Classification of Wideband Tympanometry by Deep Transfer Learning With Data Augmentation for Automatic Diagnosis of Otosclerosis.

Otosclerosis is a common disease of the middle ear leading to stapedial fixation. Its rapid and non-invasive diagnosis could be achieved through wideband tympanometry (WBT), but the interpretation of the raw data provided by this tool is complex and time-consuming. Convolutional neural networks (CNN) could potentially be applied to this situation to help the clinicians categorize WBT data. A dataset containing 135 samples from 80 patients with otosclerosis and 55 controls was obtained. We designed a lightweight CNN to categorize samples into the otosclerosis and control. Receiver operating characteristic (ROC) analysis showed an area under the curve (AUC) of 0.95 ±0.011, and the F1-score was 0.89 ±0.031 ( r=10). The performance was further improved by data augmentation schemes and transfer learning strategies (AUC: 0.97 ±0.010, F1-score: 0.94 ±0.016, , ANOVA). Finally, the most relevant diagnostic features employed by the CNN were assessed via the activation pattern heatmaps. These results are crucial for the visual interpretation of WBT graphic outputs which clinicians use in routine, and for a better understanding of the WBT signal in relation to the ossicular mechanics.

Augmented reality for inner ear procedures: visualization of the cochlear central axis in microscopic videos.

PURPOSE: Visualization of the cochlea is impossible due to the delicate and intricate ear anatomy. Augmented reality may be used to perform auditory nerve implantation by transmodiolar approach in patients with profound hearing loss. METHODS: We present an augmented reality system for the visualization of the cochlear axis in surgical videos. The system starts with an automatic anatomical landmark detection in preoperative computed tomography images based on deep reinforcement learning. These landmarks are used to register the preoperative geometry with the real-time microscopic video captured inside the auditory canal. Three-dimensional pose of the cochlear axis is determined using the registration projection matrices. In addition, the patient microscope movements are tracked using an image feature-based tracking process. RESULTS: The landmark detection stage yielded an average localization error of [Formula: see text] mm ([Formula: see text]). The target registration error was [Formula: see text] mm for the cochlear apex and [Formula: see text] for the cochlear axis. CONCLUSION: We developed an augmented reality system to visualize the cochlear axis in intraoperative videos. The system yielded millimetric accuracy and remained stable throughout the experimental study despite camera movements throughout the procedure in experimental conditions.

Augmented Reality of the Middle Ear Combining Otoendoscopy and Temporal Bone Computed Tomography.

HYPOTHESIS: Augmented reality (AR) may enhance otologic procedures by providing sub-millimetric accuracy and allowing the unification of information in a single screen. BACKGROUND: Several issues related to otologic procedures can be addressed through an AR system by providing sub-millimetric precision, supplying a global view of the middle ear cleft, and advantageously unifying the information in a single screen. The AR system is obtained by combining otoendoscopy with temporal bone computer tomography (CT). METHODS: Four human temporal bone specimens were explored by high-resolution CT-scan and dynamic otoendoscopy with video recordings. The initialization of the system consisted of a semi-automatic registration between the otoendoscopic video and the 3D CT-scan reconstruction of the middle ear. Endoscope movements were estimated by several computer vision techniques (feature detectors/descriptors and optical flow) and used to warp the CT-scan to keep the correspondence with the otoendoscopic video. RESULTS: The system maintained synchronization between the CT-scan image and the otoendoscopic video in all experiments during slow and rapid (5-10 mm/s) endoscope movements. Among tested algorithms, two feature-based methods, scale-invariant feature transform (SIFT); and speeded up robust features (SURF), provided sub-millimeter mean tracking errors (0.38 ±â€Š0.53 mm and 0.20 ±â€Š0.16 mm, respectively) and an adequate image refresh rate (11 and 17 frames per second, respectively) after 2 minutes of procedure with continuous endoscope movements. CONCLUSION: A precise augmented reality combining video and 3D CT-scan data can be applied to otoendoscopy without the use of conventional neuronavigation tracking thanks to computer vision algorithms.

Maturation of Subjective Visual Vertical in Children.

OBJECTIVE: The attraction of the subjective visual vertical (SVV) to the side of initial rod presentation has already been described in adults. The aim of this study was to evaluate this phenomenon in children and to analyze the effect of sex and maturation in this population. STUDY DESIGN: Retrospective cross-sectional study. SETTING: Tertiary referral center. PATIENTS: Six hundred and one individuals aged between 4 and 19 years. INTERVENTION: All subjects underwent a complete balance workup. SVV was measured by presenting a laser line 12 times in total darkness with a 45 degrees deviation from the vertical alternatively on the left and the right. The patient was seated and asked to replace the bar vertically with a remote control. RESULTS: On average, SVV was tilted to the side of the rod presentation at each iteration. The cumulative tilt to the side of presentation after 12 measures was higher in the 4 to 7 years age group and decreased progressively with age (25 ±â€Š2.2 degrees in 4-7 years, n = 109 versus 5 ±â€Š1.4 in 15-19 years, n = 204, p < 0.001, analysis of variance [ANOVA]). The cumulative tilt was higher in girls than in boys in the 15 to 19 years group (8 ±â€Š2.5 degrees, n = 104 versus 2 ±â€Š1.2, n = 100, respectively, p < 0.001, ANOVA). This phenomenon appeared independent from the type of vestibular disorder. CONCLUSION: Young children are highly attracted to the side of rod presentation during SVV measurements. This phenomenon gradually disappears with maturation, faster in boys than in girls.

Virtual Endoscopy to Plan Transtympanic Approach to Labyrinthine Windows.

OBJECTIVE: The aim of this study was to evaluate the position and the accessibility of labyrinthine windows through the external auditory canal (EAC) by virtual endoscopy based on computed tomographic scan images. STUDY DESIGN: Prospective cross-sectional study. SETTING: Tertiary referral center. PATIENTS: Sixty-three high-resolution temporal bone computed tomographic scans were obtained from 34 adult patients undergoing various otologic procedures. INTERVENTION: Images were analyzed by the virtual endoscopy function included in Osirix (www.osirix-viewer.com). The endoscope was constrained in the EAC. The visible surfaces of target anatomic structures were assessed on multiplanar reconstruction views. RESULTS: The optimal angles of the virtual endoscope position showed a relatively high interindividual variability in the axial plane (64 ± 2.4 degrees for the oval window [OW] and 60 ± 2.5 degrees for the round window [RW]) position, but a low interindividual variation was noted in the coronal plane (107 ± 1.5 degrees for the OW and 112 ± 1.7 degrees for the RW). The RW was accessible in 87% of cases. The OW accessibility could be staged as follows: 1, invisible stapedial superstructure (10% of cases); 2, stapedial posterior crus partly visible (33%); 3, entire posterior crus and pyramid visible (44%); 4, posterior and anterior crus visible (13%). CONCLUSION: Virtual endoscopy through the EAC can evaluate the accessibility of the OW or RW via a transcanal route. This technique seems to be helpful in preplanning minimally invasive procedures by this approach such as cochlear implantation.

Subjective Visual Vertical Tilt Attraction to the Side of Rod Presentation: Effects of Age, Sex, and Vestibular Disorders.

OBJECTIVE: The aim of this study was to evaluate the effect of initial rod position on the subjective visual vertical (SVV) tilt and to investigate the effect of sex and age on the SVV tilt induced by this initial position. STUDY DESIGN: Prospective cross-sectional study. SETTING: Tertiary referral center. PATIENTS: The study included 6598 consecutive patients with a large range of vestibular disorders and 333 control subjects. The mean age was 55 years (range, 3-97), and the sex ratio was 0.6. INTERVENTION: SVV was measured by presenting a phosphorescent rod 12 times in total darkness with a 45-degree deviation from the vertical alternatively on the right and left. The patient was asked to replace the bar vertically with a remote control. RESULTS: On average, SVV at each iteration was tilted to the side of the rod presentation. It was stronger in female subjects, in younger individuals (<20 years) and seniors (>50). It was also higher in patients with a left vestibular loss in comparison to those with a right involvement. CONCLUSIONS: These effects suggest that short-term visual memory and multisensory cortical processing interfere with SVV measurements.

Validation method of a middle ear mechanical model to develop a surgical simulator.

Ossicular surgery requires a high dexterity for the manipulation of the fragile and small middle ear components. Currently, the only efficient technique for training residents in otological surgery is through the use of temporal bone specimens, where any existing surgical simulator does not provide useful feedback. The objective of this study was to develop a finite-element model of the human ossicular chain dedicated to surgical simulation and to propose a method to evaluate its behavior. A model was developed based on human middle ear micromagnetic resonance imaging. The mechanical parameters were determined according to published data. To assess its performance, the middle ear transfer function was analyzed. The robustness of our model and the influence of different middle ear components were also evaluated at low frequency by static force pressure simulations. The mechanical behavior of our model in nominal and pathological conditions was in good agreement with published human temporal bone measurements. We showed that the cochlea influences the transfer function only at high frequency and could be omitted from a surgical simulator. In addition, surgeons were able to manipulate the validated middle ear model with a real-time haptic feedback. The computational efficiency of our approach allowed real-time interactions, making it suitable for use in a training simulator.

Is electrode-modiolus distance a prognostic factor for hearing performances after cochlear implant surgery?

The aim of this study was to evaluate electrode array position in relation to cochlear anatomy and its influence on hearing performance in cochlear implantees. Twenty-two patients (25 ears) with Med-El cochlear implants were included in this retrospective study. A negative correlation was observed between electrode-modiolus distance (EMD) at the cochlear base and monosyllabic word discrimination 6 months after implantation. We found no correlation between EMD and hearing outcome at 12 months. The insertion depth/cochlear perimeter ratio appeared to negatively influence the EMD at the base. Indeed, deep insertions in small cochleae appeared to yield smaller EMD and better hearing performance. This observation supports the idea of preplanning the surgery by adapting the electrode array to the length of the available scala tympani.

Registration of a validated mechanical atlas of middle ear for surgical simulation.

This paper is centered on the development of a new training and rehearsal simulation system for middle ear surgery. First, we have developed and validated a mechanical atlas based on finite element method of the human middle ear. The atlas is based on a microMRI. Its mechanical behavior computed in real-time has been successfully validated. In addition, we propose a method for the registration of the mechanical atlas on patient imagery. The simulation can be used for a rehearsal surgery with the geometrical anatomy of a given patient and with mechanical data that are validated. Moreover, this process does not necessitate a complete re-built of the model.

Friction force measurement during cochlear implant insertion: application to a force-controlled insertion tool design.

HYPOTHESIS: The aim of the study was to evaluate force profiles during array insertion in human cochlea specimens and to evaluate a mechatronic inserter using a 1-axis force sensor. BACKGROUND: Today, the surgical challenge in cochlear implantation is the preservation of the anatomic structures and the residual hearing. In routine practice, the electrode array is inserted manually with a limited sensitive feedback. MATERIALS AND METHODS: Hifocus 1J electrode arrays were studied. The bench test comprised a mechatronic inserter combined to a 1-axis force sensor between the inserter and the base of the array and a 6-axis force sensor beneath the cochlea model. Influence of insertion tube material, speed (0.15, 0.5, and 1.5 mm/s) and lubricant on frictions forces were studied (no-load). Different models were subsequently evaluated: epoxy scala tympani model and temporal bones. RESULTS: Frictions forces were lower in the plastic tube compared with those in the metal tube (0.09 ± 0.028 versus 0.14 ± 0.034 at 0.5 mm/s, p < 0.001) and with the use of hyaluronic acid gel. Speed did not influence frictions forces in our study. Insertion force profiles provided by the 1- and 6-axis force sensors were similar when friction forces inside the insertion tool (no-load measurements) were subtracted from the 1-axis sensor data in the epoxy and temporal bone models (mean error, 0.01 ± 0.001 N). CONCLUSION: Using a sensor included in the inserter, we were able to measure array insertion forces. This tool can be potentially used to provide real-time information to the surgeon during the procedure.

Long-term follow-up of 287 meningiomas in neurofibromatosis type 2 patients: clinical, radiological, and molecular features.

Decision-making criteria for optimal management of meningiomas in neurofibromatosis type 2 (NF2) patients is hampered by lack of robust data, particularly long-term natural history. Seventy-four NF2 patients harboring 287 cranial meningiomas followed up for a mean period of 110.2 months were studied retrospectively. The median number of meningiomas per patient was 3. The mean maximum diameter of meningiomas at diagnosis was 14.3 mm, with a mean annual growth rate of 1.5 mm. Sixty-six percent of tumors showed no or minimal growth. In a subgroup of patients with 3D MRI, 7.3% of meningiomas (28% of patients) had a volumetric growth rate 20% or more per year. Twenty-five de novo meningiomas appeared during the follow-up (8.7%) and demonstrated a higher growth rate than other meningiomas (6.6 mm/year). Fifty-six meningiomas (23%) in 34 NF2 patients (45.9%) were operated on during the follow-up period. Among symptomatic resected meningiomas, grades II and III tumors were found in 29% and 6% of cases, respectively, with a remarkable intratumor histological heterogeneity. Single nucleotide polymorphism array analysis of 22 meningioma samples in 14 NF2 patients showed increasing chromosome instability with increasing grade, the most frequent losses being on 22q, 1p, 18q, and 6p. This study provides clues to improve tailored treatment of meningiomas: de novo and brain edema-associated meningiomas require active treatment. Future clinical trials in NF2 need to focus specifically on meningiomas as the primary endpoint and should include patients with meningiomas growing 20% or more per year in order to assess new treatments.

From conception to application of a tele-operated assistance robot for middle ear surgery.

The authors' goal was to design and evaluate a robot dedicated to middle ear surgery. Specifications for dimensions, forces, and kinematics were collected, based on the otosclerosis procedure. The robot structure has a compact geometry with 3 linear and 3 rotatory motors. It is remotely piloted via a robot-surgeon interface under operative microscope. Ability to reach anatomical targets, to perform stapedectomy, and to place prosthesis in a model of stapedotomy was evaluated by 6 surgeons. Multiple anatomical targets in the middle ear could be successfully reached without damaging surrounding structures. The robot could be used under operative microscope with minimal visual field impairment or jointly with a 4-mm endoscope through the external auditory canal to perform stapedectomy in temporal bone specimens. Prosthesis could be inserted in the stapedotomy model. The assistance robot is the first prototype with 6 degrees of freedom, a kinematic structure, and dimensions optimized for tele-operated middle ear surgery.

Interaction between electric and acoustic cues in diotic condition for speech perception in quiet and noise by cochlear implantees.

OBJECTIVE: This study aimed to evaluate the interaction of electric and acoustic cues in diotic condition in cochlear implantees. MATERIALS AND METHODS: Five adult cochlear implantees with residual contralateral hearing were prospectively evaluated in hearing aid only (HA), cochlear implant only (CI), and HA + CI modes by audiometry (pure tone, dissyllabic words, and sentences), and sound quality questionnaires. CI electrodes corresponding to preserved frequencies in the contralateral ear (free-field aided thresholds, <50 dB) were then deactivated, and patients were retested after 20 to 30 days. RESULTS: Sentences in silence showed a benefit of CI and the additive effect of HA + CI. As expected, performances with CI alone decreased after apical electrode deactivation. In contrast, speech performances (Marginal Benefit from Acoustic Amplification sentences) in HA + CI mode were not altered by electrode deactivation in silence (90 ± 5.9% before versus 81 ± 10.1% after deactivation, not significant, 2-way analysis of variance) or in noise (78 ± 4.8% before versus 66 ± 11.9% after deactivation, not significant, 2-way analysis of variance). Performances for dissyllabic words confirmed these results. Questionnaires showed a significant compensation of partial electrode deactivation by the contralateral hearing. Moreover, the human voice was reported to be significantly less metallic. CONCLUSION: These results suggested a significant complementarity of acoustic and electric diotic cues but also some redundancy affecting the sound quality.

Continuous facial nerve stimulating burr for otologic surgeries.

OBJECTIVE: To evaluate a continuous facial nerve (FN) stimulating burr (the StimBurGard) during otologic/neurotologic procedures in terms of safety and reliability when drilling in contact with the Fallopian canal (FC) of the mastoid segment of the FN. STUDY DESIGN: Prospective clinical trial. SETTING: Tertiary referral center. PATIENTS: Thirty-five patients operated through translabyrinthine (TL) approach for vestibular schwannoma removal were divided into 3 groups. Group 1 (5 patients): the stimulation current was set at 3 and then at 2 mA visualizing the localization of the burr when the first response at 100-µV threshold was obtained in the mastoid cavity. Group 2 (15 patients): exposure of the FC in the mastoid segment during TL approach was stopped when the first response was obtained at 1-mA stimulation; FC thickness in the second genu and mastoid segment of the FC was evaluated on a postoperative computed tomographic (CT) scan, and FC dehiscence observed on CT scan was compared with surgical observation. Group 3 (15 patients), exposure of the FC was performed as routinely done during a TL approach and surgical observation of FC dehiscence; stimulation values in mA at the 100 µV threshold and FC thickness on postoperative CT scan were evaluated. In all cases, the stimulation value at the cerebellopontine angle root of the FN with a 100-µV response threshold was measured before tumor resection. RESULTS: Group 1: stimulation at 3 mA occurred in aditus ad antrum and at 2 mA near the FC. Group 2: mean thickness of 1.09 ± 0.69 mm with 2 cases of radiologic dehiscence of the FN. Group 3: the stimulation threshold was 0.6 ± 0.37 mA, and the thickness was 0.41 ± 0.56 mm with 9 cases of uncovered FN (p = 0.0082). In all patients, FN at brainstem was stimulated at 0.03 mA before VS dissection. CONCLUSION: Continuous FN stimulating burr by means of the StimBurGard system is a safe and effective tool for FN stimulation and identification. The integrity of FC is preserved in most cases when the stimulation intensity is 1 mA.

Visual analog scale to assess vertigo and dizziness after repositioning maneuvers for benign paroxysmal positional vertigo.

INTRODUCTION: The aim of this study was to evaluate the value of Visual Analog Scale (VAS) in evaluating vertigo and dizziness in patients with benign paroxysmal positional vertigo (BPPV). MATERIALS AND METHODS: 226 adult patients suffering from a BPPV and undergoing repositioning maneuvers were included in this prospective study. Vertigo (V) and dizziness (D) were separately evaluated from day 0 to 5 by VAS. V and D scores decreased both exponentially but with different constants. D scores were not correlated to V suggesting the independency of the ratings. D was related to the dizziness intensity level expressed at interrogation. Both D and V scores were related to patients' global satisfaction. CONCLUSION: VAS allows differentiating vertigo from dizziness and provides coherent results with other clinical indicators.

Minimally invasive computer-assisted approach for cochlear implantation: a human temporal bone study.

Computer-assisted navigation systems can now potentially guide the surgeon to the cochlea with a trajectory avoiding the facial nerve through a keyhole approach. Five temporal bone specimens, with 4 titanium screws placed in the mastoid cortex, were studied. Preoperative computed tomographic scan images were loaded on an electromagnetic computer-assisted surgery (CAS) system (Digipointeur, Collin, Bagneux, France). A drill was connected to the CAS to monitor its progression continuously. A conical approach passing through the facial recess and ending in the scala tympani was performed. A 0.5-mm wire was inserted into the cochlea. The keyhole approach was technically feasible in all cases. No facial nerve injury was observed on imaging and dissection control. The wire was positioned in the scala tympani and the position accuracy of the CAS was <0.76 mm on the target in all cases. The CAS system with fiducial markers yielded sufficient precision to allow a minimally invasive approach to the cochlea.

Functional results of Vibrant Soundbridge middle ear implants in conductive and mixed hearing losses.

OBJECTIVE: Our purpose was to evaluate the results of Vibrant Soundbridge (VSB) in conductive or mixed hearing loss. MATERIALS AND METHODS: Twenty-five adult patients (29 ears) with a mixed or conductive hearing loss and various etiologies were included in this retrospective study. The preoperative ipsilateral pure tone average was 71 ± 3.0 dB, and the average bone conduction threshold was 42 ± 2.8 dB (n = 29). The transducer was placed on the long apophysis of the incus (n = 16), in the round window (n = 10) or on the stapes (n = 3). RESULTS: No complications were noted. The bone conduction threshold remained unchanged. VSB was activated in all cases. The postoperative pure tone average without VSB was 63 ± 3.9 dB (n = 24) and with VSB in free-field condition 24 ± 2.1 dB (n = 22). CONCLUSIONS: VSB is safe and efficacious for auditory rehabilitation in conductive and mixed hearing losses.

Effects of systemic administration of methylprednisolone on residual hearing in an animal model of cochlear implantation.

CONCLUSION: Delivery of intramuscular injection of methylprednisolone around the implantation surgery improved the hearing threshold shift induced by cochlear implantation. OBJECTIVES: During electroacoustic cochlear implantation surgery, the residual hearing is not preserved in about 15% of cases. In this study, we tested the effects of intramuscular administration of methylprednisolone on the hearing loss induced by cochlear implantation in a model of guinea pig cochlear implantation. METHODS: Eleven guinea pigs with normal hearing were implanted with a 254 µm diameter silicone array through a cochleostomy, and the effects on hearing of longstanding array insertion (21 days) were assessed. Six of the implanted animals received intramuscular administration of methylprednisolone. Auditory brainstem response recordings were performed before and up to 21 days after the cochlear implantation. CT scans were performed in some animals 1 month after implantation. RESULTS: CT scans confirmed that the array was well positioned in tested animals. From days 3 to 21, a hearing loss of about 30 dB on all frequencies was observed in the implanted nontreated group. This hearing loss remained stable during the whole follow-up period. Compared with implanted nontreated animals, the hearing threshold shift decreased by 12 dB in animals treated with methylprednisolone.

Protective effect of systemic administration of erythropoietin on auditory brain stem response and compound action potential thresholds in an animal model of cochlear implantation.

OBJECTIVES: An animal model of cochlear implantation has been developed, and the hearing threshold was evaluated after different surgical procedures. The effect of perioperative systemic administration of erythropoietin on the hearing loss induced by cochlear implantation was tested. METHODS: Twenty-nine guinea pigs with normal hearing underwent implantation of a 254-microm-diameter array through a cochleostomy. The effects on hearing of cochleostomy and transient and long-term array implantation (21 days) were assessed by testing of the auditory brain stem responses and compound action potentials. Eleven implanted animals received intraperitoneal administration of erythropoietin. Selected computed tomographic scans and cochlear histologic studies were performed 1 month after implantation to confirm proper placement of the array. The erythropoietin concentration at the time of surgery was assessed in samples of perilymph, cerebrospinal fluid, and blood. RESULTS: The cochleostomy and transient array insertion had no effect on hearing thresholds. Long-term array implantation induced a stable decrease of hearing threshold (30 dB), a decrease that was reduced by 12 dB in erythropoietin-treated animals. The erythropoietin-treated animals had better hearing preservation at higher frequencies. Fibrosis surrounding the array was seen in both groups. CONCLUSIONS: The hearing loss observed was probably due to the presence of the array in the cochlea. The intraperitoneal injection of erythropoietin improved the hearing threshold shift induced by implantation.

Intraoperative electromyography and surgical observations as predictive factors of facial nerve outcome in vestibular schwannoma surgery.

OBJECTIVE: To evaluate the predictive value of electromyographic recordings for the immediate facial nerve outcome after solitary vestibular schwannoma (VS) removal. STUDY DESIGN: Prospective cohort study. SETTING: Tertiary referral center. PATIENTS AND METHODS: From November 2005 to July 2007, 120 operated solitary VSs monitored by NIM Response 2.0 (Medtronic, Xomed, Jacksonville, FL, USA) were included. Intraoperative direct facial nerve stimulation at the brainstem (proximal pre site) before resection and at the brainstem (proximal post and distal sites, respectively) and internal acoustic meatus after resection were performed. Intraoperative recordings were performed on 4 muscles: Frontalis, Orbicularis oculi, Orbicularis oris, and chin muscles (CM). Postoperative facial function (House-Brackmann classification) was evaluated at postoperative Days (D) 2, 8, and 30. Grades 1 or 2 were considered as a good result. RESULTS: A good postoperative facial function was obtained in 71, 63, and 77% at postoperative D2, D8, and D30, respectively. Maximal conduction block was found at Day 8. Poor facial nerve outcomes were observed for large tumor with a facial nerve superior to the tumor with a strong adhesion to the nerve. With a stimulation threshold at proximal site after proximal site tumor resection at less than or equal to 0.04 mA, a good facial function was observed in 77% at Day 8. Using these criteria together with CM responses greater than 800 muV to supramaximal stimulation at (proximal post site) and a proximal/distal post ratio at CM response greater than 0.6, 93% of patients presented with a good facial function at Day 8. CONCLUSION: Using CM recording stimulation and supramaximal stimulation of the facial nerve in combination with thresholds seem to increase the predictive value of the monitoring for the immediate facial nerve function for VS surgery.