Artificial intelligence innovations in neurosurgical oncology: a narrative review.

PURPOSE: Artificial Intelligence (AI) has become increasingly integrated clinically within neurosurgical oncology. This report reviews the cutting-edge technologies impacting tumor treatment and outcomes. METHODS: A rigorous literature search was performed with the aid of a research librarian to identify key articles referencing AI and related topics (machine learning (ML), computer vision (CV), augmented reality (AR), virtual reality (VR), etc.) for neurosurgical care of brain or spinal tumors. RESULTS: Treatment of central nervous system (CNS) tumors is being improved through advances across AI-such as AL, CV, and AR/VR. AI aided diagnostic and prognostication tools can influence pre-operative patient experience, while automated tumor segmentation and total resection predictions aid surgical planning. Novel intra-operative tools can rapidly provide histopathologic tumor classification to streamline treatment strategies. Post-operative video analysis, paired with rich surgical simulations, can enhance training feedback and regimens. CONCLUSION: While limited generalizability, bias, and patient data security are current concerns, the advent of federated learning, along with growing data consortiums, provides an avenue for increasingly safe, powerful, and effective AI platforms in the future.

Synthetic interpolated DSA for radiation exposure reduction via gamma variate contrast flow modeling: a retrospective cohort study.

BACKGROUND: Digital subtraction angiography (DSA) yields high cumulative radiation dosages (RD) delivered to patients. We present a temporal interpolation of low frame rate angiograms as a method to reduce cumulative RDs. METHODS: Patients undergoing interventional evaluation and treatment of cerebrovascular vasospasm following subarachnoid hemorrhage were retrospectively identified. DSAs containing pre- and post-intervention runs capturing the full arterial, capillary, and venous phases with at least 16 frames each were selected. Frame rate reduction (FRR) of the original DSAs was performed to 50%, 66%, and 75% of the original frame rate. Missing frames were regenerated by sampling a gamma variate model (GVM) fit to the contrast response curves to the reduced data. A formal reader study was performed to assess the diagnostic accuracy of the "synthetic" studies (sDSA) compared to the original DSA. RESULTS: Thirty-eight studies met inclusion criteria (average RD 1,361.9 mGy). Seven were excluded for differing views, magnifications, or motion. GVMs fit to 50%, 66%, and 75% FRR studies demonstrated average voxel errors of 2.0 ± 2.5% (mean ± standard deviation), 6.5 ± 1.5%, and 27 ± 2%, respectively for anteroposterior projections, 2.0 ± 2.2%, 15.0 ± 3.1%, and 14.8 ± 13.0% for lateral projections, respectively. Reconstructions took 0.51 s/study. Reader studies demonstrated an average rating of 12.8 (95% CI 12.3-13.3) for 75% FRR, 12.7 (12.2-13.2) for 66% FRR and 12.0 (11.5-12.5) for 50% FRR using Subjective Image Grading Scale. Kendall's coefficient of concordance resulted in W = 0.506. CONCLUSION: FRR by 75% combined with GVM reconstruction does not compromise diagnostic quality for the assessment of cerebral vasculature. RELEVANCE STATEMENT: Using this novel algorithm, it is possible to reduce the frame rate of DSA by as much as 75%, with a proportional reduction in radiation exposure, without degrading imaging quality. KEY POINTS: • DSA delivers some of the highest doses of radiation to patients. • Frame rate reduction (FRR) was combined with bolus tracking to interpolate intermediate frames. • This technique provided a 75% FRR with preservation of diagnostic utility as graded by a formal reader study for cerebral angiography performed for the evaluation of cerebral vasospasm. • This approach can be applied to other types of angiography studies.

Machine learning for automated and real-time two-dimensional to three-dimensional registration of the spine using a single radiograph.

OBJECTIVE: The goal of this work was to methodically evaluate, optimize, and validate a self-supervised machine learning algorithm capable of real-time automatic registration and fluoroscopic localization of the spine using a single radiograph or fluoroscopic frame. METHODS: The authors propose a two-dimensional to three-dimensional (2D-3D) registration algorithm that maximizes an image similarity metric between radiographic images to identify the position of a C-arm relative to a 3D volume. This work utilizes digitally reconstructed radiographs (DRRs), which are synthetic radiographic images generated by simulating the x-ray projections as they would pass through a CT volume. To evaluate the algorithm, the authors used cone-beam CT data for 127 patients obtained from an open-source de-identified registry of cervical, thoracic, and lumbar scans. They systematically evaluated and tuned the algorithm, then quantified the convergence rate of the model by simulating C-arm registrations with 80 randomly simulated DRRs for each CT volume. The endpoints of this study were time to convergence, accuracy of convergence for each of the C-arm's degrees of freedom, and overall registration accuracy based on a voxel-by-voxel measurement. RESULTS: A total of 10,160 unique radiographic images were simulated from 127 CT scans. The algorithm successfully converged to the correct solution 82% of the time with an average of 1.96 seconds of computation. The radiographic images for which the algorithm converged to the solution demonstrated 99.9% registration accuracy despite utilizing only single-precision computation for speed. The algorithm was found to be optimized for convergence when the search space was limited to a ± 45° offset in the right anterior oblique/left anterior oblique, cranial/caudal, and receiver rotation angles with the radiographic isocenter contained within 8000 cm3 of the volumetric center of the CT volume. CONCLUSIONS: The investigated machine learning algorithm has the potential to aid surgeons in level localization, surgical planning, and intraoperative navigation through a completely automated 2D-3D registration process. Future work will focus on algorithmic optimizations to improve the convergence rate and speed profile.

Case report: Minimally invasive modification of the Goel-Harms atlantoaxial fusion utilizing percutaneous screws and intra-articular cage is feasible and results in decreased blood loss.

Treatment of atlantoaxial pathology is often associated with significant morbidity and mortality. While surgical techniques for fixation are well established, approaches that minimize blood loss and muscle dissection are advantageous for expediting recovering and minimizing surgical risk. We present a 34-year-old female who presented with a Type III odontoid fracture requiring surgical fixation. She underwent a C1-2 fusion employing a novel minimally invasive modification of the Goel-Harms atlantoaxial fusion using percutaneous screws and intra-articular cage.

Sociodemographic disparities as a determinant of fetal versus postnatal surgical myelomeningocele repair.

OBJECTIVE: Relatively few women undergo open maternal-fetal surgery (OMFS) for myelomeningocele (MMC) despite the potential to reverse hindbrain herniation, reduce the rate of infant shunt-dependent hydrocephalus, and improve ambulation. These benefits have the potential to significantly reduce morbidity and lifetime medical care. In this study, the authors examined demographics and socioeconomic variables of women who were offered and opted for OMFS for MMC versus postnatal MMC surgery, with the purpose of identifying variables driving the disparity between these two patient populations. METHODS: This was a retrospective case-control study of patients who underwent evaluation for OMFS for MMC at a single academic hospital from 2015 to 2020. Race/ethnicity, primary insurance type, zip code, and BMI were collected and compared by treatment received and eligibility status for OMFS. Prevalence odds ratios were used to test for associations between each independent variable and the two outcomes. Logistical regression models were utilized to determine significant predictors of undergoing OMFS and being eligible for OMFS. RESULTS: Of 96 women, 36 underwent OMFS for MMC, 40 received postnatal repair, and 20 either terminated the pregnancy or received care at another institution. Overall, 66 (68.8%) women were White, 14 (14.6%) were Black, 13 (13.5%) were Hispanic/Latinx, 1 (1.0%) was Asian, and 2 (2.1%) identified as other or multiple races. Among women who underwent OMFS for MMC, 27 (75.0%) were White, 2 (5.6%) were Black, 4 (11.1%) were Hispanic/Latinx, 1 (2.8%) was Asian, and 2 (5.6%) identified as other or multiple races. Having private insurance or TRICARE was associated with higher odds of being eligible for OMFS compared with women who were uninsured or had Medicaid when accounting for race and income (OR 3.87, 95% CI 1.51-9.59). CONCLUSIONS: The population evaluated and treated for MMC was homogeneous and insufficiently representative of the population affected by the disease. This finding raises concern, as it suggests underlying barriers to formal evaluation for OMFS for MMC. Insurance status and BMI have a significant association between the access to and election of OMFS, revealing socioeconomic disparities. This was the first study to explore sociodemographic characteristics of patient populations who may be at risk for limited access to highly specialized fetal surgical care.

Novel use of percutaneous cervical pedicle inlet screws for supplemental posterior fixation after anterior cervical deformity correction.

INTRODUCTION: Correction of cervical deformity can be achieved using anterior cervical fixation and fusion techniques. However, supplemental posterior fixation is a critical component for ensuring biomechanical longevity and favorable patient outcomes. We present a novel percutaneous technique for posterior cervical fixation in patients where cervical pedicle (CP) screws may not be feasible and midline muscle dissection is not needed. METHODS: Three patients presented to our hospital with cervical pathology amendable to circumferential cervical fusion. After adequate deformity correction was performed through an anterior cervical decompression and fusion, staged posterior supplemental fixation was achieved using percutaneous CP inlet (CPI) screws using a percutaneous muscle-sparing approach. RESULTS: All three patients underwent CPI screw placement without postoperative neurovascular complications. Postoperative radiographic follow-up showed the desired, proper screw placement, with continued maintained cervical alignment. CONCLUSIONS: CPI screw placement may be alternative hybrid screw that achieves a advantageous safety profile while also avoiding an open midline exposure.

Feasibility of stereo electroencephalogram (SEEG) with little to no scalp bone; a case report.

Stereo electroencephalogram (SEEG) electrode placement with cranially fixed guide bolts is recognized as one of the most accurate and safest implantation strategies to sample deep and buried cortex during certain clinical scenarios involving epilepsy surgery. Bone thickness of less than 2 mm is a relative contraindication to SEEG. Here, we describe a case drug-resistant focal epilepsy where prior craniotomies, infections and radiation therapy yielded limited skull bone requiring invasive EEG monitoring. Due to the inability to use bolts over areas with limited skull bone, we successfully utilized a combination of the standard and a modified SEEG techniques for implantation and stabilization of intracranial electrodes without complications. This strategy enabled optimal intracranial EEG monitoring and surgical management of the patient's drug-resistant focal seizures.

Computational methods for visualizing and measuring verapamil efficacy for cerebral vasospasm.

Cerebral vasospasm is a dreaded sequelae of aneurysmal subarachnoid hemorrhage (aSAH), requiring timely intervention with therapeutic goals of improving brain perfusion. There are currently no standardized real-time, objective assessments of the interventional procedures performed to treat vasospasm. Here we describe real-time techniques to quantify cerebral perfusion during interventional cerebral angiography. We retrospectively analyzed 39 consecutive cases performed to treat clinical vasospasm and quantified the changes in perfusion metrics between pre- and post- verapamil administrations. With Digital Subtraction Angiography (DSA) perfusion analysis, we are able to identify hypoperfused territories and quantify the exact changes in cerebral perfusion for each individual case and vascular territory. We demonstrate that perfusion analysis for DSA can be performed in real time. This provides clinicians with a colorized map which directly visualizes hypoperfused tissue, combined with associated perfusion statistics. Quantitative thresholds and analysis based on DSA perfusion may assist with real-time dosage estimation and help predict response to treatment, however future prospective analysis is required for validation.

Pediatric nonoperative skull fractures: delayed complications and factors associated with clinic and imaging utilization.

OBJECTIVE: Skull fractures represent a common source of morbidity in the pediatric trauma population. This study characterizes the type of follow-up that these patients receive and discusses predictive factors for follow-up. METHODS: The authors reviewed cases of nonoperative pediatric skull fractures at a single academic hospital between 2007 and 2017. Clinical patient and radiological fractures were recorded. Recommended neurosurgical follow-up, follow-up appointments, imaging studies, and fracture-related complications were recorded. Statistical analyses were performed to identify predictors for outpatient follow-up and imaging. RESULTS: The study included 414 patients, whose mean age was 5.2 years; 37.2% were female, and the median length of stay was 1 day (IQR 0.9-4 days). During 438 clinic visits and a median follow-up period of 8 weeks (IQR 4-12, range 1-144 weeks), 231 imaging studies were obtained, mostly head CT scans (55%). A total of 283 patients were given recommendations to attend follow-up in the clinic, and 86% were seen. Only 12 complications were detected, including 7 growing skull fractures, 2 traumatic encephaloceles, and 3 cases of hearing loss. Primary care physician (PCP) status and insurance status were associated with a recommendation of follow-up, actual follow-up compliance, and the decision to order outpatient imaging in patients both with and without intracranial hemorrhage. PCP status remained an independent predictor in each of these analyses. Follow-up compliance was not associated with a patient's distance from home. Among patients without intracranial hemorrhage, a follow-up recommendation and actual follow-up compliance were associated with pneumocephalus and other polytraumatic injuries, and outpatient imaging was associated with a bilateral fracture. No complications were found in patients with linear fractures above the skull base in those without an intracranial hemorrhage. CONCLUSIONS: Pediatric nonoperative skull fractures drive a large expenditure of clinic and imaging resources to detect a relatively small profile of complications. Understanding the factors underlying the decision for clinic follow-up and additional imaging can decrease future costs, resource utilization, and radiation exposure. Factors related to injury severity and socioeconomic indicators were associated with outpatient imaging, the decision to follow up patients in the clinic, and patients' subsequent attendance. Socioeconomic status (PCP and insurance) may affect access to appropriate neurosurgical follow-up and deserves future research attention. Patients with no intracranial hemorrhage and with a linear fracture above the skull base do not appear to be at risk for delayed complications and could be candidates for reduced follow-up and imaging.

Predictors of Prostatic Artery Embolization Technical Outcomes: Patient and Procedural Factors.

PURPOSE: To identify technical factors that significantly change prostatic artery embolization (PAE) technical outcomes and to derive and test technical outcome predictive models. MATERIALS AND METHODS: Retrospective analysis of PAEs performed by 2 operators (OPs) was performed: OP1, between April 2014 and May 2017 (n = 150); OP2, between February 2017 and December 2017 (n = 67). Multivariate analysis with mixed-effects modeling was used to test significance and derive predictive models. Mean difference was used to analyze prediction accuracy. RESULTS: Moderate versus none subjective iliac tortuosity grade (SITG) and the presence of internal iliac atherosclerosis (PIIAA) versus none were associated with the following respective technical outcome increases: procedure time (PT): 43% (P < .01), 16% (P < .01); fluoroscopy time (FT): 47% (P < .01), 25% (P < .01); contrast volume (CV): 25.6 mL (P < .001), 13.7 mL (P = .01); and dose area product (DAP) 52% (P < .01), 20% (P = 0.03). Prostatic artery origin left obturator versus left superior vesical was associated with a 24% (P = .01) DAP decrease. For every 1 cc that prostate volume increased, CV decreased on average by 0.1 mL (P = .05). For every 1-cm decrease in patient height and 1-kg increase in weight, DAP increased on average by 0.02% (P < .01) for each. Unilateral versus bilateral versus 3-vessel embolization resulted in a 16.3-mL CV decrease on average for each additional vessel embolized (P = .03). The mean absolute differences between predicted and measured technical outcome values were: PT: 16 minutes, FT: 7 minutes, CV: 25 mL, and DAP: 44 Gy·cm2. CONCLUSIONS: In this study, higher SITGs and PIIAA most likely contributed to higher technical outcomes when controlling for the 2 OPs.

Ultradeep analysis of tumor heterogeneity in regions of somatic hypermutation.

Tumor heterogeneity is of growing importance in the treatment of cancers. Mutational hot spots are prime locations for determining number and proportions of low variant allele frequency (VAF) tumor subclones by next generation sequencing. Low VAF detection is complicated by poor mapping efficiency in regions with high mutation density. Our Deep-Drilling with iterative Mapping (DDiMAP) method retains variant allele patterns to aid in single nucleotide variation detection and generation of additional reference alleles, with remapping increasing coverage of highly mutated regions to capture data critical to heterogeneity analysis and enhancing sensitivity. DDiMAP outputs variant patterns with frequencies, enabling rapid phylogenetic analysis of ongoing mutation.

Intraclonal diversity in follicular lymphoma analyzed by quantitative ultradeep sequencing of noncoding regions.

Cancers are characterized by genomic instability, and the resulting intraclonal diversity is a prerequisite for tumor evolution. Therefore, metrics of tumor heterogeneity may prove to be clinically meaningful. Intraclonal heterogeneity in follicular lymphoma (FL) is apparent from studies of somatic hypermutation (SHM) caused by activation-induced deaminase (AID) in IGH. Aberrant SHM (aSHM), defined as AID activity outside of the IG loci, predominantly targets noncoding regions causing numerous "passenger" mutations, but it has the potential to generate rare significant "driver" mutations. The quantitative relationship between SHM and aSHM has not been defined. To measure SHM and aSHM, ultradeep sequencing (>20,000-fold coverage) was performed on IGH (~1650 nt) and nine other noncoding regions potentially targeted by AID (combined 9411 nt), including the 5' untranslated region of BCL2. Single-nucleotide variants (SNVs) were found in 12/12 FL specimens (median 136 SHMs and 53 aSHMs). The aSHM SNVs were associated with AID motifs (p < 0.0001). The number of SNVs at BCL2 varied widely among specimens and correlated with the number of SNVs at eight other potential aSHM sites. In contrast, SHM at IGH was not predictive of aSHM. Tumor heterogeneity is apparent from SNVs at low variant allele frequencies; the relative number of SNVs with variable allele frequency < 5% varied with clinical grade, indicating that tumor heterogeneity based on aSHM reflects a clinically meaningful parameter. These data suggest that genome-wide aSHM may be estimated from aSHM of BCL2 but not SHM of IGH. The results demonstrate a practical approach to the quantification of intratumoral genetic heterogeneity for clinical specimens.