Reproducibility of laryngeal force sensor measurements.

OBJECTIVE: The laryngeal force sensor (LFS) measures force during suspension microlaryngoscopy (SML) procedures, and has been previously shown to predict postoperative complications. Reproducibility of its measurements has not been described. STUDY DESIGN: Prospective cohort study. SETTING: Academic medical center. METHODS: 291 adult patients had force data collected from 2017 to 2021 during various SML procedures. 94 patients had passive LFS monitoring (surgeon blinded to intraoperative recordings) and 197 had active LFS monitoring (surgeon able to see LFS recordings). 27 of these patients had repeat procedures, with unique LFS metrics for each procedure. The 27 patients were divided into three groups. Group 1 had passive use for both procedures, group 2 had passive use for the first procedure and active use for the second, and group 3 had active use for both procedures. Force metrics from the two procedures were compared with a paired samples t-test. RESULTS: For airway dilation procedures and cancer resection procedures, average force variances were significantly lower with active versus passive use of the LFS. Group 1-no significant changes in maximum force (procedure 1 = 163.8 N, procedure 2 = 133.8 N, p = 0.324) or average force (procedure 1 = 93.6 N, procedure 2 = 78.3 N, p = 0.617). Group 2-maximum force dropped by 35 % between procedures 1 (219.2 N) and 2 (142.5 N), p = 0.013. Average force dropped by 42.5 % between procedures 1 (147.2 N) and 2 (84.6 N), p = 0.007. Group 3-no significant changes in maximum force (procedure 1 = 158.6 N, procedure 2 = 158.2 N, p = 0.986) or average force (procedure 1 = 94.2, procedure 2 = 81.8, p = 0.419). CONCLUSIONS: LFS measurements were reproducible for similar procedures in the same patient when the type of LFS monitoring was not a confounder.

Multiple roles for peptidylglycine α-amidating monooxygenase in the response to hypoxia.

The biosynthesis of many of the peptides involved in homeostatic control requires peptidylglycine α-amidating monooxygenase (PAM), an ancient, highly conserved copper- and ascorbate-dependent enzyme. Using the production of amidated chromogranin A to monitor PAM function in tumor cells, physiologically relevant levels of hypoxia were shown to inhibit this monooxygenase. The ability of primary pituitary cells exposed to hypoxic conditions for 4 h to produce amidated chromogranin A was similarly inhibited. The affinity of the purified monooxygenase for oxygen (Km = 99 ± 19 µM) was consistent with this result. The ability of PAM to alter secretory pathway behavior under normoxic conditions required its monooxygenase activity. Under normoxic conditions, hypoxia-inducible factor 1a levels in dense cultures of corticotrope tumor cells expressing high levels of PAM exceeded those in control cells; expression of inactive monooxygenase did not have this effect. The effects of hypoxia on levels of two PAM-regulated genes (activating transcription factor 3 [Atf3] and FK506 binding protein 2 [Fkbp2]) differed in cells expressing high versus low levels of PAM. Putative hypoxia response elements occur in both human and mouse PAM, and hPAM has consistently been identified as one of the genes upregulated in response to hypoxia. Expression of PAM is also known to alter gene expression. A quarter of the genes consistently upregulated in response to hypoxia were downregulated following increased expression of PAM. Taken together, our data suggest roles for PAM and amidated peptide secretion in the coordination of tissue-specific responses to hypoxia.

A pH-sensitive luminal His-cluster promotes interaction of PAM with V-ATPase along the secretory and endocytic pathways of peptidergic cells.

The biosynthetic and endocytic pathways of secretory cells are characterized by progressive luminal acidification, a process which is crucial for posttranslational modifications and membrane trafficking. This progressive fall in luminal pH is mainly achieved by the vacuolar-type-H+ ATPase (V-ATPase). V-ATPases are large, evolutionarily ancient rotary proton pumps that consist of a peripheral V1 complex, which hydrolyzes ATP, and an integral membrane V0 complex, which transports protons from the cytosol into the lumen. Upon sensing the desired luminal pH, V-ATPase activity is regulated by reversible dissociation of the complex into its V1 and V0 components. Molecular details of how intraluminal pH is sensed and transmitted to the cytosol are not fully understood. Peptidylglycine α-amidating mono-oxygenase (PAM; EC 1.14.17.3), a secretory pathway membrane enzyme which shares similar topology with two V-ATPase accessory proteins (Ac45 and prorenin receptor), has a pH-sensitive luminal linker region. Immunofluorescence and sucrose gradient analysis of peptidergic cells (AtT-20) identified distinct subcellular compartments exhibiting spatial co-occurrence of PAM and V-ATPase. In vitro binding assays demonstrated direct binding of the cytosolic domain of PAM to V1H. Blue native PAGE identified heterogeneous high-molecular weight complexes of PAM and V-ATPase. A PAM-1 mutant (PAM-1/H3A) with altered pH sensitivity had diminished ability to form high-molecular weight complexes. In addition, V-ATPase assembly status was altered in PAM-1/H3A expressing cells. Our analysis of the secretory and endocytic pathways of peptidergic cells supports the hypothesis that PAM serves as a luminal pH-sensor, regulating V-ATPase action by altering its assembly status.

Identifying bias in models that detect vocal fold paralysis from audio recordings using explainable machine learning and clinician ratings.

Introduction: Detecting voice disorders from voice recordings could allow for frequent, remote, and low-cost screening before costly clinical visits and a more invasive laryngoscopy examination. Our goals were to detect unilateral vocal fold paralysis (UVFP) from voice recordings using machine learning, to identify which acoustic variables were important for prediction to increase trust, and to determine model performance relative to clinician performance. Methods: Patients with confirmed UVFP through endoscopic examination (N=77) and controls with normal voices matched for age and sex (N=77) were included. Voice samples were elicited by reading the Rainbow Passage and sustaining phonation of the vowel "a". Four machine learning models of differing complexity were used. SHapley Additive explanations (SHAP) was used to identify important features. Results: The highest median bootstrapped ROC AUC score was 0.87 and beat clinician's performance (range: 0.74 - 0.81) based on the recordings. Recording durations were different between UVFP recordings and controls due to how that data was originally processed when storing, which we can show can classify both groups. And counterintuitively, many UVFP recordings had higher intensity than controls, when UVFP patients tend to have weaker voices, revealing a dataset-specific bias which we mitigate in an additional analysis. Conclusion: We demonstrate that recording biases in audio duration and intensity created dataset-specific differences between patients and controls, which models used to improve classification. Furthermore, clinician's ratings provide further evidence that patients were over-projecting their voices and being recorded at a higher amplitude signal than controls. Interestingly, after matching audio duration and removing variables associated with intensity in order to mitigate the biases, the models were able to achieve a similar high performance. We provide a set of recommendations to avoid bias when building and evaluating machine learning models for screening in laryngology.

Identifying bias in models that detect vocal fold paralysis from audio recordings using explainable machine learning and clinician ratings.

Detecting voice disorders from voice recordings could allow for frequent, remote, and low-cost screening before costly clinical visits and a more invasive laryngoscopy examination. Our goals were to detect unilateral vocal fold paralysis (UVFP) from voice recordings using machine learning, to identify which acoustic variables were important for prediction to increase trust, and to determine model performance relative to clinician performance. Patients with confirmed UVFP through endoscopic examination (N = 77) and controls with normal voices matched for age and sex (N = 77) were included. Voice samples were elicited by reading the Rainbow Passage and sustaining phonation of the vowel "a". Four machine learning models of differing complexity were used. SHapley Additive exPlanations (SHAP) was used to identify important features. The highest median bootstrapped ROC AUC score was 0.87 and beat clinician's performance (range: 0.74-0.81) based on the recordings. Recording durations were different between UVFP recordings and controls due to how that data was originally processed when storing, which we can show can classify both groups. And counterintuitively, many UVFP recordings had higher intensity than controls, when UVFP patients tend to have weaker voices, revealing a dataset-specific bias which we mitigate in an additional analysis. We demonstrate that recording biases in audio duration and intensity created dataset-specific differences between patients and controls, which models used to improve classification. Furthermore, clinician's ratings provide further evidence that patients were over-projecting their voices and being recorded at a higher amplitude signal than controls. Interestingly, after matching audio duration and removing variables associated with intensity in order to mitigate the biases, the models were able to achieve a similar high performance. We provide a set of recommendations to avoid bias when building and evaluating machine learning models for screening in laryngology.

Producing amyloid fibrils in vitro: A tool for studying AL amyloidosis.

Amyloid light-chain (AL) amyloidosis is the second most common form of systemic amyloidosis which is characterized by a high level of mortality and no effective treatment to remove fibril deposition. This disorder is caused by malfunctioning of B-cells resulting in production of abnormal protein fibrils composed of immunoglobulin light chain fragments that tend to deposit on various organs and tissues. AL amyloidosis is set apart from other forms of amyloidosis in that no specific sequences have been identified in the immunoglobulin light chains that are amyloid fibril formation causative and patient specific. This unusual feature hinders the therapeutic progress and requires either direct access to patient samples (which is not always possible) or a source of in vitro produced fibrils. While isolated reports of successful AL amyloid fibril formation from various patient-specific protein sequences can be found in literature, no systematic research on this topic was performed since 1999. In the present study we have developed a generalized approach to in vitro fibril production from various types of previously reported [[1], [2], [3]] amyloidogenic immunoglobulin light chains and their fragments. We describe the procedure from selection and generation of starting material, through finding of optimal assay conditions, to applying a panel of methods to confirm successful fibril formation. Procedure details are discussed in the light of the most recent findings and theories on amyloid fibril formation. The reported protocol produces high quality AL amyloid fibrils that can subsequently be used in the development of the much-needed amyloid-targeting diagnostic and therapeutic approaches.

Detecting schizophrenia with 3D structural brain MRI using deep learning.

Schizophrenia is a chronic neuropsychiatric disorder that causes distinct structural alterations within the brain. We hypothesize that deep learning applied to a structural neuroimaging dataset could detect disease-related alteration and improve classification and diagnostic accuracy. We tested this hypothesis using a single, widely available, and conventional T1-weighted MRI scan, from which we extracted the 3D whole-brain structure using standard post-processing methods. A deep learning model was then developed, optimized, and evaluated on three open datasets with T1-weighted MRI scans of patients with schizophrenia. Our proposed model outperformed the benchmark model, which was also trained with structural MR images using a 3D CNN architecture. Our model is capable of almost perfectly (area under the ROC curve = 0.987) distinguishing schizophrenia patients from healthy controls on unseen structural MRI scans. Regional analysis localized subcortical regions and ventricles as the most predictive brain regions. Subcortical structures serve a pivotal role in cognitive, affective, and social functions in humans, and structural abnormalities of these regions have been associated with schizophrenia. Our finding corroborates that schizophrenia is associated with widespread alterations in subcortical brain structure and the subcortical structural information provides prominent features in diagnostic classification. Together, these results further demonstrate the potential of deep learning to improve schizophrenia diagnosis and identify its structural neuroimaging signatures from a single, standard T1-weighted brain MRI.

Color Match Following Free Flap Surgery in Head and Neck Reconstruction: A Colorimetric and Aesthetic Analysis.

BACKGROUND: Many options for free tissue transfer have been described for head and neck reconstruction. While functional outcomes remain paramount, aesthetic considerations like color match can be equally consequential for patient quality of life. It is important to understand differences in color match based on flap donor site for head and neck reconstruction. METHODS: A retrospective review was performed of patients who underwent head and neck reconstruction with free tissue transfer at a tertiary care academic medical center between November 2012 and November 2020. Patients with documented pictures of their reconstruction and external skin paddles were considered. Patient demographics and surgery specific factors were recorded. Objective differences in color match were obtained by calculating the International Commission on Illumination Delta E 2000 (dE2000) score. Standard univariate descriptive statistics and multivariable statistical analyses were performed. RESULTS: Lateral arm, parascapular, and medial sural artery perforator (MSAP) free tissue transfer performed favorably compared to other donor sites, whereas anterolateral thigh flaps had the highest average dE2000 scores. Differences in dE2000 scores were mitigated by post-operative radiation to the flap site and with increasing time beyond 6 months post-operatively. CONCLUSIONS: We provide an objective assessment of external skin color match in patients undergoing free tissue transfer for head and neck cancer by donor site. MSAP, lateral arm, and parascapular free flaps performed well compared to traditional donor sites. These differences are more significant at the face and mandible when compared to the neck, but diminish 6 months after surgery and with post-operative radiation to the free flap skin paddle.

Surgical factors affecting regionally metastatic cutaneous squamous cell carcinoma to the parotid gland.

BACKGROUND: Understanding the impact of surgical treatment on regionally metastatic cutaneous squamous cell carcinoma (cSCC). METHODS: Retrospective series of 145 patients undergoing parotidectomy and neck dissection for regionally metastatic cSCC to the parotid. Overall survival (OS), disease-specific survival (DSS), and disease-free survival (DFS) analyzed over 3 years. Multivariate analysis was completed using Cox proportional hazard models. RESULTS: OS was 74.5%, DSS was 85.5% and DFS was 64.8%. On multivariate analysis, immune status (HR = 3.225[OS], 5.119[DSS], 2.071[DFS]) and lymphovascular invasion (HR = 2.380[OS], 5.237[DSS], 2.595[DFS]) were predictive for OS, DSS, and DFS. Margin status (HR = 2.296[OS], 2.499[DSS]) and ≥18 resected nodes (HR = 0.242[OS], 0.255[DSS]) were predictive of OS and DSS, while adjuvant therapy was predictive of DSS (p = 0.018). CONCLUSIONS: Immunosuppression and lymphovascular invasion portended worse outcomes in patients with metastatic cSCC to the parotid. Microscopically positive margins and <18 nodes resected are associated with worse OS and DSS, while patients receiving adjuvant therapy had improved DSS.

Design, development, and face validation of an intubation simulation device using real-time force data feedback.

Objectives: To develop a novel laryngoscope device capable of dynamically measuring force and torque measurements in real-time during intubation and to explore the efficacy of such a device through a face validation simulation. Methods: The torque sensor laryngoscope is designed for use during intubation and is modeled after a standard, single-use plastic laryngoscope. After device calibration, a face validation study was performed with intubation experts in the field. Quantitative data (intubation force metrics) and qualitative data (expert feedback on the device) were collected from three intubations using a Mac blade and three intubations with the Miller blade. Results: Three experts (two anesthesiologists and one otolaryngologist) participated in the study. The mean maximum force exerted with the Mac blade was 24.5 N (95% confidence interval [CI], 22.3-26.8). The average force exerted was 13.6 N (95% CI, 11.7-15.5). The average total suspension time was 13.1 s (95% CI, 10.4-15.8). The average total impulse was 164.6 N·s (95% CI, 147.9-181.4). The mean maximum force exerted with the Miller blade was 31.6 N (95% CI, 26.4-36.8). The average force exerted was 15.8 N (95% CI, 13.8-17.9). The average total suspension time was 11.3 s (95% CI, 9.9-12.6). The average total impulse was 216.2 N·s (95% CI, 186.5-245.9). The mean maximum force (p = .0265) and total impulse (p = .009) were significantly higher in the Miller blade trials than in the Mac blade trials. Survey results found that this device, while bulky, intubated similarly to standard-use models and has potential as an intubation teaching tool. Conclusion: The torque sensor laryngoscope can measure and display real-time intubation force metrics for multiple laryngoscope blades. Initial validation studies showed a significantly lower maximum force and total impulse when intubating with the Mac blade than with the Miller blade. Face validation survey results were positive and suggested the potential for this device as a teaching tool. Level of Evidence: Level 5.

Clinical, Radiologic, and Endolaryngeal Findings in Laryngeal Fractures: A 15-Year Case Series.

OBJECTIVE: Laryngeal fractures are rare injuries; recent data describing these injuries and associated examination findings are limited. This study aims to describe injury etiology and outcomes associated with laryngeal fractures. STUDY DESIGN: Retrospective case series. SETTING: Academic tertiary center. METHODS: Patients with laryngeal fractures from 2005 to 2020 were identified in a retrospective chart review. Patient demographics, injury mechanisms, management, and voice outcomes were examined. Fracture type, radiologic, and endolaryngeal examination findings were analyzed for associations between fracture etiology and examination characteristics. RESULTS: Laryngeal fractures most commonly occurred at the thyroid cartilage. Fractures were most commonly due to sport-related injuries. Mechanism of injury was not associated with specific radiologic or endolaryngeal findings. Mechanism of injury was additionally not significantly associated with the need for intubation, surgical intervention, or tracheotomy. Fracture location was significantly associated with intubation requirement (P = .015), with 40% of patients with concomitant thyroid and cricoid fractures requiring intubation. Mechanism of injury significantly correlated with dysphonia at follow-up (P = .033). Mechanism of injury, fracture location, and surgical management were not associated with increased vocal fold injury or dysphonia. CONCLUSION: There are no significant correlations between injury mechanism and fracture location, characteristics, radiologic findings, or endolaryngeal findings. These features emphasize the importance of a thorough and comprehensive laryngeal examination.

Improving across-dataset brain tissue segmentation for MRI imaging using transformer.

Brain tissue segmentation has demonstrated great utility in quantifying MRI data by serving as a precursor to further post-processing analysis. However, manual segmentation is highly labor-intensive, and automated approaches, including convolutional neural networks (CNNs), have struggled to generalize well due to properties inherent to MRI acquisition, leaving a great need for an effective segmentation tool. This study introduces a novel CNN-Transformer hybrid architecture designed to improve brain tissue segmentation by taking advantage of the increased performance and generality conferred by Transformers for 3D medical image segmentation tasks. We first demonstrate the superior performance of our model on various T1w MRI datasets. Then, we rigorously validate our model's generality applied across four multi-site T1w MRI datasets, covering different vendors, field strengths, scan parameters, and neuropsychiatric conditions. Finally, we highlight the reliability of our model on test-retest scans taken in different time points. In all situations, our model achieved the greatest generality and reliability compared to the benchmarks. As such, our method is inherently robust and can serve as a valuable tool for brain related T1w MRI studies. The code for the TABS network is available at: https://github.com/raovish6/TABS.

Laryngeal Subsite Analysis of Granulomatosis With Polyangiitis (Wegener's).

OBJECTIVE: To analyze specific intralaryngeal findings associated with granulomatosis with polyangiitis (GPA). STUDY DESIGN: Retrospective chart review. SETTING: Tertiary referral center. METHODS: A retrospective chart review was performed on all patients diagnosed with GPA who were evaluated at the laryngology division of Massachusetts Eye and Ear Infirmary between January 2006 and September 2019. RESULTS: Forty-four patients (14 male, 30 female) were evaluated for laryngeal pathology. The mean age at onset was 48 years. Nine patients (21%) were identified with only vocal fold disease, 11 (25%) with subglottic disease, and 8 (18%) with disease at the glottis and subglottis (transglottic). The remaining 16 patients (36%) had a normal airway upon examination although they presented with laryngeal symptoms. Patients with glottic disease had statistically significantly lower voice-related quality of life scores than patients with isolated subglottic stenosis. CONCLUSIONS: Although laryngeal manifestations of GPA is often described as a subglottic disease presenting with respiratory symptoms, subsite analysis show that only 25% of patients had subglottic disease alone, with similar rates of glottic disease alone. Laryngeal subsites have different epithelial mucosa, function, and physiology, and understanding the specific sites of involvement will determine symptoms and enable better analysis of the underlying mechanisms of disease. Glottic disease is associated with a reduction in vocal fold motion and voice changes. Subglottic involvement presents more frequently with airway symptoms. Further research is necessary to better define the specific regions of laryngeal involvement in patients diagnosed with GPA.

Laryngeal Force Sensor for Suspension Microlaryngoscopy: A Prospective Controlled Trial.

OBJECTIVES: The laryngeal force sensor (LFS) provides real-time force data for suspension microlaryngoscopy. This study investigates whether active use of the LFS can prevent the development of complications. STUDY DESIGN: Prospective controlled trial. SETTING: Academic tertiary center. METHODS: The LFS and custom software were developed to track intraoperative force metrics. A consecutive series of 100 patients had force data collected with operating surgeons blinded to intraoperative readings. The subsequent 100 patients had surgeons actively use the LFS monitoring system. Patients were prospectively enrolled, completing pre- and postoperative surveys to assess the development of tongue pain, paresthesia, paresis, dysgeusia, or dysphagia. RESULTS: On univariate analysis, the active monitoring group had lower total impulse (P < .001) and fewer extralaryngeal complications (P < .01). On multiple logistic regression, maximum force (odds ratio [OR], 1.08; 95% CI, 1.01-1.16; P = .02) was a significant predictive variable for the development of postoperative complications. Similarly, active LFS monitoring showed a 29.1% (95% CI, 15.7%-42.4%; P < .001) decrease in the likelihood of developing postoperative complications. These effects persisted at the first postoperative visit for maximum force (P = .04) and active LFS monitoring (P = .01). Maximum force (OR, 1.11; 95% CI, 1.04-1.18; P < .01) and active LFS monitoring (16.6%; 95% CI, 2.7%-30.5%; P = .02) were also predictive for the development of an abnormal 10-item Eating Assessment Tool score. These effects also persisted at the first postoperative visit for maximum force (P = .01) and active LFS monitoring (P = .01). CONCLUSION: Maximum force is predictive of the development postoperative complications. Active monitoring with the LFS is able to mitigate these forces and prevent postoperative complications. LEVEL OF EVIDENCE: 2.

Force Metrics and Suspension Times for Microlaryngoscopy Procedures.

OBJECTIVE: To determine the difference in force metrics measured by the laryngeal force sensor for various suspension microlaryngoscopy (SML) procedures and their perioperative narcotic requirements. STUDY DESIGN: Prospective observational study. SETTING: Academic tertiary center. METHODS: The laryngeal force sensoris a force sensor designed for SML procedures. Prospectively enrolled patients had dynamic recordings of maximum force, average force, suspension time, and total impulse. Procedures were grouped into excision of striking zone lesions, nonstriking zone lesions, endoscopic cancer surgery with margin control, and airway dilation. Narcotic administration in the intraoperative period and postanesthesia care unit was also recorded and converted into IV morphine equivalents. Surgeons were blinded to the force recordings during surgery to prevent operator bias. RESULTS: In total, 110 patients completed the study. There was no significant difference in average force across different procedures, however, a significant difference was seen for maximum force (P = 0.025), suspension time (P < 0.001), and total impulse (P = 0.002). The highest values were seen for endoscopic cancer surgeries with margin control with a mean maximum force of 49.4 lbf (95%CI, 37.1-61.7), mean suspension time of 60.2 minutes (95%CI, 40.5-79.9), and mean total impulse of 31.3 ton*s (95%CI, 15.2-47.3). A significant difference (P < 0.01) in perioperative narcotic requirements was also seen, with endoscopic cancer surgery cases having the highest requirements at 27.6 mg of ME (95%CI, 16.1-39.2 mg). CONCLUSION: Significant differences in force metrics exist between various SML procedures. Endoscopic cancer surgery is associated with higher force metrics and perioperative narcotic requirements.

Role of Kalirin and mouse strain in retention of spatial memory training in an Alzheimer's disease model mouse line.

Nontransgenic and 3xTG transgenic mice, which express mutant transgenes encoding human amyloid precursor protein (hAPP) along with Alzheimer's disease-associated versions of hTau and a presenilin mutation, acquired the Barnes Maze escape task equivalently at 3-9 months of age. Although nontransgenics retested at 6 and 9 months acquired the escape task more quickly than naïve mice, 3xTG mice did not. Deficits in Kalirin, a multidomain protein scaffold and guanine nucleotide exchange factor that regulates dendritic spines, has been proposed as a contributor to the cognitive decline observed in Alzheimer's disease. To test whether deficits in Kalirin might amplify deficits in 3xTG mice, mice heterozygous/hemizygous for Kalirin and the 3xTG transgenes were generated. Mouse strain, age and sex affected cortical expression of key proteins. hAPP levels in 3xTG mice increased total APP levels at all ages. Kalirin expression showed strong sex-dependent expression in C57 but not B6129 mice. Decreasing Kalirin levels to half had no effect on Barnes Maze task acquisition or retraining in 3xTG hemizygous mice.