be used in a range of computer assisted interventions and automation in surgical robotics. While deep learning architectures have rapidly advanced the robustness and performance of segmentation models, most are still reliant on supervision and large quantities of labelled data. In this paper, we present a novel method for surgical image generation that can fuse robotic instrument simulation and recent domain adaptation techniques to synthesize artificial surgical images to train surgical instrument segmentation models. We integrate attention modules into well established image generation pipelines and propose a novel cost function to support supervision from simulation frames in model training. We provide an extensive evaluation of our method in terms of segmentation performance along with a validation study on image quality using evaluation metrics. Additionally, we release a novel segmentation dataset from real surgeries that will be shared for research purposes. Both binary and semantic segmentation have been considered, and we show the capability of our synthetic images to train segmentation models compared with the latest methods from the literature.

An attention-guided network for surgical instrument segmentation from endoscopic images

Accurate surgical instrument segmentation can provide the precise location and pose information to the surgeons, assisting the surgeon to accurately judge the follow-up operation during the robot-assisted surgery procedures. Due to strong context extraction ability, there have been significant advances in research of automatic surgical instrument segmentation, especially U-Net and its variant networks. However, there are still some problems to affect segmentation accuracy, like insufficient processing of local features, class imbalance issue, etc. To deal with these problems, with the typical encoder–decoder structure, an effective surgical instrument segmentation network is proposed for providing an end-to-end detection scheme. Specifically, aimed at the problem of insufficient processing of local features, the residual path is introduced for the full feature extraction to strengthen the backward propagation of low-level features. Further, to achieve feature enhancement of local feature maps, a non-local attention block is introduced to insert into the bottleneck layer to acquire global contexts. Besides, to highlight the pixel areas of the surgical instruments, a dual-attention module (DAM) is introduced to make full use of the high-level features extracted from decoder unit and the low-level features delivered by the encoder unit to acquire the attention features and suppress the irrelevant features. To prove the effectiveness and superiority of the proposed segmentation model, experiments are conducted on two public surgical instrument segmentation data sets, including Kvasir-instrument set and Endovis2017 set, which could acquire a 95.77% Dice score and 92.13% mIOU value on Kvasir-instrument set, and simultaneously reach 95.60% Dice score and 92.74% mIOU value on Endovis2017 set respectively. Experimental results show that the proposed segmentation model realizes a superior performance on surgical instruments in comparison to other advanced models, which could provide a good reference for further development of intelligent surgical robots. The source code is provided at https://github.com/lyangucas92/Surg_Net.

MRI changes of adjacent segments after transforaminal lumbar interbody fusion (TLIF) and foraminal endoscopy: A case-control study.

Background:Intervertebral foramen endoscopy has developed rapidly, but compared with transforaminal lumbar interbody fusion (TLIF), the progress of degeneration is unknown. We aim to compare the changes of intervertebral disc and intervertebral foramen in adjacent segments after TLIF and endoscopic discectomy for patients with lumbar disc herniation (LDH).Methods:From 2014 to 2017, 87 patients who were diagnosed with single-level LDH and received surgery of TLIF (group T, n = 43) or endoscopic discectomy (group F, n = 44) were retrospectively analyzed. X-ray, MRI, CT and clinical symptoms were recorded before operation and at the last follow-up (FU). The neurological function was originally evaluated by the Japanese Orthopaedic Association (JOA) scores. Radiological evaluation included the height of intervertebral space (HIS), intervertebral foramen height (FH), intervertebral foramen area (FA), lumbar lordosis (CA) and intervertebral disc degeneration Pfirrmann scores.Results:There was no significant difference in baseline characteristics, JOA improvement rate, reoperation rate and complications between the two groups. The age, average blood loss, average hospital stays and average operation time in group F were lower than those in group T. During the last FU, HIS, CA and FA decreased in both groups, and the changes in group T were more significant than those in group F (P < .05). There was no significant difference in FH changes between the two groups (P > .05).Conclusion:Both TLIF and endoscopic surgery can achieve good results in the treatment of LDH, but the risk of lumbar disc height loss and intervertebral foramina reduction in the adjacent segment after endoscopic surgery is lower.

Real-time instance segmentation of surgical instruments using attention and multi-scale feature fusion.

Precise instrument segmentation aids surgeons to navigate the body more easily and increases patient safety. While accurate tracking of surgical instruments in real-time plays a crucial role in minimally invasive computer-assisted surgeries, it is a challenging task to achieve, mainly due to: (1) a complex surgical environment, and (2) model design trade-off in terms of both optimal accuracy and speed. Deep learning gives us the opportunity to learn complex environment from large surgery scene environments and placements of these instruments in real world scenarios. The Robust Medical Instrument Segmentation 2019 challenge (ROBUST-MIS) provides more than 10,000 frames with surgical tools in different clinical settings. In this paper, we propose a light-weight single stage instance segmentation model complemented with a convolutional block attention module for achieving both faster and accurate inference. We further improve accuracy through data augmentation and optimal anchor localization strategies. To our knowledge, this is the first work that explicitly focuses on both real-time performance and improved accuracy. Our approach out-performed top team performances in the most recent edition of ROBUST-MIS challenge with over 44% improvement on area-based multi-instance dice metric MI_DSC and 39% on distance-based multi-instance normalized surface dice MI_NSD. We also demonstrate real-time performance (>60 frames-per-second) with different but competitive variants of our final approach.

Dejeneratif Lomber Hastalıklarda Uygulanan Spinal Enstrümantasyon Sonrası Gelişen Komşu Segment Dejenerasyonu: İnsidans ve Risk Faktörleri

Posterior spinal enstrümantasyon, dejeneratif omurga patolojilerinin tedavisi için oldukça yaygın uygulanan cerrahi bir prosedürdür. Spinal enstrümantasyon sonrası komşu segment dejenerasyonu (KSD) gelişimi önemli bir problem olarak karşımıza çıkmaktadır. Bu çalışmanın amacı KSD gelişimi için risk faktörlerinin değerlendirilmesidir. Bu çalışmada posterior segmental enstrümantasyon uygulanan ve cerrahi işlemi üzerinden 4 yıldan fazla geçen 126 hasta retrospektif olarak incelenmiştir. Bu hastalar KSD ve n-KSD olarak iki gruba ayrılmıştır. İki grubun hasta karakteristikleri, preoperatif ve postoperatif radyolojik parametreleri ve cerrahi farklılıkları karşılaştırılarak KSD gelişimi için prediktif faktörler ortaya konulmaya çalışıldı. Çalışmaya dahil edilen hastaların on beşinde (%11.9) KSD geliştiği görüldü. Her iki grup arasında cinsiyet, diyabetes mellitus, sigara kullanımı ve osteoporoz açısından anlamlı farklılık yok idi (p&gt;0.05). Lojistik regresyon analizine göre preoperatif yüksek vücut kitle indeksi, preoperatif komşu segmentteki faset dejenerasyonu varlığı, postoperatif lomber lordozda azalma ve 4 seviyeden daha fazla posterior enstürman uygulanması KSD gelişimi için bağımsız risk faktörleridir. Spinal enstrümantasyon cerrahisi öncesi yukarıda bahsedilen risk faktörlerinin bilinmesi, cerrahi açıdan daha uygun sonuçlar alınabilmesi için önlem almaya imkan tanımaktadır. Hastaların ve işlemi uygulayacak cerrahların modifiye edilebilecek risk faktörleri açısından gerekli önlemleri almaları uzun dönem komplikasyonları azaltabileceğini düşünmekteyiz.

59. Long-term survival and clinical outcomes in patients managed with emergent surgery for a spinal epidural abscess with associated sepsis and neurologic decline

<h3>BACKGROUND CONTEXT</h3> Patients with a spinal epidural abscess (SEA) and associated sepsis and neurologic deterioration require immediate evaluation and prompt surgical intervention. The goals of surgery are life preservation, eradication of the infectious nidus, and restoration of neurologic function. The focus of treatment in these critically ill patients is to achieve these immediate goals. There has been limited attention given to the long-term status of patients who have survived this life threatening condition. <h3>PURPOSE</h3> To identify the survivability and long-term outcome of patients who underwent emergent surgical management for a SEA with associated sepsis and neurologic decline. <h3>STUDY DESIGN/SETTING</h3> A retrospective review of all consecutive patients with a SEA, sepsis and neurologic decline treated at a single, tertiary referral medical center from 2000-2010. <h3>PATIENT SAMPLE</h3> Fifty-four patients. <h3>OUTCOME MEASURES</h3> Survival/mortality, neurologic improvement (ASIA Grade), need for further spine surgery. <h3>METHODS</h3> All patients were clinically septic and had documented neurologic deterioration directly related to a spinal epidural abscess. Surgery was performed emergently from the time of notification of the spine service. Surgical approach was determined by the location of the abscess in relation to the thecal sac; 8 occurred in the cervical spine (all anterior), 20 in the thoracic spine (7 anterior, 13 posterior) and 26 in the lumbar spine (9 anterior, 17 posterior). All cervical SEA underwent an anterior decompression and fusion with structural grafting and plating, while anterior SEA within the thoracic or lumbar spines underwent anterior decompression and fusion with structural grafting followed by a staged, posterior instrumented fusion. Posterior SEA in the thoracic or lumbar spine underwent decompression and abscess evacuation only. Preoperative neurologic status was ASIA B in 20 patients, C in 20, and D in 10. The mean time of neurologic deterioration documented in the medical record was 24 hours (range: 6 hours-5 days). Stepwise multiple regression analysis was performed to identify risk factors for survival and neurologic improvement. Long-term survival was assessed using the institution's electronic medical record and subsequently confirmed by direct patient contact. <h3>RESULTS</h3> All patients survived the surgery, although 6 (11%) died within one month. Significant risk factors for early mortality included age >70 (p=.01), hospitalization >5 days prior to surgery (p=.02), ASIA D (p=.02), diabetes (p=.02), end stage renal disease (p=.02), and MRSA sepsis (p=.02). Of the surviving patients, the mean neurologic improvement was 2.4 ASIA grades. Of surviving patients, 92% (44/48) regained the ability to ambulate. Significant factors for neurologic improvement included age <70 (p=.01), documented neurologic decline <24 hours (p=.01), non-diabetic (p=.02), a cervical abscess (p=.01), a thoracic or lumbar abscess not requiring fusion (p=.02), and preoperative ASIA D (p=.01). At the 10-year follow-up, 88% (42/48) of the patients who survived their perioperative course were still alive. The six non-surviving patients had died of causes unrelated to their spine infections. At the 10-year follow-up, all 42 patients were noted to have maintained their level of neurologic recovery, with none having evidence of recurrent neurologic deficits or loss of ambulatory ability. Ten percent (4/42) of the long-term surviving patients had undergone additional spine surgery, 2 for removal of painful segmental instrumentation and 2 for extension of fusions for symptomatic adjacent level disease. None of the long-term survivors required additional surgery for a recurrent spine infection. Advanced imaging (MRI or CT scan) was available for 30 of the surviving patients. Aside from the 2 patients requiring surgery for adjacent level stenosis, there were no instances of residual infectious spinal column involvement, segmental instability or sagittal plane deformity in any patient. <h3>CONCLUSIONS</h3> Spinal epidural abscesses in patients with sepsis and neurologic deterioration can be effectively managed when diagnosed promptly and treated with emergent surgery. In this series, the majority of patients survived, had neurologic recovery and regained independent ambulation. However, there was a notable early mortality rate (11%), with older, diabetic patients with prolonged neurologic deficits most susceptible. Long-term survival occurred in 88% of patients, with evidence of sustained neurologic improvement, permanent eradication of the spine infection, and relatively uncommon need for additional surgery. <h3>FDA DEVICE/DRUG STATUS</h3> This abstract does not discuss or include any applicable devices or drugs.

Angled Ultrasonic Bone Curette-Assisted Circumferential Decompression for Thoracic Myelopathy Caused by Severely Anterior Ossification.

ObjectiveThoracic myelopathy caused by severe anterior ossification is often progressive and fails to respond to conservative treatment. Removal of the compressing ossification is the most effective method but is hard to operate. In this study, we describe a novel one‐stage posterior circumferential decompressive procedure assisted by an angled ultrasonic bone curette (UBC) for thoracic myelopathy caused by severe anterior ossification and evaluate its safety and efficacy.MethodsThe current study enrolled 15 consecutive patients (five men and 10 women) with thoracic myelopathy caused by severely anterior ossification between January 2017 and December 2019. All patients underwent posterior circumferential decompression assisted by angled UBC and segmental instrumentation with interbody fusion. At the time of surgery, the average age was 58.6 ± 6.3 years (47–70 years). Before and after surgery, the patient data, clinical manifestation, operative levels, blood loss, operative time, perioperative complications, Japanese Orthopaedic Association (JOA) score were recorded and analyzed retrospectively.ResultsAll patients had successful one‐stage posterior circumferential decompression to remove anterior ossifications directly. There were 12 cases of OPLL, two cases of a calcified giant herniated disc, and one case of osteophyte. The average operation time was 153.4 ± 53.4 min (77–242 min), with a mean blood loss of 463.5 ± 155.8 mL (240–780 mL). The average length of stay in the hospital was 14.3 ± 4.7 days (9–25 days) and the mean follow‐up duration was 20.8 ± 8.8 months (12–39 months). Almost all patients had subjective improvement in motor power and gait. The average preoperative JOA score was 4.5 ± 1.6, which improved to 9.0 ± 1.8 at the final follow‐up. Postoperative differences in the overall JOA scores showed significant improvement (F = 105.446, p < 0.01). The overall recovery rate at the final examination scored 70.9% ± 25.0%. According to Hirabayashi's classification, eight cases were rated as excellent, four as good, two as fair, and one as unchanged. No patient was graded as deteriorated. Two patients (13.3%) experienced intraoperative cerebrospinal fluid leakage, while two cases (13.3%) experienced unilateral intercostal neuralgia, and only one (6.7%) encountered acute neurological deterioration. All these patients were treated conservatively and their neurological function improved significantly. At the follow‐up, there was no evidence of neurological deterioration.ConclusionCircumferential decompression assisted by angled UBC can preserve more posterior elements of the involved levels, maintaining an intact pleura and reducing the operation time and blood loss for thoracic myelopathy caused by severe anterior ossification. It is a safe, effective, and technically feasible method to provide surgeons with a new option for thoracic spinal circumferential decompression.

DRR-Net: A Dense-Connected Residual Recurrent Convolutional Network for Surgical Instrument Segmentation From Endoscopic Images

The precise segmentation of surgical instruments is the key link for the stable and reasonable operation of surgical robots. However, accurate surgical instrument segmentation is still a challenging task due to the complex surgical environment in endoscopic images, low contrast between surgical instruments and tissues, and the diversity of surgical instruments and their morphological variability. In recent years, deep learning has been widely applied into medical image segmentation and achieved a certain achievements, especially U-Net and its variants. However, existing surgical instrument segmentation networks still suffer from some shortcomings, such as insufficient processing of local feature maps, lack of temporal modeling information, etc. To address the above issues, in order to effectively improve the segmentation accuracy of surgical instruments, based on an encoder-decoder network structure, a novel dense residual recurrent convolutional network, called DRR-Net, is proposed in this paper for automatic and accurate surgical instrument segmentation from endoscopic images. Faced with lack of temporal modeling information, inspired by the recurrent neural networks (RNNs), an attention dense-connected recurrent convolutional block (ADRCB) is proposed to optimize the backbone network to obtain temporal information and learn the correspond pixel relationship between frames. To address the insufficient processing issue of local feature maps, to replace the simple skip connections, a residual path is proposed to enhance the context feature representation. Meanwhile, it could also reduce the effect of semantic gap issue. Further, to improve the segmentation accuracy on segmented objects with different sizes, a context fusion block (CFB) is proposed to embed into the bottleneck layer to extract multi-scale attention context features. Multiple public data sets on surgical instrument segmentation are adopted for model evaluation and comparison, including kvasir-instrument set and UW-Sinus-Surgery-C/L set. Experimental results demonstrate that proposed DRR-Net network could achieve an excellent performance compared with other advanced segmentation models.

Simulation Training in Pedicle Preparation and Screw Insertion in Pediatric Spinal Deformity

Pedicle screw segmental instrumentation has become the gold standard for spinal deformity correction for both the pediatric and adult populations. The thought of placing screws through tiny boney channels in the spinal column can be a daunting task for trainees. The goal in spine surgery instrumentation is similar to other orthopaedic subspecialities—safe passage of screws through the bone. Yet, the potential consequences of pedicle breaches and malpositioned instruments/screws are what give pedicle preparation and screw insertion an appropriately heightened sense of care and apprehension. Important anatomical structures lie around the pedicles of the spine, with the spinal cord, nerve roots, epidural veins, lungs, kidneys, and major vessels being just some of the structures that lurk outside of the spine’s osseous anatomy. In 2007, we implemented power pedicle screw preparation and screw insertion to improve safety for both the surgeon and patient. A recent multicenter study of early adopters demonstrated the safety of this technique in 22,209 screws across six institutions. As we have taught this technique to fellow surgeons throughout the world, inevitably the question would always arise, “Do you let your residents do that?”. The answer was always an emphatic, YES. However, we began to explore ways in which we could improve our trainees’ education on pedicle screw insertion, both manual and power, given the complexity of the osseous anatomy in pediatric spinal deformity cases.

Measurable Lumbar Motion Remains 1 Year After Vertebral Body Tethering.

Vertebral body tethering (VBT) is growing in popularity for skeletally immature patients with scoliosis because of presumed preservation of spinal motion. Although results have shown preserved thoracic motion, there is minimal data to support motion over the lumbar instrumented segments after VBT. The purpose of this study was to analyze the range of motion of the thoracolumbar and lumbar spine after lumbar VBT. Retrospective review of patients treated with lumbar VBT underwent low-dose biplanar flexion-extension and lateral bending radiographs at 1 year after surgery to assess motion. Coronal motion at 1 year was compared with preoperative side-bending radiographs. The angle subtended by the screws at the upper instrumented vertebra and lower instrumented vertebra was measured on left-bending and right-bending radiographs to evaluate the coronal arc of motion and was compared with preoperative values over the same levels measured from the end plates. At 1 year postoperatively, the sagittal angle was measured over the instrumented levels on flexion and extension radiographs. Of the 71 scoliosis patients who underwent VBT at our center eligible for 1-year follow-up, 20 had lumbar instrumentation, all of whom had lumbar bending films available at 1 year after surgery. Seven patients had both thoracic and lumbar VBT on the same day and 13 had lumbar or thoracolumbar tether only. Mean age was 13.5±1.9 years. Mean preoperative major coronal curve measured 52+8 degrees (range: 42 to 70) and mean 27 degrees (range: 13 to 40) at latest follow-up. Mean levels instrumented was 8 (range: 5 to 12), with the lowest instrumented level typically L3 (N=14). The mean preoperative coronal arc of motion over the instrumented segments was 38±13 degrees (range: 19 to 73 degrees) and decreased after surgery to a mean arc of 17±7 degrees (range: 7 to 31 degrees). However, 19 of the 20 (95%) had at least a 10-degree coronal arc of motion. Patients maintained on average 46% (range: 22% to 100%) of their preoperative coronal arc of lumbar motion over the instrumented lumbar segments. On flexion-extension lateral radiographs taken at 1 year postoperatively, there was a mean postoperative arc of motion of 30±13 degrees. Lumbar VBT resulted in preserved flexion and extension motion at 1 year postoperatively. We also noted some preserved coronal plane motion, but this was decreased compared with preoperative values by ~50%. These findings provide proof of concept that some spinal motion is preserved after lumbar VBT in contrast to lumbar fusion where no motion is retained over the instrumented segments.

Limited generalizability of single deep neural network for surgical instrument segmentation in different surgical environments

Clarifying the generalizability of deep-learning-based surgical-instrument segmentation networks in diverse surgical environments is important in recognizing the challenges of overfitting in surgical-device development. This study comprehensively evaluated deep neural network generalizability for surgical instrument segmentation using 5238 images randomly extracted from 128 intraoperative videos. The video dataset contained 112 laparoscopic colorectal resection, 5 laparoscopic distal gastrectomy, 5 laparoscopic cholecystectomy, and 6 laparoscopic partial hepatectomy cases. Deep-learning-based surgical-instrument segmentation was performed for test sets with (1) the same conditions as the training set; (2) the same recognition target surgical instrument and surgery type but different laparoscopic recording systems; (3) the same laparoscopic recording system and surgery type but slightly different recognition target laparoscopic surgical forceps; (4) the same laparoscopic recording system and recognition target surgical instrument but different surgery types. The mean average precision and mean intersection over union for test sets 1, 2, 3, and 4 were 0.941 and 0.887, 0.866 and 0.671, 0.772 and 0.676, and 0.588 and 0.395, respectively. Therefore, the recognition accuracy decreased even under slightly different conditions. The results of this study reveal the limited generalizability of deep neural networks in the field of surgical artificial intelligence and caution against deep-learning-based biased datasets and models.Trial Registration Number: 2020-315, date of registration: October 5, 2020.

Analysis of Current Deep Learning Networks for Semantic Segmentation of Anatomical Structures in Laparoscopic Surgery.

Semantic segmentation of anatomical structures in laparoscopic videos is a crucial task to enable the development of new computer-assisted systems that can assist surgeons during surgery. However, this is a difficult task due to artifacts and similar visual characteristics of anatomical structures on the laparoscopic videos. Recently, deep learning algorithms have been showed promising results on the segmentation of laparoscopic instruments. However, due to the lack of large public datasets for semantic segmentation of anatomical structures, there are only a few studies on this task. In this work, we evaluate the performance of five networks, namely U-Net, U-Net++, DynUNet, UNETR and DeepLabV3+, for segmentation of laparoscopic cholecystectomy images from the recently released CholecSeg8k dataset. To the best of our knowledge, this is the first benchmark performed on this dataset. Training was performed with dice loss. The networks were evaluated on segmentation of 8 anatomical structures and instruments, performance was quantified through the dice coefficient, intersection over union, recall, and precision. Apart from the U-Net, all networks obtained scores similar to each other, with the U-Net++ being the network with the best overall score with a mean Dice value of 0.62. Overall, the results show that there is still room for improvement in the segmentation of anatomical structures from laparoscopic videos. Clinical Relevance- The results of this study show the potential of deep neural networks for the segmentation of anatomical structures in laparoscopic images which can later be incorporated into computer-aided systems for surgeons.

MVD-Net: Semantic Segmentation of Cataract Surgery Using Multi-View Learning.

Semantic segmentation of surgery scenarios is a fundamental task for computer-aided surgery systems. Precise segmentation of surgical instruments and anatomies contributes to capturing accurate spatial information for tracking. However, uneven reflection and class imbalance lead the segmentation in cataract surgery to a challenging task. To desirably conduct segmentation, a network with multi-view decoders (MVD-Net) is proposed to present a generalizable segmentation for cataract surgery. Two discrepant decoders are implemented to achieve multi-view learning with the backbone of U-Net. The experiment is carried out on the Cataract Dataset for Image Segmentation (CaDIS). The ablation study verifies the effectiveness of the proposed modules in MVD-Net, and superior performance is provided by MVD-Net in the comparison with the state-of-the-art methods. The source code will be publicly released.

Automated Microsurgical Tool Segmentation and Characterization in Intra-Operative Neurosurgical Videos.

Checklist based routine evaluation of surgical skills in any medical school demands quality time and effort from the supervising expert and is highly influenced by assessor bias. Alternatively, automated video based surgical skill assessment is a simple and viable method to analyse surgical dexterity offline without the need for acute presence of an expert surgeon throughout the surgery. In this paper, a novel approach and results for the automated segmentation of microsurgical instruments from the real-world neurosurgical video dataset was presented. The proposed tool segmentation model showcased mean average precision of 96.7% in detecting, and localizing five surgical instruments from the real-world neurosurgical videos. Accurate detection and characterization of motion features of the microsurgical tool from the novel annotated neurosurgical video dataset forms the key step towards automated surgical skill evaluation. Clinical Relevance- Tool segmentation, localization, and characterization in neurosurgical video, has several applications including assessing surgeons skills, training novice surgeons, understanding critical operating procedures post surgery, characterizing any critical anatomical response to the tool that leads to the success or failure of the surgery, and building models for conducting autonomous robotic surgery. Semantic segmentation, and characterization of the microsurgical tools forms the basis of the modern neurosurgery.