Multi-perspective region-based CNNs for vertebrae labeling in intraoperative long-length images.

PURPOSE: Effective aggregation of intraoperative x-ray images that capture the patient anatomy from multiple view-angles has the potential to enable and improve automated image analysis that can be readily performed during surgery. We present multi-perspective region-based neural networks that leverage knowledge of the imaging geometry for automatic vertebrae labeling in Long-Film images - a novel tomographic imaging modality with an extended field-of-view for spine imaging. METHOD: A multi-perspective network architecture was designed to exploit small view-angle disparities produced by a multi-slot collimator and consolidate information from overlapping image regions. A second network incorporates large view-angle disparities to jointly perform labeling on images from multiple views (viz., AP and lateral). A recurrent module incorporates contextual information and enforce anatomical order for the detected vertebrae. The three modules are combined to form the multi-view multi-slot (MVMS) network for labeling vertebrae using images from all available perspectives. The network was trained on images synthesized from 297 CT images and tested on 50 AP and 50 lateral Long-Film images acquired from 13 cadaveric specimens. Labeling performance of the multi-perspective networks was evaluated with respect to the number of vertebrae appearances and presence of surgical instrumentation. RESULTS: The MVMS network achieved an F1 score of >96% and an average vertebral localization error of 3.3 mm, with 88.3% labeling accuracy on both AP and lateral images - (15.5% and 35.0% higher than conventional Faster R-CNN on AP and lateral views, respectively). Aggregation of multiple appearances of the same vertebra using the multi-slot network significantly improved the labeling accuracy (p < 0.05). Using the multi-view network, labeling accuracy on the more challenging lateral views was improved to the same level as that of the AP views. The approach demonstrated robustness to the presence of surgical instrumentation, commonly encountered in intraoperative images, and achieved comparable performance in images with and without instrumentation (88.9% vs. 91.2% labeling accuracy). CONCLUSION: The MVMS network demonstrated effective multi-perspective aggregation, providing means for accurate, automated vertebrae labeling during spine surgery. The algorithms may be generalized to other imaging tasks and modalities that involve multiple views with view-angle disparities (e.g., bi-plane radiography). Predicted labels can help avoid adverse events during surgery (e.g., wrong-level surgery), establish correspondence with labels in preoperative modalities to facilitate image registration, and enable automated measurement of spinal alignment metrics for intraoperative assessment of spinal curvature.

Post-operative infection after minimally invasive versus open transforaminal lumbar interbody fusion (TLIF): literature review and cost analysis.

INTRODUCTION: Surgical site infection (SSI) in the setting of lumbar fusion is associated with significant morbidity and medical resource utilization. To date, there have been no studies conducted with sufficient power to directly compare the incidence of SSI following minimally invasive (MIS) vs. open TLIF procedures. Furthermore, studies are lacking that quantify the direct medical cost of SSI following fusion procedures. We set out to determine the incidence of SSI in patients undergoing MIS vs. open TLIF reported in the literature and to determine the direct hospital cost associated with the treatment of SSI following TLIF at our institution. METHODS: A systematic Medline search was performed to identify all published studies assessing SSI after MIS or open TLIF. The cumulative incidence of SSI was calculated from all reported cohorts and compared between MIS vs. open TLIF. In order to determine the direct hospital costs associated with the treatment of SSI following TLIF, we retrospectively reviewed 120 consecutive TLIFs performed at our institution, assessed the incidence of SSI, and calculated the SSI-related hospital costs from accounting and billing records. RESULTS: To date, there have been 10 MIS-TLIF cohorts (362 patients) and 20 open-TLIF cohorts (1 133 patients) reporting incidences of SSI. The cumulative incidence of reported SSI was significantly lower for MIS vs. open-TLIF (0.6% vs. 4.0%, p=0.0005). In our experience with 120 open TLIF procedures, SSI occurred in 6 (5.0%) patients. The mean hospital cost associated with the treatment of SSI following TLIF was $ 29,110 in these 6 cases. The 3.4% decrease in reported incidence of SSI for MIS vs. open-TLIF corresponds to a direct cost savings of $ 98,974 per 100 MIS-TLIF procedures performed. CONCLUSIONS: Post-operative wound infections following TLIF are costly complications. MIS vs. open TLIF is associated with a decreased reported incidence of SSI in the literature and may be a valuable tool in reducing hospital costs associated with spine care.

CT fluoroscopically guided percutaneous placement of transiliosacral rod for sacral insufficiency fracture: case report and technique.

Treatment of sacral insufficiency fractures (SIFs) has traditionally been conservative, but several patients have been treated with percutaneous sacroplasty. Unfortunately, in the setting of severe, bilateral SIFs, cement may not withstand shear forces present at the lumbosacral junction, and surgical hardware may not provide adequate fixation in osteoporotic, cancellous bone of the sacrum, leading to eventual pseudarthrosis. Thus, we propose a novel technique in which guidance with CT fluoroscopy allows placement of a transiliosacral bar in conjunction with sacroplasty.

Comprehensive assessment of hemorrhage risks and outcomes after stereotactic brain biopsy.

OBJECT: Stereotactic brain biopsy has played an integral role in the diagnosis and management of brain lesions. At most centers, imaging studies following biopsy are rarely performed. The authors prospectively determined the acute hemorrhage rate after stereotactic biopsy by performing immediate postbiopsy intraoperative computerized tomography (CT) scanning. They then analyzed factors that may influence the risk of hemorrhage and the diagnostic accuracy rate. METHODS: Five hundred consecutive patients undergoing stereotactic brain biopsy underwent immediate postbiopsy intraoperative CT scanning. Before surgery, routine preoperative coagulation studies were performed in all patients. All medical charts, laboratory results, preoperative imaging studies, and postoperative imaging studies were reviewed. In 40 patients (8%) hemorrhage was detected using immediate postbiopsy intraoperative CT scanning. Neurological deficits developed in six patients (1.2%) and one patient (0.2%) died. Symptomatic delayed neurological deficits developed in two patients (0.4%), despite the fact that the initial postbiopsy CT scans in these cases did not show acute hemorrhage. Both patients had large intracerebral hemorrhages that were confirmed at the time of repeated imaging. The results of a multivariate logistic regression analysis of the risk of postbiopsy hemorrhage of any size showed a significant correlation only with the degree to which the platelet count was below 150,000/mm3 (p = 0.006). The results of a multivariate analysis of a hemorrhage measuring greater than 5 mm in diameter also showed a correlation between the risk of hemorrhage and a lesion location in the pineal region (p = 0.0086). The rate at which a nondiagnostic biopsy specimen was obtained increased as the number of biopsy samples increased (p = 0.0073) and in accordance with younger patient age (p = 0.026). CONCLUSIONS: Stereotactic brain biopsy was associated with a low likelihood of postbiopsy hemorrhage. The risk of hemorrhage increased steadily as the platelet count fell below 150,000/mm3. The authors found a small but definable risk of delayed hemorrhage, despite unremarkable findings on an immediate postbiopsy head CT scan. This risk justifies an overnight hospital observation stay for all patients after having undergone stereotactic brain biopsy.

L-buthionine sulfoximine potentiates the antitumor effect of 4-hydroperoxycyclophosphamide when administered locally in a rat glioma model.

OBJECTIVE: L-buthionine sulfoximine (BSO) inhibits glutathione synthesis and may modulate tumor resistance to some alkylating agents, but it has not been proven effective in the treatment of intracranial neoplasms. To evaluate this drug for the treatment of brain tumors, we studied the use of BSO for potentiating the antineoplastic effect of 4-hydroxyperoxycyclophosphamide (4-HC) in the rat 9L glioma model. METHODS: The survival of male Fischer 344 rats with intracranial 9L gliomas was measured after implantation of controlled-release polymers containing one of the following: no drug, BSO, 4-HC, or both BSO and 4-HC. The efficacy of intracranial 4-HC treatment was assessed with and without serial systemic intraperitoneal BSO injections. Tissue glutathione levels were measured in the brains, tumors, and livers of animals treated with intraperitoneal injections or local delivery of BSO. RESULTS: The median survival of animals treated with intracranial polymers containing 4-HC was 2.3 times greater than that of controls. This survival benefit was doubled by local delivery of BSO. In contrast, systemic BSO therapy did not improve survival time. In animals that were treated systemically, both liver and tumor glutathione levels were significantly lower than they were in control animals. In the locally treated animals, glutathione levels were reduced in the brain tumor but not in the liver. CONCLUSION: These results demonstrate that local but not systemic delivery of BSO enhances the antineoplastic effect of 4-HC in this rat 9L glioma model. In addition, because local delivery of BSO within the brain did not deplete glutathione levels systemically, this method of treatment may be safer than systemic administration of BSO.

Relationship between type of health insurance and time to inpatient rehabilitation placement for surgical subspecialty patients.

A significant proportion of patients on a neurosurgical service require inpatient rehabilitation placement after discharge. The relationship between the type of health insurance of the patient at the time of admission and the time to placement of patients has not previously been addressed. We prospectively studied all patients on the adult neurosurgical service at our hospital to determine whether the type of health insurance carried by patients is related to the time necessary to arrange acceptance into inpatient rehabilitation facilities. Ninety-one patients (51 men, 40 women; mean age, 56 years) admitted to the neurosurgery service during a 6-month period required inpatient rehabilitation placement after discharge. The time in days between the request for placement into a rehabilitation facility and the acceptance of the patient was examined. The mean time for placement of patients with and without health insurance at the time of admission was 0.8 days and 2.1 days, respectively (overall mean, 1.1 days) (P < .002). No statistically significant associations were found between age, sex, or race of the patient and the time to placement. In addition, there was no difference in the time to placement between those patients admitted as a result of trauma and those patients admitted for reasons other than trauma. These results indicate that among patients on a neurosurgical service, patients with private health insurance are accepted into inpatient rehabilitation approximately 1 day sooner than patients without private health insurance. Patients without private health insurance are delayed in their transfer to inpatient rehabilitation facilities and more aggressive inpatient rehabilitation. How this finding translates into an increase in cost of care or a decrease in patient outcomes is unknown.

Combination of stereotactic radiosurgery and cytokine gene-transduced tumor cell vaccination: a new strategy against metastatic brain tumors.

OBJECT: To determine if the combination of radiosurgery and tumor cell vaccine would enhance the therapy of metastatic lesions of the central nervous system (CNS), the authors examined the antitumoral effects of radiosurgery and cytokine-transduced tumor cell vaccine. METHODS: Fifty-five rats underwent intracranial implantation of 5 x 10(3) MADB 106 cells. On Day 3 after tumor implantation, 34 rats were inoculated in the flank with nonirradiated MADB 106 cells that had been retrovirally transduced to express granulocyte-macrophage colony-stimulating factor or interleukin-4. Twenty-seven rats (17 animals that had received the vaccine and 10 that had not) underwent radiosurgery performed using a gamma knife at maximum doses of 32 Gy on Day 5. No animals in the untreated group or in the vaccine-alone groups survived longer than 21 days. Animals treated by ra diosurgery alone displayed prolonged survival in comparison with untreated animals (p < 0.0001), but only one of 10 animals survived longer than 55 days. In contrast, 14 of 17 animals that received the combination therapy of radiosurgery and vaccination survived longer than 55 days (p = 0.0003 compared with animals that underwent radiosurgery alone). On Day 55, the long-term survivors were challenged by parental MADB 106 cells, which were implanted in the contralateral hemisphere. All animals from the combination therapy groups survived longer than 50 days after this challenge, but the single survivor from the radiosurgery-alone group died of tumor growth in 27 days. CONCLUSIONS: The combination of radiosurgery and cytokine gene-transduced tumor cell vaccine markedly prolonged animal survival and protected animals from a subsequent challenge by parental tumor cells placed in the CNS. The data provided by this study indicate that this combination therapy represents a strategy that may have clinical applicability for single and/or multiple metastatic brain tumors.

Unruptured cerebral aneurysm producing a thunderclap headache.

A sudden and severe headache is the most common presentation of an acutely ruptured cerebral aneurysm. A similar headache in the absence of subarachnoid blood has rarely been ascribed to an unruptured cerebral aneurysm, but may result from acute aneurysm expansion and indicate a high risk of future rupture. We present a patient who developed a sudden, severe, "thunderclap" headache, with no associated neurological deficit. Computed tomogram and lumbar cerebral spinal fluid obtained 5.5 hours after headache onset were negative for subarachnoid hemorrhage. The patient underwent cerebral angiography which revealed a posterior communicating artery aneurysm with an associated daughter aneurysm. Craniotomy and clip obliteration of the aneurysm were performed. The aneurysm dome was very thin and there was no evidence of recent or old hemorrhage. A "thunderclap" headache without subarachnoid hemorrhage may be an important harbinger of a cerebral aneurysm with the potential for future rupture. Early recognition and neurovascular imaging of aneurysms presenting in this rare fashion are warranted.

Survival of patients with high grade glioma treated with intrathecal thiotriethylenephosphoramide for ependymal or leptomeningeal gliomatosis.

BACKGROUND: The diagnosis of leptomeningeal dissemination of malignant glioma (meningeal gliomatosis) is associated with poor survival. Intrathecal (IT) chemotherapeutic agents used to achieve tumor control and improve survival include methotrexate, cytosine arabinoside (ara-C), thiotriethylenephosphoramide (thio-TEPA), neocarzinostatin, and 3-[(4-amino-2-methyl-5-pyrimidinyl)methyl]-1-(2-chloroethyl)-1-nitros ourea hydrochloride (ACNU). Little information exists about survival following administration of IT chemotherapy. The authors report survival data from a series of patients with supratentorial anaplastic astrocytoma (AA) or glioblastoma multiforme (GBM) treated for ependymal or leptomeningeal gliomatosis with IT thio-TEPA. METHODS: The authors reviewed the records of 14 patients treated between 1991 and 1997 (GBM: n = 9; AA: n = 5). All patients were diagnosed with ependymal (n = 8) or leptomeningeal (n = 6) dissemination of tumor on the basis of clinical signs and symptoms, ependymal or leptomeningeal contrast enhancement on magnetic resonance imaging (MRI), and/or cerebrospinal fluid analysis. All 14 patients underwent placement of a ventricular reservoir system and subsequent instillation of IT thio-TEPA on a weekly basis for 6-12 weeks. Response to treatment was evaluated clinically and by MRI at intervals of 1-3 months and 3-6 months from the initiation of IT thio-TEPA. Data on survival from the time of diagnosis of dissemination was assessed. RESULTS: The median survival, from the time of diagnosis of ependymal or leptomeningeal dissemination, of patients who received IT thio-TEPA was 10 months (AA = 19 months; GBM = 10 months). Five of 14 patients had a radiographic response to treatment within 6 months. The median survival of patients with a radiographic response was 15.5 months, compared with 10 months for nonresponders. No significant neurotoxicity or myelopathy was observed. CONCLUSIONS: Early treatment with IT thio-TEPA may result in improved survival with minimal morbidity. Radiographic response may predict prolonged survival.

Morbidity and survival after 1,3-bis(2-chloroethyl)-1-nitrosourea wafer implantation for recurrent glioblastoma: a retrospective case-matched cohort series.

OBJECTIVE: To determine the risks and survival benefit associated with implantation of an absorbable, 1,3-bis(2chloroethyl)-1-nitrosourea-impregnated polymer wafer, we prospectively studied patients with recurrent glioblastoma multiforme and compared them with a demographically matched cohort group. METHODS: Over a 29-month period, 62 patients underwent operations. All had tumor growth despite standard treatment, a Karnofsky performance score of > or =70, and histopathological confirmation of glioblastoma. Seventeen patients underwent gross total resection with placement of 1,3-bis(2-chloroethyl)-1-nitrosourea wafers (wafer group) at a median 44 weeks from diagnosis (6 women, 11 men; median age, 56 years). A cohort group of 45 patients undergoing surgery for recurrent glioblastoma during the same time period, but not receiving wafers, was identified. Surgery was performed at a median 47 weeks from diagnosis (14 women, 31 men; median age, 54 years). RESULTS: Within 6 weeks of surgery, 13 complications were identified in 8 patients in the wafer group. In the cohort group, 6 patients sustained 8 complications. We were unable to identify any survival advantage using Kaplan-Meier analysis. In the wafer group, median survival was 58 weeks from diagnosis and 14 weeks from wafer implantation. In the cohort group, median survival was 97 weeks from diagnosis and 50 weeks from operation. CONCLUSION: 1,3-bis(2-chloroethyl)-1-Nitrosourea wafer implantation for recurrent glioblastoma was associated with a higher risk of postoperative complications, particularly those related to infection and wound healing. No clear survival benefit associated with wafer implantation was identified.

Near-infrared spectroscopy monitored cerebral venous thrombolysis.

BACKGROUND: Cerebral venous thrombosis is a clinical entity which is readily diagnosed with the advent of modern imaging techniques. Anticoagulation is now a standard therapy, but more recent treatment strategies have included endovascular thrombolysis. While the endpoint of this intervention both clinically and radiographically has not been defined, noninvasive monitoring techniques may add further objective measures of treatment response. CLINICAL PRESENTATION: We present a patient with a four day history of worsening headache and papilledema on exam. Superior sagittal, straight, and bilateral transverse sinus thromboses were identified on computed tomography and angiography. INTERVENTION: Emergent endovascular thrombolysis by a transvenous approach re-established venous patency and resulted in immediate resolution of the patient's symptoms. Cerebral oximetry by near-infrared spectroscopy was utilized during the procedure, and changes in chromophore concentrations correlated directly with angiographic and clinical resolution of the thrombosis. CONCLUSION: Near-infrared spectroscopy can provide continuous feedback during thrombolytic therapy in cerebral venous thrombosis and may help define endpoints of such intervention.

A new subarachnoid hemorrhage grading system based on the Glasgow Coma Scale: a comparison with the Hunt and Hess and World Federation of Neurological Surgeons Scales in a clinical series.

OBJECTIVE: Although the Hunt and Hess Scale (HHS) and World Federation of Neurological Surgeons Scale (WFNSS) are the most widely used subarachnoid hemorrhage (SAH) grading systems, neither system has achieved universal acceptance. We propose a simplified grading system based entirely on the Glasgow Coma Scale (GCS), which compresses the 15-point GCS into five grades that are comparable with those of the HHS and WFNSS. We refer to this system as the GCS grading system and present a direct comparison with the HHS and WFNSS for predictive value regarding patient outcome and interrater reliability. METHODS: We reviewed 291 consecutive patients with aneurysms treated at our institution between January 1992 and January 1996 and compared the admission grades from the GCS, WFNSS, and HHS with outcome measures at discharge from hospitalization. The Glasgow Outcome score was used as the major outcome measure to evaluate the predictive value of the three scales. Mortality and length of stay (LOS) were also evaluated as outcome measures. The predictive value of each scale was tested with an ordinal logistic regression model for Glasgow Outcome score, a logistic regression model for mortality data, and a linear regression model for LOS. RESULTS: Using the logistic regression model, the GCS was the best predictor of discharge Glasgow Outcome score, with an odds ratio of 2.585 (P = 0.0001), compared with 2.311 (P = 0.0001) for the WFNSS and 2.262 (P = 0.0001) for the HHS. Using mortality data in the logistic model, the HHS was the best predictor, with an odds ratio of 3.391 (P = 0.0001), compared with 2.859 (P = 0.0001) for the GCS and 2.560 (P = 0.0001) for the WFNSS. Each of the three scales had a high predictive value for LOS, using a linear model. We discuss, however, the problematic nature of LOS as an outcome measure for SAH. Interrater reliability for each scale was evaluated using kappa statistics, based on 15 additional patients evaluated prospectively, and showed that the GCS grade also had the greatest interrater reliability, with a kappa of 0.46 (P = 0.0002), compared with 0.41 (P = 0.0005) for the HHS and 0.27 (P = 0.027) for the WFNSS. CONCLUSION: We conclude that the GCS grade has equal or greater predictive value regarding outcome after SAH than do the currently used grading systems and that it has greater reproducibility across observers. Broader familiarity with the GCS among medical and paramedical personnel may further enhance the usefulness of the GCS grade over the HHS and WFNSS in providing a standardized, universally accepted grading system for SAH.

Isolation and characterization of sets of anti-L-fucose and anti-bovine serum albumin antibodies directed against a glycoconjugate of L-fucose and bovine serum albumin.

A set of anti-carbohydrate antibodies and a set of anti-protein antibodies were isolated from the serum of rabbits immunized with a glycoconjugate of L-fucose and bovine serum albumin. The sets were separated by affinity chromatography by a two-column method on adsorbents with L-fucose or bovine serum albumin ligands. Isoelectrofocusing results showed that the anti-carbohydrate antibodies consisted of 11 molecular species and the anti-bovine serum albumin antibodies consisted of seven molecular species. The anti-carbohydrate antibodies are all of the IgG type while the anti-protein antibodies contain three types of globulin molecules, IgA, IgG, and IgM. The former antibodies should be useful as markers for unique glycoproteins of diseased cells and the latter antibodies may be useful for investigating the mechanism of simultaneous synthesis of three types of immunoglobulins.

Affinity chromatography of two sets of isomeric antibodies having specificity for different oligosaccharide units of gum arabic.

Two sets of antibodies directed against different carbohydrate units of gum arabic were isolated from the sera of rabbits immunized intramuscularly with gum arabic and Freund's complete adjuvant. The isolation was effected by affinity chromatography on two columns attached in series and containing an absorbent of AH-Sepharose 4B with ligands of partially hydrolyzed gun arabic in the first column and an adsorbent of AH-Sepharose 4B with ligands of native gum arabic in the second column. The two populations of anti-gum arabic antibodies were obtained and have been designated as Set 1 and Set 2 on the basis of their mobilities on agar diffusion. The antibodies of Set 1 consisted of 4 isomeric antibodies and those of Set 2 consisted of 11 isomeric antibodies. Native gum arabic samples were oxidized with periodate or reduced with sodium borohydride and carbodiimide under standard conditions and the modified samples were totally inactive in the precipitin test. On the basis of methylation data and immunological results it was concluded that terminal disaccharide moieties of the gum having the structure beta-D-glucosyluronic acid-(1----6)-D-galactose and alpha-L-arabinofuranosyl-(1----4)-D-glucuronic acid were the immunodeterminant groups for Set 1 and Set 2 antibodies, respectively.