Lucian Sulica

Objective: To evaluate the perceptual realism of synthetic videostroboscopic laryngeal images and determine the relationship between training duration and clinician classification accuracy. Methods: Synthetic images were generated using StyleGAN3 from two age‐stratified datasets: Dataset A (≥ 65 years) and Dataset B (< 65 years). A total of 114 clinicians evaluated a randomized, balanced‐controlled 36‐image survey drawn from a 144‐image pool containing real frames and five StyleGAN3 training intervals (5120–25,000 kimg). Accuracy was analyzed across image classes, experience levels, and viewing devices. Results: Clinicians demonstrated significantly higher mean accuracy identifying true clinical frames than synthetic generations (70.9% vs. 57.8%, p < 0.001). Overall survey accuracy was 59.9% ± 14.2%, with realism peaking at 20,120 kimg (44.8% accuracy). Accuracy dropped significantly between 5120 kimg (79.6%) and 10,120 kimg (59.1%, p < 0.001), with diminishing returns thereafter. No significant accuracy difference existed between Datasets A and B ( p = 0.27). The effect of specialty reached borderline significance ( p = 0.059), though limited by severe subgroup imbalances (e.g., n = 5 fellows, n = 2 residents). Computer users (63.9%) were significantly more accurate than phone users (56.2%, p = 0.014). A confounder analysis confirmed specialty and device choice were statistically independent ( p = 0.164). Conclusion: StyleGAN3 produces synthetic laryngeal images demonstrating high perceptual realism. The significant real‐versus‐synthetic accuracy gap highlights that while true frames remain distinct, synthetic generations achieve profound ambiguity. Perceptual realism plateaus early (10,120–20,120 kimg), suggesting moderate training durations can produce human‐perceived realism while reducing computational and environmental demands. Level of Evidence: N/A.

Objective: To develop and evaluate a deep learning object detection system for identifying vocal fold polyps in stroboscopic video frames using You Only Look Once ( YOLO ), and to assess the added benefit of temporal tracking on detection performance. Methods: A retrospective dataset of 12,742 annotated frames from 55 laryngoscopy video recordings was annotated with bounding boxes identifying vocal fold polyps. Pretrained YOLO11 and YOLO12 models were fine‐tuned to detect the polyps in the frames. A temporal tracking algorithm was further developed to propagate missed detections across adjacent frames. Results: YOLO12 outperformed YOLO11 across all metrics. On the hold‐out test set, YOLO12 reached a precision of 83.1% and an F 1 score of 67.6%, with a mean average precision at 0.5 (mAP@0.5) of 64.1%. By comparison, YOLO11 achieved a precision of 67.3% and an F 1 score of 56.2%, with a mAP@0.5 of 56.0%. Incorporating temporal tracking increased mAP@0.5 to 70.4% with YOLO 12, while maintaining a detection speed of 21.4 frames per second (fps), close to real time (30 fps). Conclusions: Using YOLO 12 for vocal fold polyp detection in stroboscopy was enhanced with temporal tracking, achieving a mAP@0.5 of 70.4% with near real time performance. These results demonstrate the potential of real‐time AI‐assisted detection of vocal fold lesions. Level of Evidence: 4.

Objective: To develop and validate a multimodal deep‐learning classifier trained on stroboscopic image, voice, and clinicodemographic data, differentiating between three different vocal fold (VF) states: healthy (HVF), unilateral paralysis (UVFP), and VF lesions, including benign and malignant pathologies. Methods: Patients with UVFP ( n = 54), VF lesions ( n = 42), and HVF ( n = 41) were retrospectively identified. Image frames and voice samples were extracted from stroboscopic videos. Clinicodemographic variables were collected from the electronic health record. Patient‐level data was independently divided into training (80%) and testing (20%). Visual features were extracted using a transformer DINOv2 and acoustic features were extracted using Librosa. All three feature modalities were combined using a custom multilayer perceptron. Unimodality models using only image or only voice data were trained for comparison. Accuracy and F 1 scores were used to validate the models. Results: On a hold‐out test set, the multimodal classifier demonstrated stronger performance (76.9% accuracy) compared to the image classifier (61.5% accuracy) and audio classifier (65.4% accuracy). On an external dataset, the multimodal classifier accuracy dropped to 45%, though still an improvement compared to accuracies of 42% and 31% for the video‐only and audio‐only modalities, respectively. Conclusions: In this proof‐of‐concept study, we successfully developed a multimodal dataset and classifier for VF pathology, demonstrating the potential of combining stroboscopic frames, voice and text data. The multimodal classifier achieved higher accuracy than the image‐only model and audio‐only models. Future models should validate these findings on larger datasets. Level of Evidence: 4

Objectives: Previous investigations have identified laryngeal diagnoses associated with age and sex, with professional singers experiencing a higher incidence of vocal injury relative to the rest of the population. However, no previous studies have examined injury patterns within the population of professional and student singers only. The objective of this study was to describe patterns of clinical presentations for laryngologic care among professional and student singers according to age, sex, and diagnosis. Methods: This retrospective case review collected primary diagnosis, age, sex, and self‐reported professional or student status for singers who presented with voice complaints from January 1, 2017 to May 1, 2024. Iatrogenic injuries, malignancy, and neurodegenerative disease processes were excluded. Results: Records of 1139 singers were included ( n = 1139, 416 male, 723 female, median age 30, age range 17–79). Presentations were predominantly professional female singers (63% F, 85% professional) between ages 20 and 35. Peaks in presentations were seen at ages 22, 26, and 30, in support of prior literature and clinical observations. A peak in both sexes was also seen at age 52, potentially coinciding with menopause and effects of senescence. There were strongly significant associations between age group and diagnosis and sex and diagnosis ( p < 0.001), especially bilateral midfold masses and female sex ( p < 0.001) across all age groups ( p < 0.001). Conclusion: These data inform counseling for professional singers and support screening exams at early career milestones to reduce healthcare burden and costs over the lifetime for this high‐risk population. Level of Evidence: 4.

Objectives/Hypotheses: To compare symptom characteristics between benign vocal fold lesions for insight into pathophysiology. Study Design: Cross‐sectional study. Methods: Adults who were diagnosed with a benign vocal fold lesion between June 2023 and June 2024 were included. Symptom features including acuity, duration, and inciting event were noted, along with prior history of voice problems, gender, and occupational voice load. Patients with cysts, polyps, and midfold lesions (fibrous mass, pseudocyst, non‐specific midfold mass) were compared to those with lesions known to be acute (edema, hemorrhage) and lesions known to be chronic (sulcus vergeture, sulcus vocalis, scar, Reinke's edema). Results: One hundred and fifty‐one adults (44M:106F:1Other; mean age 39.1 ± 15.8 years) were included. Midfold lesions (49.3%) and cysts (53.3%) had similar rates of gradual onset to known chronic lesions (50.0%), whereas polyps (30.4%) and known acute lesions had lower rates (8.3%, Exact Test = 16.6, p = 0.002). Similarly, midfold masses (37.3%), cysts (13.3%), and known chronic lesions (42.9%) had lower associations with an inciting event compared to polyps (52.2%) and known acute lesions (87.5%, Exact Test = 26.8, p < 0.001). While not statistically significant ( p = 0.054), cysts (46.7%), midfold lesions (57.3%), and known chronic lesions (78.6%) had a trend for more prior voice problems compared to polyps (30.4%) and known acute lesions (50.0%). Conclusions: These findings are consistent with models of phonotraumatic lesion development that suggest midfold lesions develop as a result of chronic cumulative damage, whereas polyps are more likely to develop acutely. Cysts may be structural lesions, as demonstrated by their lack of association with occupational voice use and chronic nature. Level of Evidence: 4.

Background: Vocal fold sulci are depressions that may be linear; sulcus vergeture (SVE) or like a pit; sulcus vocalis (SVO). The occurrence of SVE in family groups and the preponderance of reports from specific geographical areas suggest a hereditary etiology. Our objectives are to audit the possibility of a geographical variability between SVE and SVO and to study the patterns of presentation and treatment. Methods: Following ethics clearance, a 3‐year retrospective case‐series review of the demographics of vocal fold sulcus was conducted at voice centers in India, USA, Germany, and Australia. Sulcus cases were noted with details of associated lesions, sex, performers, age, and treatment given. The denominator was benign mucosal glottic lesions such as cysts and polyps. Results: Of a total of 351 cases of sulcus, 239 were SVE, with associated lesions in 56% (47% in SVO). A total of 50/239 SVE and 56/112 SVO were found in vocal performers ( p < 0.00001). A total of 160/239 SVE and 64/112 SVO were found in men. Sulcus of both types made up 20% of benign lesions in Sydney, 17% in Mumbai, which were significantly higher than New York (4%) and Germany (3.5%). SVE comprised 82% of cases (Mumbai), 72% (Sydney), 70% (New York) and 8.5% (Germany) with a significantly higher geographical variability than SVO. Conclusion: Both SVO and SVE were found predominantly in men. SVO was significantly more prevalent in performers, suggesting phonotrauma in its formation. SVE showed significant geographical variation, suggesting regional variation in its prevalence, supporting heredity in its formation, unlike SVO. Level of Evidence: 4

Objective: To develop and validate a deep‐learning classifier trained on voice data extracted from videolaryngostroboscopy recordings, differentiating between three different vocal fold (VF) states: healthy (HVF), unilateral paralysis (UVFP), and VF lesions, including benign and malignant pathologies. Methods: Patients with UVFP ( n = 105), VF lesions ( n = 63), and HVF ( n = 41) were retrospectively identified. Voice samples were extracted from stroboscopic videos (Pentax Laryngeal Strobe Model 9400), including sustained /i/ phonation, pitch glide, and /i/ sniff task. Extracted audio files were converted into Mel‐spectrograms. Voice samples were independently divided into training (80%), validation (10%), and test (10%) by patient. Pretrained ResNet18 models were trained to classify (1) HVF and pathological VF (lesions and UVFP), and (2) HVF, UVFP, and VF lesions. Both classifiers were further validated on an external dataset consisting of 12 UVFP, 13 VF lesions, and 15 HVF patients. Model performances were evaluated by accuracy and F1‐score. Results: When evaluated on a hold‐out test set, the binary classifier demonstrated stronger performance compared to the multi‐class classifier (accuracy 83% vs. 40%; F1‐score 0.90 vs. 0.36). When evaluated on an external dataset, the binary classifier achieved an accuracy of 63% and F1‐score of 0.48, compared to 35% and 0.25 for the multi‐class classifier. Conclusions: Deep‐learning classifiers differentiating HVF, UVFP, and VF lesions were developed using voice data from stroboscopic videos. Although healthy and pathological voice were differentiated with moderate accuracy, multi‐class classification lowered model performance. The model performed poorly on an external dataset. Voice captured in stroboscopic videos may have limited diagnostic value, though further studies are needed. Level of Evidence: 4 Laryngoscope , 135:2428–2436, 2025

Background: Unilateral vocal fold pseudocysts have been hypothesized to result from vocal fold paresis, but no explanation has been proposed for bilateral lesions. This study compares patients with unilateral to those with bilateral pseudocysts for insights into pathogenesis.Methods: Adults with unilateral and bilateral pseudocysts evaluated between 2018 and 2023 were retrospectively studied. Patient demographics, laryngeal stroboscopic findings, management strategies employed, and treatment outcomes were recorded. Fisher's exact and student's t‐tests were performed to assess unilateral and bilateral cohorts for differences.Results: One hundred ninety‐six patients (109 with bilateral and 87 with unilateral pseudocysts) were studied. The average age was 29 years (bilateral: 29 years, unilateral: 30 years; p = 0.3846). The groups differed with respect to sex (172 females: 105 bilateral, 67 unilateral; 24 males: 4 bilateral, 20 unilateral; p < 0.0001) and clinical diagnosis of paresis (bilateral: 13.8%, unilateral: 34.5%; p = 0.0010). Treatment was similar between cohorts for rates of voice therapy (bilateral: 67.0%, unilateral 63.2%; p = 0.6511) and surgery (bilateral: 12.8%, unilateral 17.2%; p = 0.4228). There were a total of six recurrences (bilateral: 2, unilateral: 4; p = 0.3898).Conclusions: Bilateral pseudocysts occur almost exclusively in women and with a relative absence of paresis. Unilateral pseudocysts are more likely to occur in the presence of paresis and in a significantly higher proportion of men. This suggests that unilateral and bilateral disease evolve in different clinical conditions, although they may share glottic insufficiency as a predisposing factor.Level of Evidence: III Laryngoscope, 2024

Objective(s): The objective of this study is to compare treatment outcomes for vocal fold polyps (VFPs) between operating room microlaryngoscopy and office‐based photoangiolysis with the potassium titanyl phosphate (KTP) laser.Methods: Prospective nonrandomized cohort study of patients with VFPs undergoing microlaryngoscopy (“OR group”) or KTP laser photoangiolysis (“KTP group”). Voice outcomes (patient‐reported outcome measures [Voice Handicap Index‐10 (VHI‐10) and Singing VHI‐10 (SVHI‐10)], auditory‐perceptual measures [Consensus Auditory‐Perceptual Evaluation of Voice (CAPE‐V)], videostroboscopic characteristics [Voice‐Vibratory Assessment of Laryngeal Imaging (VALI)], and acoustic and aerodynamic measures) were performed at baseline and regular intervals after intervention.Results: Forty‐four subjects (17 OR group, 27 KTP group) with VFPs were enrolled. Mean VHI‐10 significantly improved from baseline to each follow‐up interval in both groups, except for the 1–2‐week interval in the OR group. Mean SVHI‐10 improved for both groups at some intervals. Growth curve models and time‐to‐event analyses for patient‐reported outcomes did not differ between groups. There were significant improvements in all categories of auditory‐perceptual voice quality and some categories of videostroboscopic characteristics in both groups. No significant trends were identified in acoustic and aerodynamic measures. Improvements in most outcomes did not significantly differ between groups or based on polyp size. There were no major complications.Conclusions: Significant improvements in patient‐reported voice outcomes measures, auditory‐perceptual voice evaluation, and videostroboscopic characteristics occur following surgical treatment of vocal fold polyps with either microlaryngoscopy or office‐based KTP laser. Long‐term voice outcomes do not significantly differ between treatment modalities.Level of Evidence: 3 Laryngoscope, 134:S1–S20, 2024