Andy S. Ding
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Andy S. Ding has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
Francis X. Creighton
Department of Otolaryngology-Head and Neck Surgery, Johns Hopkins University School of Medicine, Baltimore, USA
7 shared publicationsRussell H. Taylor
Department of Computer Science, Johns Hopkins University, Baltimore, USA
6 shared publicationsZhaoshuo Li
Department of Computer Science Johns Hopkins University Baltimore Maryland USA
5 shared publicationsAlexander Lu
Department of Otolaryngology–Head and Neck Surgery Johns Hopkins University School of Medicine Baltimore Maryland USA
4 shared publicationsDeepa Galaiya
Department of Otolaryngology–Head and Neck Surgery Johns Hopkins University School of Medicine Baltimore Maryland USA
4 shared publicationsJeffrey H. Siewerdsen
Department of Computer Science Johns Hopkins University Baltimore Maryland USA
4 shared publications
Latest Publications
- 26 October 2026
- Published by Wolters Kluwer Health in Otology & Neurotology
- Vol. 47 (10) , 1213-1223
The Evolution and Application of Robotic Systems in Neurotologic Surgery: A Scoping Review
Objective: To map the technological evolution and clinical feasibility of robotic systems in neurotologic and lateral skull-base surgery, identifying bottlenecks limiting widespread clinical translation. Databases Reviewed: A systematic search was conducted across Embase, Scopus, PubMed/MEDLINE, Web of Science, and the Cochrane Central Register of Controlled Trials. Review Methods: A Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews–compliant scoping review was conducted. Inclusion criteria comprised peer-reviewed, English-language original research evaluating the feasibility, accuracy, and clinical performance of robotic systems in human subjects or cadaveric models undergoing neurotologic procedures. Data regarding study characteristics, spatial accuracy, system kinematics, and safety outcomes were extracted for qualitative synthesis. Results: Fifty-seven studies met the inclusion criteria. Notably, the yielded literature was skewed toward otologic applications rather than broader lateral skull-base resections. Two dominant clinical applications emerged: image-guided bone milling (eg, direct cochlear access) and robotic-assisted microinstrumentation (eg, cochlear implant electrode insertion). Robotic milling systems consistently achieved submillimetric accuracy (frequently <0.15 mm at the facial recess) by integrating multilayer safety architectures, including force-torque sensing and continuous electromyography. For electrode insertion, motorized actuation standardized ultra-slow velocities (0.1 to 0.25 mm/s), significantly reducing scalar translocations and maximum insertion forces compared with manual techniques. Primary barriers to broad adoption include prolonged operative setup times, reliance on invasive fiducial registration, and an absence of high-fidelity active haptic feedback. Conclusion: Robotic systems in neurotology successfully mitigate physiological human tremor and enforce strict kinematic safety boundaries, enabling advanced, minimally invasive approaches. Widespread clinical integration will depend on transitioning to markerless registration, validating in vivo dynamic models, and overcoming economic and workflow barriers.
- 26 October 2026
- Published by Wolters Kluwer Health in Otology & Neurotology
- Vol. 47 (10) , 1213-1223
The Evolution and Application of Robotic Systems in Neurotologic Surgery: A Scoping Review
Objective: To map the technological evolution and clinical feasibility of robotic systems in neurotologic and lateral skull-base surgery, identifying bottlenecks limiting widespread clinical translation. Databases Reviewed: A systematic search was conducted across Embase, Scopus, PubMed/MEDLINE, Web of Science, and the Cochrane Central Register of Controlled Trials. Review Methods: A Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews–compliant scoping review was conducted. Inclusion criteria comprised peer-reviewed, English-language original research evaluating the feasibility, accuracy, and clinical performance of robotic systems in human subjects or cadaveric models undergoing neurotologic procedures. Data regarding study characteristics, spatial accuracy, system kinematics, and safety outcomes were extracted for qualitative synthesis. Results: Fifty-seven studies met the inclusion criteria. Notably, the yielded literature was skewed toward otologic applications rather than broader lateral skull-base resections. Two dominant clinical applications emerged: image-guided bone milling (eg, direct cochlear access) and robotic-assisted microinstrumentation (eg, cochlear implant electrode insertion). Robotic milling systems consistently achieved submillimetric accuracy (frequently <0.15 mm at the facial recess) by integrating multilayer safety architectures, including force-torque sensing and continuous electromyography. For electrode insertion, motorized actuation standardized ultra-slow velocities (0.1 to 0.25 mm/s), significantly reducing scalar translocations and maximum insertion forces compared with manual techniques. Primary barriers to broad adoption include prolonged operative setup times, reliance on invasive fiducial registration, and an absence of high-fidelity active haptic feedback. Conclusion: Robotic systems in neurotology successfully mitigate physiological human tremor and enforce strict kinematic safety boundaries, enabling advanced, minimally invasive approaches. Widespread clinical integration will depend on transitioning to markerless registration, validating in vivo dynamic models, and overcoming economic and workflow barriers.
- Open Access
- 20 November 2025
- Published by JMIR Publications Inc. in JMIR Human Factors
- Vol. 12 (1) , e64224
Factors Shaping Public Perceptions of a Range of Robotic Technologies in Surgery: Cross-Sectional Web-Based Survey
Background: Within the surgical field, there has been an evolution in the application of robotic technology. Fully automatic robotic systems and augmented visualization tools are being introduced and may eventually replace existing surgical extenders such as the da Vinci surgical system. The literature on public perception of robotic surgery is growing, though specific drivers of these attitudes remain under investigation. Objective: The aim of this study is to investigate the underlying motivators of public perceptions toward robotic surgeries with varying levels of autonomy through a formal technology acceptance model. Methods: An online survey was distributed via the Amazon Mechanical Turk platform. Survey participants were provided definitions of a continuum of robotic technologies: robotic surgical extenders (technology without independent actions), semiautonomous robotic surgery (technology that provides guidance to the surgeon and requires surgeon input), and fully autonomous robotic surgery (technology that performs tasks autonomously without direct human interaction). The survey assessed overall attitudes toward each application of robotic technology in surgery and included questions delineating specific receptivity based on (1) perceived usefulness, (2) social risk, (3) time risk, (4) personal risk, and (5) reliability. A technology acceptance model was built to identify associations between these factors and overall attitudes toward robotic and semiautonomous surgeries. Results: A total of 1221 survey responses were recorded (mean age 38, SD 12 y; females: n=635, 52%). Individuals were more willing to accept robotic surgical extenders and semiautonomous robotic surgery compared to autonomous robotic surgery. Higher levels of education and better self-reported health were correlated with more positive attitudes toward autonomous robotic surgery. Perceptions of these technologies were not associated with age, gender, or income. Overall, attitudes toward robotic technologies in surgery were driven by views on the reliability, safety, and efficiency of the procedures. There was less concern regarding time risk and social risk associated with robotic and semirobotic surgeries. Conclusions: The public is more accepting of semiautonomous surgery and surgical extenders than fully autonomous surgery. General perceptions of the reliability, safety, and efficiency of these technologies drive variations in attitude. Time and social risk do not appear to have a significant impact on receptivity. Understanding these perspectives can help guide education within an advancing surgical field.
- 20 November 2025
- Published by JMIR Publications Inc. in JMIR Human Factors
- Vol. 12 (1) , e64224
Open AccessFactors Shaping Public Perceptions of a Range of Robotic Technologies in Surgery: Cross-Sectional Web-Based Survey
Background: Within the surgical field, there has been an evolution in the application of robotic technology. Fully automatic robotic systems and augmented visualization tools are being introduced and may eventually replace existing surgical extenders such as the da Vinci surgical system. The literature on public perception of robotic surgery is growing, though specific drivers of these attitudes remain under investigation. Objective: The aim of this study is to investigate the underlying motivators of public perceptions toward robotic surgeries with varying levels of autonomy through a formal technology acceptance model. Methods: An online survey was distributed via the Amazon Mechanical Turk platform. Survey participants were provided definitions of a continuum of robotic technologies: robotic surgical extenders (technology without independent actions), semiautonomous robotic surgery (technology that provides guidance to the surgeon and requires surgeon input), and fully autonomous robotic surgery (technology that performs tasks autonomously without direct human interaction). The survey assessed overall attitudes toward each application of robotic technology in surgery and included questions delineating specific receptivity based on (1) perceived usefulness, (2) social risk, (3) time risk, (4) personal risk, and (5) reliability. A technology acceptance model was built to identify associations between these factors and overall attitudes toward robotic and semiautonomous surgeries. Results: A total of 1221 survey responses were recorded (mean age 38, SD 12 y; females: n=635, 52%). Individuals were more willing to accept robotic surgical extenders and semiautonomous robotic surgery compared to autonomous robotic surgery. Higher levels of education and better self-reported health were correlated with more positive attitudes toward autonomous robotic surgery. Perceptions of these technologies were not associated with age, gender, or income. Overall, attitudes toward robotic technologies in surgery were driven by views on the reliability, safety, and efficiency of the procedures. There was less concern regarding time risk and social risk associated with robotic and semirobotic surgeries. Conclusions: The public is more accepting of semiautonomous surgery and surgical extenders than fully autonomous surgery. General perceptions of the reliability, safety, and efficiency of these technologies drive variations in attitude. Time and social risk do not appear to have a significant impact on receptivity. Understanding these perspectives can help guide education within an advancing surgical field.
- 29 May 2025
- Published by Springer Nature in International Journal of Computer Assisted Radiology and Surgery
- p. 1-11
Manual and automated facial de-identification techniques for patient imaging with preservation of sinonasal anatomy
Purpose Facial recognition of reconstructed computed tomography (CT) scans poses patient privacy risks, necessitating reliable facial de-identification methods. Current methods obscure sinuses, turbinates, and other anatomy relevant for otolaryngology. We present a facial de-identification method that preserves these structures, along with two automated workflows for large-volume datasets. Methods A total of 20 adult head CTs from the New Mexico Decedent Image Database were included. Using 3D Slicer, a seed-growing technique was performed to label the skin around the face. This label was dilated bidirectionally to form a 6-mm mask that obscures facial features. This technique was then automated using: (1) segmentation propagation that deforms an atlas head CT and corresponding mask to match other scans and (2) a deep learning model (nnU-Net). Accuracy of these methods against manually generated masks was evaluated with Dice scores and modified Hausdorff distances (mHDs). Results Manual de-identification resulted in facial match rates of 45.0% (zero-fill), 37.5% (deletion), and 32.5% (re-face). Dice scores for automated face masks using segmentation propagation and nnU-Net were 0.667 ± 0.109 and 0.860 ± 0.029, respectively, with mHDs of 4.31 ± 3.04 mm and 1.55 ± 0.71 mm. Match rates after de-identification using segmentation propagation (zero-fill: 42.5%; deletion: 40.0%; re-face: 35.0%) and nnU-Net (zero-fill: 42.5%; deletion: 35.0%; re-face: 30.0%) were comparable to manual masks. Conclusion We present a simple facial de-identification approach for head CTs, as well as automated methods for large-scale implementation. These techniques show promise for preventing patient identification while preserving underlying sinonasal anatomy, but further studies using live patient photographs are necessary to fully validate its effectiveness.
- 29 May 2025
- Published by Springer Nature in International Journal of Computer Assisted Radiology and Surgery
- p. 1-11
Manual and automated facial de-identification techniques for patient imaging with preservation of sinonasal anatomy
Purpose Facial recognition of reconstructed computed tomography (CT) scans poses patient privacy risks, necessitating reliable facial de-identification methods. Current methods obscure sinuses, turbinates, and other anatomy relevant for otolaryngology. We present a facial de-identification method that preserves these structures, along with two automated workflows for large-volume datasets. Methods A total of 20 adult head CTs from the New Mexico Decedent Image Database were included. Using 3D Slicer, a seed-growing technique was performed to label the skin around the face. This label was dilated bidirectionally to form a 6-mm mask that obscures facial features. This technique was then automated using: (1) segmentation propagation that deforms an atlas head CT and corresponding mask to match other scans and (2) a deep learning model (nnU-Net). Accuracy of these methods against manually generated masks was evaluated with Dice scores and modified Hausdorff distances (mHDs). Results Manual de-identification resulted in facial match rates of 45.0% (zero-fill), 37.5% (deletion), and 32.5% (re-face). Dice scores for automated face masks using segmentation propagation and nnU-Net were 0.667 ± 0.109 and 0.860 ± 0.029, respectively, with mHDs of 4.31 ± 3.04 mm and 1.55 ± 0.71 mm. Match rates after de-identification using segmentation propagation (zero-fill: 42.5%; deletion: 40.0%; re-face: 35.0%) and nnU-Net (zero-fill: 42.5%; deletion: 35.0%; re-face: 30.0%) were comparable to manual masks. Conclusion We present a simple facial de-identification approach for head CTs, as well as automated methods for large-scale implementation. These techniques show promise for preventing patient identification while preserving underlying sinonasal anatomy, but further studies using live patient photographs are necessary to fully validate its effectiveness.
- 25 April 2025
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 173 (1) , 126-133
A Novel Natural Language Processing Model for Triaging Head and Neck Patient Appointments
Objective: Inaccurate patient triage contributes to suboptimal clinical capacity management and delays in patient care, which in cancer patients may significantly increase morbidity and mortality. We developed a natural language processing (NLP) model as an adjunctive tool for head and neck (H&N) patient triage workflows. This study assesses the model's ability to categorize and triage patient appointments based on available documentation.Study Design: A retrospective cohort study.Setting: An academic institution.Methods: A total of 83 new patients seeing an H&N surgeon from January to April 2024 with at least 1 referral record (clinic note, imaging, or pathology report) available were included in this study. Referral clinic, imaging, and pathology reports were entered into the NLP model to predict pathology type (non‐endocrine H&N neoplasm, thyroid, parathyroid, and benign lesions), malignancy risk, and appointment urgency. The gold standard was the final diagnosis from pathology reports or surgeons' clinic notes.Results: The NLP model achieved an accuracy of 81.9% for pathology type and 86.8% for urgency level. Sensitivity was high for non‐endocrine H&N neoplasms (88.9%), thyroid pathology (88.9%), and parathyroid pathology (100%), although lower for benign lesions (67.9%). Specificity was 86.8% for non‐endocrine H&N neoplasms, 91.9% for thyroid pathology, 97.6% for parathyroid pathology, and 96.4% for benign lesions. Prediction of appointment urgency achieved a Matthews correlation coefficient of 0.698, reflecting strong predictive performance.Conclusion: This novel NLP model demonstrated robust performance characteristics for predicting H&N diagnoses based on referring documents and excelled at identifying patients requiring urgent care based on malignancy risk. This tool may help H&N practice coordinators screen referrals, potentially optimizing patient care.
- 25 April 2025
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 173 (1) , 126-133
A Novel Natural Language Processing Model for Triaging Head and Neck Patient Appointments
Objective: Inaccurate patient triage contributes to suboptimal clinical capacity management and delays in patient care, which in cancer patients may significantly increase morbidity and mortality. We developed a natural language processing (NLP) model as an adjunctive tool for head and neck (H&N) patient triage workflows. This study assesses the model's ability to categorize and triage patient appointments based on available documentation.Study Design: A retrospective cohort study.Setting: An academic institution.Methods: A total of 83 new patients seeing an H&N surgeon from January to April 2024 with at least 1 referral record (clinic note, imaging, or pathology report) available were included in this study. Referral clinic, imaging, and pathology reports were entered into the NLP model to predict pathology type (non‐endocrine H&N neoplasm, thyroid, parathyroid, and benign lesions), malignancy risk, and appointment urgency. The gold standard was the final diagnosis from pathology reports or surgeons' clinic notes.Results: The NLP model achieved an accuracy of 81.9% for pathology type and 86.8% for urgency level. Sensitivity was high for non‐endocrine H&N neoplasms (88.9%), thyroid pathology (88.9%), and parathyroid pathology (100%), although lower for benign lesions (67.9%). Specificity was 86.8% for non‐endocrine H&N neoplasms, 91.9% for thyroid pathology, 97.6% for parathyroid pathology, and 96.4% for benign lesions. Prediction of appointment urgency achieved a Matthews correlation coefficient of 0.698, reflecting strong predictive performance.Conclusion: This novel NLP model demonstrated robust performance characteristics for predicting H&N diagnoses based on referring documents and excelled at identifying patients requiring urgent care based on malignancy risk. This tool may help H&N practice coordinators screen referrals, potentially optimizing patient care.
- Open Access
- 15 March 2024
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 171 (1) , 45-53
Long‐term Quality of Life After Thyroidectomy: Transoral Endoscopic Thyroidectomy Vestibular Approach Versus Transcervical Approach
Objective: To compare long‐term health‐related quality of life (HRQOL) after Transoral Endoscopic Thyroidectomy Vestibular Approach (TOETVA) and transcervical approach (TCA) thyroidectomy.Study Design: Prospective cohort study.Setting: Tertiary referral center.Methods: A web‐based survey was distributed to patients at our institution who met the criteria for TOETVA and underwent thyroidectomy by TOETVA or TCA between August 2017 and October 2021. All survey participants were at least 6 months postsurgery. Minors, non‐English speakers, and patients who received concomitant neck dissection or reoperative thyroidectomy were excluded from the study. The survey assessed quality of life through 4 standardized instruments: the Dermatology Life Quality Index (DLQI), the Eating Assessment Tool (EAT‐10), the Voice Handicap Index (VHI‐10), and the Short Form Health Survey (SF‐36).Results: A total of 108 TOETVA and 129 TCA patients were included in the study. The median age of respondents was 44 (36, 54; 25th, 75th percentile) years and median time from surgery to survey was 35 (22, 45; 25th, 75th percentile) months. TOETVA group DLQI (0.63 vs 0.99; P = .17), VHI‐10 (1.94 vs 1.67; P = .35), EAT‐10 (2.14 vs 2.32; P = .29), SF‐36 physical component (52.25 vs 51.00; P = .25), and SF‐36 mental component (47.74 vs 47.29; P = .87) scores were all similar to those of the TCA group. Scrutinizing specific DLQI questions, individuals in the TOETVA group were less self‐conscious of their skin as compared to the TCA group (Q2; 0.08 vs 0.26, P = .03).Conclusion: Long‐term HRQOL after TOETVA is similar to TCA, with significantly lower skin‐related self‐consciousness.
- 15 March 2024
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 171 (1) , 45-53
Open AccessLong‐term Quality of Life After Thyroidectomy: Transoral Endoscopic Thyroidectomy Vestibular Approach Versus Transcervical Approach
Objective: To compare long‐term health‐related quality of life (HRQOL) after Transoral Endoscopic Thyroidectomy Vestibular Approach (TOETVA) and transcervical approach (TCA) thyroidectomy.Study Design: Prospective cohort study.Setting: Tertiary referral center.Methods: A web‐based survey was distributed to patients at our institution who met the criteria for TOETVA and underwent thyroidectomy by TOETVA or TCA between August 2017 and October 2021. All survey participants were at least 6 months postsurgery. Minors, non‐English speakers, and patients who received concomitant neck dissection or reoperative thyroidectomy were excluded from the study. The survey assessed quality of life through 4 standardized instruments: the Dermatology Life Quality Index (DLQI), the Eating Assessment Tool (EAT‐10), the Voice Handicap Index (VHI‐10), and the Short Form Health Survey (SF‐36).Results: A total of 108 TOETVA and 129 TCA patients were included in the study. The median age of respondents was 44 (36, 54; 25th, 75th percentile) years and median time from surgery to survey was 35 (22, 45; 25th, 75th percentile) months. TOETVA group DLQI (0.63 vs 0.99; P = .17), VHI‐10 (1.94 vs 1.67; P = .35), EAT‐10 (2.14 vs 2.32; P = .29), SF‐36 physical component (52.25 vs 51.00; P = .25), and SF‐36 mental component (47.74 vs 47.29; P = .87) scores were all similar to those of the TCA group. Scrutinizing specific DLQI questions, individuals in the TOETVA group were less self‐conscious of their skin as compared to the TCA group (Q2; 0.08 vs 0.26, P = .03).Conclusion: Long‐term HRQOL after TOETVA is similar to TCA, with significantly lower skin‐related self‐consciousness.
- 12 September 2023
- Published by Wiley in The Laryngoscope
- Vol. 134 (4) , 1847-1853
Longitudinal Performance of Cochlear Implants in Neurofibromatosis Type 2
Objective: Cochlear implants (CIs) are a well‐established treatment modality for hearing loss due to neurofibromatosis type 2 (NF2). Our aim is to investigate variables that affect longitudinal performance of CIs among patients with NF2.Study Design: Retrospective review at a single academic institution consisting of patients who have received cochlear implants following hearing loss due to NF2.Methods: The primary outcome examined was CI disuse or explantation. Associated clinical and surgical variables were analyzed using descriptive statistics. These included postoperative pure tone average (PTA) at 500, 1000, and 2000 Hz, tumor size, previous surgery, and comorbid depression.Results: A total of 12 patients and 14 cochlear implants received at our institution from 2001 to 2022 were included. Notably, 35.7% of CIs (5 out of 14 cases) resulted in disuse or explantation. The average interval until explant was 9.4 years (range 3–14 years). In explanted CI cases, 20% had previous surgery and 80% had a diagnosis of comorbid depression as compared to 22.2% and 22.2%, respectively, in intact CI cases. Maximum tumor diameter was the only variable found to impact CI usage outcome (p = 0.028). Long‐term data showed that on average, patients benefit from 13.85 years of CI utility and a maximum PTA improvement of 45.0 ± 29.0 dB.Conclusion: Despite the recurrent nature of NF2, patients continue to receive audiological benefit from cochlear implants. We found that larger tumor size may be associated with longitudinal CI failure.Level of Evidence: 4 Laryngoscope, 134:1847–1853, 2024
- 12 September 2023
- Published by Wiley in The Laryngoscope
- Vol. 134 (4) , 1847-1853
Longitudinal Performance of Cochlear Implants in Neurofibromatosis Type 2
Objective: Cochlear implants (CIs) are a well‐established treatment modality for hearing loss due to neurofibromatosis type 2 (NF2). Our aim is to investigate variables that affect longitudinal performance of CIs among patients with NF2.Study Design: Retrospective review at a single academic institution consisting of patients who have received cochlear implants following hearing loss due to NF2.Methods: The primary outcome examined was CI disuse or explantation. Associated clinical and surgical variables were analyzed using descriptive statistics. These included postoperative pure tone average (PTA) at 500, 1000, and 2000 Hz, tumor size, previous surgery, and comorbid depression.Results: A total of 12 patients and 14 cochlear implants received at our institution from 2001 to 2022 were included. Notably, 35.7% of CIs (5 out of 14 cases) resulted in disuse or explantation. The average interval until explant was 9.4 years (range 3–14 years). In explanted CI cases, 20% had previous surgery and 80% had a diagnosis of comorbid depression as compared to 22.2% and 22.2%, respectively, in intact CI cases. Maximum tumor diameter was the only variable found to impact CI usage outcome (p = 0.028). Long‐term data showed that on average, patients benefit from 13.85 years of CI utility and a maximum PTA improvement of 45.0 ± 29.0 dB.Conclusion: Despite the recurrent nature of NF2, patients continue to receive audiological benefit from cochlear implants. We found that larger tumor size may be associated with longitudinal CI failure.Level of Evidence: 4 Laryngoscope, 134:1847–1853, 2024
- Open Access
- 17 May 2023
- Published by Wiley in The Laryngoscope
- Vol. 134 (1) , 108-112
Utilizing a Functional Lumen Imaging Probe for Evaluation of the Pediatric Airway, A Pilot Study
Objectives: Accurate and reproducible measurements of the pediatric airway are critical for diagnostic evaluation and management of subglottic and tracheal stenosis. The endoluminal functional lumen imaging probe (EndoFLIP) is a catheter‐based imaging probe which utilizes impedance planimetry to calculate luminal parameters, including cross‐sectional area and compliance. Herein, we demonstrate the feasibility of this system for multidimensional evaluation of the pediatric airway.Methods: 3D‐printed pediatric laryngotracheal models were created based on computed tomography scans, then artificially deformed to simulate both circumferential and posterior subglottic stenosis. Two observers made six measurements of the minimum cross‐sectional area (MCSA) and length of stenosis of each model with EndoFLIP. Agreement between observer measurements and model dimensions was evaluated using Lin's concordance correlation coefficient; inter‐observer reliability was assessed using intraclass correlation.Results: Four models were created: two without pathology (MCSA: 132.4, 44.3 mm2) and two with subglottic stenosis (MCSA: 28.7, 59.7 mm2, stenotic length 27.8, 24.4 mm). Observer measurements of MCSA and length of stenosis demonstrated high concordance with the models (r = 0.99, 0.95, p < 0.001) with a mean error of 4.5% and 18.2% respectively. There was a low coefficient of variation (0.6%–2.8%) for measurements, indicating high precision. Interrater reliability was high for both MCSA and stenotic length (ICC: 0.99, 0.98).Conclusions: The EndoFLIP system allows for accurate and reproducible measurements of cross‐sectional area and stenotic length in pediatric airway models. This method may provide further advantages in the evaluation of airway distensibility, as well as measurements of asymmetric airway pathology.Level of Evidence: NA Laryngoscope, 134:108–112, 2024
- 17 May 2023
- Published by Wiley in The Laryngoscope
- Vol. 134 (1) , 108-112
Open AccessUtilizing a Functional Lumen Imaging Probe for Evaluation of the Pediatric Airway, A Pilot Study
Objectives: Accurate and reproducible measurements of the pediatric airway are critical for diagnostic evaluation and management of subglottic and tracheal stenosis. The endoluminal functional lumen imaging probe (EndoFLIP) is a catheter‐based imaging probe which utilizes impedance planimetry to calculate luminal parameters, including cross‐sectional area and compliance. Herein, we demonstrate the feasibility of this system for multidimensional evaluation of the pediatric airway.Methods: 3D‐printed pediatric laryngotracheal models were created based on computed tomography scans, then artificially deformed to simulate both circumferential and posterior subglottic stenosis. Two observers made six measurements of the minimum cross‐sectional area (MCSA) and length of stenosis of each model with EndoFLIP. Agreement between observer measurements and model dimensions was evaluated using Lin's concordance correlation coefficient; inter‐observer reliability was assessed using intraclass correlation.Results: Four models were created: two without pathology (MCSA: 132.4, 44.3 mm2) and two with subglottic stenosis (MCSA: 28.7, 59.7 mm2, stenotic length 27.8, 24.4 mm). Observer measurements of MCSA and length of stenosis demonstrated high concordance with the models (r = 0.99, 0.95, p < 0.001) with a mean error of 4.5% and 18.2% respectively. There was a low coefficient of variation (0.6%–2.8%) for measurements, indicating high precision. Interrater reliability was high for both MCSA and stenotic length (ICC: 0.99, 0.98).Conclusions: The EndoFLIP system allows for accurate and reproducible measurements of cross‐sectional area and stenotic length in pediatric airway models. This method may provide further advantages in the evaluation of airway distensibility, as well as measurements of asymmetric airway pathology.Level of Evidence: NA Laryngoscope, 134:108–112, 2024
- 8 March 2023
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 169 (4) , 988-998
A Self‐Configuring Deep Learning Network for Segmentation of Temporal Bone Anatomy in Cone‐Beam CT Imaging
Objective: Preoperative planning for otologic or neurotologic procedures often requires manual segmentation of relevant structures, which can be tedious and time‐consuming. Automated methods for segmenting multiple geometrically complex structures can not only streamline preoperative planning but also augment minimally invasive and/or robot‐assisted procedures in this space. This study evaluates a state‐of‐the‐art deep learning pipeline for semantic segmentation of temporal bone anatomy.Study Design: A descriptive study of a segmentation network.Setting: Academic institution.Methods: A total of 15 high‐resolution cone‐beam temporal bone computed tomography (CT) data sets were included in this study. All images were co‐registered, with relevant anatomical structures (eg, ossicles, inner ear, facial nerve, chorda tympani, bony labyrinth) manually segmented. Predicted segmentations from no new U‐Net (nnU‐Net), an open‐source 3‐dimensional semantic segmentation neural network, were compared against ground‐truth segmentations using modified Hausdorff distances (mHD) and Dice scores.Results: Fivefold cross‐validation with nnU‐Net between predicted and ground‐truth labels were as follows: malleus (mHD: 0.044 ± 0.024 mm, dice: 0.914 ± 0.035), incus (mHD: 0.051 ± 0.027 mm, dice: 0.916 ± 0.034), stapes (mHD: 0.147 ± 0.113 mm, dice: 0.560 ± 0.106), bony labyrinth (mHD: 0.038 ± 0.031 mm, dice: 0.952 ± 0.017), and facial nerve (mHD: 0.139 ± 0.072 mm, dice: 0.862 ± 0.039). Comparison against atlas‐based segmentation propagation showed significantly higher Dice scores for all structures (p < .05).Conclusion: Using an open‐source deep learning pipeline, we demonstrate consistently submillimeter accuracy for semantic CT segmentation of temporal bone anatomy compared to hand‐segmented labels. This pipeline has the potential to greatly improve preoperative planning workflows for a variety of otologic and neurotologic procedures and augment existing image guidance and robot‐assisted systems for the temporal bone.
- 8 March 2023
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 169 (4) , 988-998
A Self‐Configuring Deep Learning Network for Segmentation of Temporal Bone Anatomy in Cone‐Beam CT Imaging
Objective: Preoperative planning for otologic or neurotologic procedures often requires manual segmentation of relevant structures, which can be tedious and time‐consuming. Automated methods for segmenting multiple geometrically complex structures can not only streamline preoperative planning but also augment minimally invasive and/or robot‐assisted procedures in this space. This study evaluates a state‐of‐the‐art deep learning pipeline for semantic segmentation of temporal bone anatomy.Study Design: A descriptive study of a segmentation network.Setting: Academic institution.Methods: A total of 15 high‐resolution cone‐beam temporal bone computed tomography (CT) data sets were included in this study. All images were co‐registered, with relevant anatomical structures (eg, ossicles, inner ear, facial nerve, chorda tympani, bony labyrinth) manually segmented. Predicted segmentations from no new U‐Net (nnU‐Net), an open‐source 3‐dimensional semantic segmentation neural network, were compared against ground‐truth segmentations using modified Hausdorff distances (mHD) and Dice scores.Results: Fivefold cross‐validation with nnU‐Net between predicted and ground‐truth labels were as follows: malleus (mHD: 0.044 ± 0.024 mm, dice: 0.914 ± 0.035), incus (mHD: 0.051 ± 0.027 mm, dice: 0.916 ± 0.034), stapes (mHD: 0.147 ± 0.113 mm, dice: 0.560 ± 0.106), bony labyrinth (mHD: 0.038 ± 0.031 mm, dice: 0.952 ± 0.017), and facial nerve (mHD: 0.139 ± 0.072 mm, dice: 0.862 ± 0.039). Comparison against atlas‐based segmentation propagation showed significantly higher Dice scores for all structures (p < .05).Conclusion: Using an open‐source deep learning pipeline, we demonstrate consistently submillimeter accuracy for semantic CT segmentation of temporal bone anatomy compared to hand‐segmented labels. This pipeline has the potential to greatly improve preoperative planning workflows for a variety of otologic and neurotologic procedures and augment existing image guidance and robot‐assisted systems for the temporal bone.
- 1 July 2022
- Published by Wolters Kluwer Health in Otology & Neurotology
- Vol. 43 (6) , e679-e687
Statistical Shape Model of the Temporal Bone Using Segmentation Propagation
Hypothesis: Automated image registration techniques can successfully determine anatomical variation in human temporal bones with statistical shape modeling. Background: There is a lack of knowledge about inter-patient anatomical variation in the temporal bone. Statistical shape models (SSMs) provide a powerful method for quantifying variation of anatomical structures in medical images but are time-intensive to manually develop. This study presents SSMs of temporal bone anatomy using automated image-registration techniques. Methods: Fifty-three cone-beam temporal bone CTs were included for SSM generation. The malleus, incus, stapes, bony labyrinth, and facial nerve were automatically segmented using 3D Slicer and a template-based segmentation propagation technique. Segmentations were then used to construct SSMs using MATLAB. The first three principal components of each SSM were analyzed to describe shape variation. Results: Principal component analysis of middle and inner ear structures revealed novel modes of anatomical variation. The first three principal components for the malleus represented variability in manubrium length (mean: 4.47 mm; ±2-SDs: 4.03–5.03 mm) and rotation about its long axis (±2-SDs: -1.6° to 1.8° posteriorly). The facial nerve exhibits variability in first and second genu angles. The bony labyrinth varies in the angle between the posterior and superior canals (mean: 88.9°; ±2-SDs: 83.7°–95.7°) and cochlear orientation (±2-SDs: -4.0° to 3.0° anterolaterally). Conclusions: SSMs of temporal bone anatomy can inform surgeons on clinically relevant inter-patient variability. Anatomical variation elucidated by these models can provide novel insight into function and pathophysiology. These models also allow further investigation of anatomical variation based on age, BMI, sex, and geographical location.
- 1 July 2022
- Published by Wolters Kluwer Health in Otology & Neurotology
- Vol. 43 (6) , e679-e687
Statistical Shape Model of the Temporal Bone Using Segmentation Propagation
Hypothesis: Automated image registration techniques can successfully determine anatomical variation in human temporal bones with statistical shape modeling. Background: There is a lack of knowledge about inter-patient anatomical variation in the temporal bone. Statistical shape models (SSMs) provide a powerful method for quantifying variation of anatomical structures in medical images but are time-intensive to manually develop. This study presents SSMs of temporal bone anatomy using automated image-registration techniques. Methods: Fifty-three cone-beam temporal bone CTs were included for SSM generation. The malleus, incus, stapes, bony labyrinth, and facial nerve were automatically segmented using 3D Slicer and a template-based segmentation propagation technique. Segmentations were then used to construct SSMs using MATLAB. The first three principal components of each SSM were analyzed to describe shape variation. Results: Principal component analysis of middle and inner ear structures revealed novel modes of anatomical variation. The first three principal components for the malleus represented variability in manubrium length (mean: 4.47 mm; ±2-SDs: 4.03–5.03 mm) and rotation about its long axis (±2-SDs: -1.6° to 1.8° posteriorly). The facial nerve exhibits variability in first and second genu angles. The bony labyrinth varies in the angle between the posterior and superior canals (mean: 88.9°; ±2-SDs: 83.7°–95.7°) and cochlear orientation (±2-SDs: -4.0° to 3.0° anterolaterally). Conclusions: SSMs of temporal bone anatomy can inform surgeons on clinically relevant inter-patient variability. Anatomical variation elucidated by these models can provide novel insight into function and pathophysiology. These models also allow further investigation of anatomical variation based on age, BMI, sex, and geographical location.
- 8 February 2022
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 167 (4) , 731-738
Automated Extraction of Anatomical Measurements From Temporal Bone CT Imaging
Objective: Proposed methods of minimally invasive and robot‐assisted procedures within the temporal bone require measurements of surgically relevant distances and angles, which often require time‐consuming manual segmentation of preoperative imaging. This study aims to describe an automatic segmentation and measurement extraction pipeline of temporal bone cone‐beam computed tomography (CT) scans.Study Design: Descriptive study of temporal bone measurements.Setting: Academic institution.Methods: A propagation template composed of 16 temporal bone CT scans was formed with relevant anatomical structures and landmarks manually segmented. Next, 52 temporal bone CT scans were autonomously segmented using deformable registration techniques from the Advanced Normalization Tools Python package. Anatomical measurements were extracted via in‐house Python algorithms. Extracted measurements were compared to ground truth values from manual segmentations.Results: Paired t test analyses showed no statistical difference between measurements using this pipeline and ground truth measurements from manually segmented images. Mean (SD) malleus manubrium length was 4.39 (0.34) mm. Mean (SD) incus short and long processes were 2.91 (0.18) mm and 3.53 (0.38) mm, respectively. The mean (SD) maximal diameter of the incus long process was 0.74 (0.17) mm. The first and second facial nerve genus had mean (SD) angles of 68.6 (6.7) degrees and 111.1 (5.3) degrees, respectively. The facial recess had a mean (SD) span of 3.21 (0.46) mm. Mean (SD) minimum distance between the external auditory canal and tegmen was 3.79 (1.05) mm.Conclusions: This is the first study to automatically extract relevant temporal bone anatomical measurements from CT scans using segmentation propagation. Measurements from these models can streamline preoperative planning, improve future segmentation techniques, and help develop future image‐guided or robot‐assisted systems for temporal bone procedures.
- 8 February 2022
- Published by Wiley in Otolaryngology–Head and Neck Surgery
- Vol. 167 (4) , 731-738
Automated Extraction of Anatomical Measurements From Temporal Bone CT Imaging
Objective: Proposed methods of minimally invasive and robot‐assisted procedures within the temporal bone require measurements of surgically relevant distances and angles, which often require time‐consuming manual segmentation of preoperative imaging. This study aims to describe an automatic segmentation and measurement extraction pipeline of temporal bone cone‐beam computed tomography (CT) scans.Study Design: Descriptive study of temporal bone measurements.Setting: Academic institution.Methods: A propagation template composed of 16 temporal bone CT scans was formed with relevant anatomical structures and landmarks manually segmented. Next, 52 temporal bone CT scans were autonomously segmented using deformable registration techniques from the Advanced Normalization Tools Python package. Anatomical measurements were extracted via in‐house Python algorithms. Extracted measurements were compared to ground truth values from manual segmentations.Results: Paired t test analyses showed no statistical difference between measurements using this pipeline and ground truth measurements from manually segmented images. Mean (SD) malleus manubrium length was 4.39 (0.34) mm. Mean (SD) incus short and long processes were 2.91 (0.18) mm and 3.53 (0.38) mm, respectively. The mean (SD) maximal diameter of the incus long process was 0.74 (0.17) mm. The first and second facial nerve genus had mean (SD) angles of 68.6 (6.7) degrees and 111.1 (5.3) degrees, respectively. The facial recess had a mean (SD) span of 3.21 (0.46) mm. Mean (SD) minimum distance between the external auditory canal and tegmen was 3.79 (1.05) mm.Conclusions: This is the first study to automatically extract relevant temporal bone anatomical measurements from CT scans using segmentation propagation. Measurements from these models can streamline preoperative planning, improve future segmentation techniques, and help develop future image‐guided or robot‐assisted systems for temporal bone procedures.