Automatic summarization of endoscopic skull base surgical videos through object detection and hidden Markov modeling.

Endoscopic endonasal surgery is a medical procedure that utilizes an endoscopic video camera to view and manipulate a surgical site accessed through the nose. Despite these surgeries being video recorded, these videos are seldom reviewed or even saved in patient files due to the size and length of the video file. Editing to a manageable size may necessitate viewing 3 h or more of surgical video and manually splicing together the desired segments. We suggest a novel multi-stage video summarization procedure utilizing deep semantic features, tool detections, and video frame temporal correspondences to create a representative summarization. Summarization by our method resulted in a 98.2% reduction in overall video length while preserving 84% of key medical scenes. Furthermore, resulting summaries contained only 1% of scenes with irrelevant detail such as endoscope lens cleaning, blurry frames, or frames external to the patient. This outperformed leading commercial and open source summarization tools not designed for surgery, which only preserved 57% and 46% of key medical scenes in similar length summaries, and included 36% and 59% of scenes containing irrelevant detail. Experts agreed that on average (Likert Scale = 4) that the overall quality of the video was adequate to share with peers in its current state.

Implementing a "Distance Traveled" Question to Improve Resident Diversity: Process and Feasibility.

Increasing diversity in the physician workforce is important to improving racial and ethnic disparities in health outcomes in the United States. We describe the implementation of a "distance traveled" question (DTQ) in our residency application process. For the 2021-2022 cycle, all applicants to the University of Washington otolaryngology residency program were allowed to complete an optional DTQ. Responses were shared with the application review committee. Following the distribution of interview invites, an anonymous survey was sent to all faculty reviewers. The response rate was 26 of 36 (72%). Among respondents, 20 (77%) felt that the DTQ helped them learn something new about the applicant, and 19 (73%) reported that the DTQ influenced their decision making about the applicant. Thus, a DTQ may provide faculty with new and influential information regarding residency applicants.

Evidence of Specialty-Specific Gender Disparities in Resident Recruitment and Selection.

BACKGROUND: Specialty-specific gender disparities are multifactorial, yet one area that is lacking from this discussion is the impact of recruitment and selection. OBJECTIVE: Customized data reports were utilized to compare trends in the gender representation of applicants and residents within 11 surgical and medical specialties between 2013 and 2018. METHODS: Applicant data was obtained from the Electronic Residency Application Service (ERAS) and resident data from the Accreditation Council for Graduate Medical Education (ACGME). Eleven specialties with the highest number of applications per applicant were included (dermatology, emergency medicine, general surgery, neurological surgery, obstetrics and gynecology [OB/GYN], orthopedic surgery, otolaryngology, plastic surgery, radiation oncology, radiology, urology). A Cochran-Armitage trend test assessed for changes in the proportion of females within the total applicant group and the corresponding matched resident group. A t test was utilized to compare the mean proportion of females for ERAS and ACGME data. RESULTS: Otolaryngology, plastic surgery, radiation oncology, and urology had no significant changes over the study period. Dermatology, general surgery, and orthopedic surgery demonstrated increased gender diversity in applicants over time, while OB/GYN demonstrated decreased gender diversity. General surgery and neurological surgery showed increased gender diversity in resident representation over time. Emergency medicine and radiology had increased gender diversity, and OB/GYN had decreased gender diversity in matched residents compared to applicants. CONCLUSIONS: Our findings provide baseline data, but also illustrate evident gaps in our understanding and attempts to improve gender diversity. A multifaceted approach to obtaining and assessing data from all stages of residency recruitment and selection is necessary to support these efforts.

Multicenter Interrater Reliability in the Endoscopic Assessment of Velopharyngeal Function Using a Video Instruction Tool.

OBJECTIVE: Assess interrater agreement of endoscopic assessment of velopharyngeal (VP) function before and after viewing the video instruction tool (VIT). We hypothesized improvement in interrater agreement using the Golding-Kushner scale (GKS) after viewing the VIT. STUDY DESIGN: Prospective study. SETTING: Multi-institutional. METHODS: Sixteen fellowship-trained pediatric otolaryngologists who treat velopharyngeal insufficiency (VPI) rated 50 video segments using the GKS before and after watching the VIT. Raters assessed gap size percentage and lateral pharyngeal wall (LPW), soft palate (SP), and posterior pharyngeal wall (PPW) movement. Intraclass correlation coefficient was estimated for these continuous measures. Raters also indicated the presence of a palatal notch, Passavant's ridge, and aberrant pulsations (categorical variables). Fleiss κ coefficient was used for categorical variables. Wilcoxon signed-rank test was performed on the difference between the pre/post individual video ratings. RESULTS: Reliability improved for all continuous variables after watching the instructional video. The improvement was significant for PPW (0.22-0.30, P < .001), SP (left: 0.63-0.68, P < .001 and right: 0.64-0.68, P = .001), and LPW (left: 0.49-0.54, P = .01 and right: 0.49-0.54, P = .09) but not significant for gap size (0.65-0.69, P = .36). Among categorical variables, agreement on Passavant's ridge significantly improved (0.30-0.36, P = .03). CONCLUSION: Exposure to a video instruction tool improves interrater agreement of endoscopic assessment of VP function. Significant improvement was observed in our primary end points, specifically posterior pharyngeal wall movement, soft palate movement, and lateral pharyngeal wall movement. There was less impact of the VIT on the interrater agreement of the categorical variables, palatal notch, Passavant's ridge, and aberrant pulsations.

Standardizing Laryngeal Cleft Evaluations: Reliability of the Interarytenoid Assessment Protocol.

OBJECTIVE: While the Benjamin-Inglis classification system is widely used to categorize laryngeal clefts, it does not clearly differentiate a type 1 cleft from normal anatomy, and there is no widely accepted or validated protocol for systematically evaluating interarytenoid mucosal height. We sought to propose the interarytenoid assessment protocol as a method to standardize the description of the interarytenoid anatomy and to test its reliability. STUDY DESIGN: Retrospective review of endoscopic videos. SETTING: Pediatric academic center. SUBJECTS AND METHODS: The interarytenoid assessment protocol comprises 4 steps for evaluation of the interarytenoid region relative to known anatomic landmarks in the supraglottis, glottis, and subglottis. Thirty consecutively selected videos of the protocol were reviewed by 4 otolaryngologists. The raters were blinded to identifying information, and the video order was randomized for each review. We assessed protocol completion times and calculated Cohen's linear-weighted κ coefficient between blinded expert raters and with the operating surgeon to evaluate interrater/intrarater reliability. RESULTS: Median age was 4.9 years (59 months; range, 1 month to 20 years). Median completion time was 144 seconds. Interrater and intrarater reliability showed substantial agreement (interrater κ = 0.71 [95% confidence interval (CI), 0.55-0.87]; intrarater mean κ = 0.70 [95% CI, 0.59-0.92/rater 1, 0.47-0.85/rater 2]; P < .001). Comparing raters to the operating surgeon demonstrated substantial agreement (mean κ = 0.62; 95% CI, 0.31-0.79/rater 1, 0.48-0.89/rater 2; P < .001). CONCLUSION: The interarytenoid assessment protocol appears reliable in describing interarytenoid anatomy. Rapid completion times and substantial interrater/intrarater reliability were demonstrated. Incorporation of this protocol may provide important steps toward improved standardization in the anatomic description of the interarytenoid region in pediatric dysphagia.

Automated Surgical Approach Planning for Complex Skull Base Targets: Development and Validation of a Cost Function and Semantic At-las.

Successful multidisciplinary treatment of skull base pathology requires precise preoperative planning. Current surgical approach (pathway) selection for these complex procedures depends on an individual surgeon's experiences and background training. Because of anatomical variation in both normal tissue and pathology (eg, tumor), a successful surgical pathway used on one patient is not necessarily the best approach on another patient. The question is how to define and obtain optimized patient-specific surgical approach pathways? In this article, we demonstrate that the surgeon's knowledge and decision making in preoperative planning can be modeled by a multiobjective cost function in a retrospective analysis of actual complex skull base cases. Two different approaches- weighted-sum approach and Pareto optimality-were used with a defined cost function to derive optimized surgical pathways based on preoperative computed tomography (CT) scans and manually designated pathology. With the first method, surgeon's preferences were input as a set of weights for each objective before the search. In the second approach, the surgeon's preferences were used to select a surgical pathway from the computed Pareto optimal set. Using preoperative CT and magnetic resonance imaging, the patient-specific surgical pathways derived by these methods were similar (85% agreement) to the actual approaches performed on patients. In one case where the actual surgical approach was different, revision surgery was required and was performed utilizing the computationally derived approach pathway.

Anatomical Region Segmentation for Objective Surgical Skill Assessment with Operating Room Motion Data.

Background Most existing objective surgical motion analysis schemes are limited to structured surgical tasks or recognition of motion patterns for certain categories of surgeries. Analyzing instrument motion data with respect to anatomical structures can break the limit, and an anatomical region segmentation algorithm is required for the analysis. Methods An atlas was generated by manually segmenting the skull base into nine regions, including left/right anterior/posterior ethmoid sinuses, frontal sinus, left and right maxillary sinuses, nasal airway, and sphenoid sinus. These regions were selected based on anatomical and surgical significance in skull base and sinus surgery. Six features, including left and right eye center, nasofrontal beak, anterior tip of nasal spine, posterior edge of hard palate at midline, and clival body at foramen magnum, were used for alignment. The B-spline deformable registration was adapted to fine tune the registration, and bony boundaries were automatically extracted for final precision improvement. The resultant deformation field was applied to the atlas, and the motion data were clustered according to the deformed atlas. Results Eight maxillofacial computed tomography scans were used in experiments. One was manually segmented as the atlas. The others were segmented by the proposed method. Motion data were clustered into nine groups for every dataset and outliers were filtered. Conclusions The proposed algorithm improved the efficiency of motion data clustering and requires limited human interaction in the process. The anatomical region segmentations effectively filtered out the portion of motion data that are out of surgery sites and grouped them according to anatomical similarities.

Quantitative Analysis of Transnasal Anterior Skull Base Approach: Report of Technology for Intraoperative Assessment of Instrument Motion.

OBJECTIVE: To develop a method to measure intraoperative surgical instrument motion. This model will be applicable to the study of surgical instrument kinematics including surgical training, skill verification, and the development of surgical warning systems that detect aberrant instrument motion that may result in patient injury. DESIGN: We developed an algorithm to automate derivation of surgical instrument kinematics in an endoscopic endonasal skull base surgery model. Surgical instrument motion was recorded during a cadaveric endoscopic transnasal approach to the pituitary using a navigation system modified to record intraoperative time-stamped Euclidian coordinates and Euler angles. Microdebrider tip coordinates and angles were referenced to the cadaver's preoperative computed tomography scan allowing us to assess surgical instrument kinematics over time. A representative cadaveric endoscopic endonasal approach to the pituitary was performed to demonstrate feasibility of our algorithm for deriving surgical instrument kinematics. CONCLUSIONS: Technical feasibility of automatically measuring intraoperative surgical instrument motion and deriving kinematics measurements was demonstrated using standard navigation equipment.

Crowd-sourced assessment of surgical skills in cricothyrotomy procedure.

BACKGROUND: Objective assessment of surgical skills is resource intensive and requires valuable time of expert surgeons. The goal of this study was to assess the ability of a large group of laypersons using a crowd-sourcing tool to grade a surgical procedure (cricothyrotomy) performed on a simulator. The grading included an assessment of the entire procedure by completing an objective assessment of technical skills survey. MATERIALS AND METHODS: Two groups of graders were recruited as follows: (1) Amazon Mechanical Turk users and (2) three expert surgeons from University of Washington Department of Otolaryngology. Graders were presented with a video of participants performing the procedure on the simulator and were asked to grade the video using the objective assessment of technical skills questions. Mechanical Turk users were paid $0.50 for each completed survey. It took 10 h to obtain all responses from 30 Mechanical Turk users for 26 training participants (26 videos/tasks), whereas it took 60 d for three expert surgeons to complete the same 26 tasks. RESULTS: The assessment of surgical performance by a group (n = 30) of laypersons matched the assessment by a group (n = 3) of expert surgeons with a good level of agreement determined by Cronbach alpha coefficient = 0.83. CONCLUSIONS: We found crowd sourcing was an efficient, accurate, and inexpensive method for skills assessment with a good level of agreement to experts' grading.