Fully automated preoperative segmentation of temporal bone structures from clinical CT scans.

Middle- and inner-ear surgery is a vital treatment option in hearing loss, infections, and tumors of the lateral skull base. Segmentation of otologic structures from computed tomography (CT) has many potential applications for improving surgical planning but can be an arduous and time-consuming task. We propose an end-to-end solution for the automated segmentation of temporal bone CT using convolutional neural networks (CNN). Using 150 manually segmented CT scans, a comparison of 3 CNN models (AH-Net, U-Net, ResNet) was conducted to compare Dice coefficient, Hausdorff distance, and speed of segmentation of the inner ear, ossicles, facial nerve and sigmoid sinus. Using AH-Net, the Dice coefficient was 0.91 for the inner ear; 0.85 for the ossicles; 0.75 for the facial nerve; and 0.86 for the sigmoid sinus. The average Hausdorff distance was 0.25, 0.21, 0.24 and 0.45 mm, respectively. Blinded experts assessed the accuracy of both techniques, and there was no statistical difference between the ratings for the two methods (p = 0.93). Objective and subjective assessment confirm good correlation between automated segmentation of otologic structures and manual segmentation performed by a specialist. This end-to-end automated segmentation pipeline can help to advance the systematic application of augmented reality, simulation, and automation in otologic procedures.

Mid-frequency sensorineural hearing loss: aetiology and prognosis.

OBJECTIVES: An audiometric finding of mid-frequency sensorineural hearing loss (MFSNHL), or a U-shaped pattern, is uncommon. The objective of this study is to investigate the aetiology and prognostic significance of MFSNHL. DESIGN: Tertiary academic referral centre-based retrospective case review and review of audiograms to determine the prevalence of this audiometric finding. METHODS: Patients with a pure tone threshold average at 1, 2, and 4 kHz at least 10 dB greater than the average at 0.5 and 8 kHz were included in this study; 35 patients met these criteria. The mean age of the patients was 34.6 years old (range 4-71 years). Twelve patients (33 per cent) were under 18 years of age. Serial audiograms were obtained for 14 patients. The notes were reviewed for any pertinent otologic history, subsequent diagnoses, management and disease course. RESULTS: The prevalence of MFSNHL in this practice setting is less than 1 per cent. The average hearing threshold in the mid-frequencies was 44 dB, which was 17 dB and 20 dB lower than at 0.5 Hz and 8 kHz, respectively. The pure tone average (0.5, 1, 2 kHz) was 40 dB. Sixteen patients (44 per cent) required amplification. Of all patients, 22 had hereditary hearing loss, eight had idiopathic hearing loss, and five adults had vestibular schwannomas. CONCLUSIONS: MFSNHL is an infrequent audiometric finding. The great majority of these cases are of presumed hereditary or idiopathic aetiology, although 22 per cent of adults had vestibular schwannomas. This series presents the causes and prognosis of this audiometric pattern.

Combined transpetrosal-subtemporal craniotomy for clival tumors with extension into the posterior fossa.

Tumors of the clivus, such as chordoma and chondrosarcoma, are generally amenable to an anterior surgical approach. However, approaches that traverse the pharynx or paranasal sinuses do not adequately expose tumor posterolateral to the horizontal course of the intrapetrous carotid artery. In addition, when tumor extends into the posterior fossa, supplemental exposure of neurovascular structures is necessary. A combination petrosectomy and subtemporal craniotomy can provide simultaneous access to the entire clivus as well as the lateral aspect of the midbrain, pons, and upper medulla. The extent of petrosectomy performed depends on a number of factors including status of hearing, facial nerve function, and degree of brainstem compression. In our experience with three patients (two chordomas and one chondrosarcoma), using either the retrolabyrinthine-subtemporal or transcochlear-subtemporal approach, excellent resection was achieved with acceptable morbidity considering the extensive nature of the disease.

Facial paralysis due to benign parotid tumors.

On rare occasions, facial paralysis associated with a parotid tumor need not denote malignancy. We present two cases in which, contrary to appropriate conventional wisdom, facial paralysis resulted from benign mixed tumors. Each patient presented over 8 years following primary surgical excision. In neither patient was a mass palpable, and facial paralysis was the sole sign of recurrent disease. Each patient had been followed up for several months with a presumptive diagnosis of Bell's palsy prior to discovery of recurrent tumor by radiologic imaging. In each case, at operation the tumor was found to infiltrate the temporal bone via the stylomastoid foramen. Facial paralysis presumably resulted from extrinsic compression of the facial nerve. These two cases add to the few previous reports of facial paralysis due to benign parotid gland tumors.

Exposure of the lateral extremity of the internal auditory canal through the retrosigmoid approach: a radioanatomic study.

The recent trend toward earlier diagnosis of acoustic neuroma has substantially increased the number of candidates suitable for surgery with an attempt at hearing preservation. Although the retrosigmoid approach affords the possibility of saving hearing in selected cases, it is associated with a somewhat greater morbidity that other approaches, in terms of persistent headache, cerebrospinal fluid leakage, and cerebellar dysfunction. For this reason, it is best used selectively, when the probability of success in hearing conservation is high. Only a portion of the internal auditory canal can be exposed through the retrosigmoid approach without violating the inner ear, a maneuver that greatly reduces the chance of preserving residual hearing. Substantial variability exists between individuals as to just how far laterally the internal auditory canal may be opened without compromising labyrinthine integrity. To assess the magnitude of this variability, measurements were obtained from 60 high-resolution temporal bone computed tomography scans with a schema intended to model the surgical angle of view used during the retrosigmoid procedure. Intraoperative measurements in a series of cases established that the actual surgical point of view is situated along a line that passes approximately 1.5 cm behind the sigmoid sinus. In this typical surgical position, these data predict that an average of 3.0 mm (32% of the internal auditory canal length) must be left unexposed to avoid labyrinthine injury, with a range between 1.1 mm and 5.3 mm (9% to 58% of the internal auditory canal). Each additional 1-cm retraction on the cerebellum beyond that customarily used affords approximately 1 mm (10% of the internal auditory canal) further exposure of the canal. When considering the retrosigmoid approach to an acoustic neuroma, the clinician is urged to evaluate each patient individually to estimate the amount of internal auditory canal accessible without the removal of a portion of the inner ear. This can be ascertained from an axially oriented, gadolinium-enhanced magnetic resonance imaging scan in the internal auditory canal plane by drawing a line that originates 1.5 cm behind the posterior margin of the sigmoid sinus and passes tangential to the most medial extent of the labyrinth. If this line intersects the posterior margin of the internal auditory canal at least 2 mm lateral to the deepest point of tumor penetration, then adequate exposure with preservation of the labyrinth is likely an achievable goal.

Chronic myringitis: prevalence, presentation, and natural history.

OBJECTIVE: The aim of this study was to examine the clinical presentation and natural history of chronic myringitis (CM). STUDY DESIGN: Retrospective case review. SETTING: Tertiary referral center. PATIENTS: Chronic myringitis is defined as a loss of tympanic membrane epithelium for >1 month without disease within the tympanic cavity. Seven hundred fifty patient records were reviewed to determine the prevalence of CM in an academic otology practice. The records of 40 patients (45 ears) with CM seen between 1995 and 1999 inclusive were reviewed. MAIN OUTCOME MEASURES: The series was reviewed with attention to previous medical and otologic history, the nature and duration of symptoms, the physical findings, and management. RESULTS: The prevalence of CM was found to be -1% (approximately one fourth as common as cholesteatoma). Symptoms were often present for many years before the diagnosis of CM, with CM often mistaken for chronic otitis media. Sixty percent of patients had undergone previous otologic procedures. There did not appear to be an association between CM and systemic disease. Physical findings were varied, with granulation tissue and tympanic membrane perforations often occurring transiently. The clinical course of CM is typified by recurrent episodes of symptoms, often interspersed with long asymptomatic periods. A subset of CM can result in an acquired atresia. The most effective treatment appeared to be prolonged topical medications, surgery being reserved for only the most refractory cases. CONCLUSIONS: Chronic myringitis is often mistaken for chronic otitis media. Such confusion prolongs the initiation of appropriate management and sometimes leads to needless tympanomastoid surgery. The otologist should be aware of this clinical entity and its varied presentation.

Are acoustic neuromas encapsulated tumors?

In articles and chapters on the subject of acoustic neuroma, it is almost invariably stated that they are well-encapsulated tumors. During surgical procedures, blunt mechanical dissection defines a natural subsurface cleavage plane that leaves intact a several millimeter thick rind of tumor surface. Occasionally, as a concession to neural integrity, less than complete resection is elected, leaving behind this "capsular" remnant. To clarify the nature of the surface of acoustic neuromas and to test whether this long held description is indeed correct, a microscopic analysis of 10 surgical specimens was performed. A wedge was harvested from the free surface of the tumor in the mid cerebellopontine angle that included a large, undisturbed section of the tumor surface. Histologic analysis showed that for most of the tumor surface only an extremely thin (3 to 5 microm) layer of connective tissue envelops the tumor. Neoplastic Schwann cells, which extend essentially to the margin of the tumor, were found to be somewhat flattened and compressed in the vicinity of the surface. Although acoustic neuromas are surrounded by a continuous layer of connective tissue, it is so exceptionally thin (on average less than the diameter of a red blood cell) that its edge cannot be visualized intraoperatively by a surgeon. Because the pathologic definition of a capsule is a thick, enveloping layer of connective tissue that is both micro- and macroscopically evident, it must be concluded that acoustic neuromas are nonencapsulated, at least in the conventional sense of the term. The surface peel observed intraoperatively is surgically produced during tumor debulking by cleaving of the looser central component from the more compressed portion of neoplastic cells that lies immediately beneath the free margin of the lesion.

CT angiography as a screening tool for dural arteriovenous fistula in patients with pulsatile tinnitus: feasibility and test characteristics.

BACKGROUND AND PURPOSE: The diagnosis of intracranial DAVF with noninvasive cross-sectional imaging such as CTA is challenging. We sought to determine the sensitivity and specificity of CTA compared with cerebral angiography for DAVF in patients presenting with PT. MATERIALS AND METHODS: Following approval of the institutional review board, we reviewed all patients who underwent CTA for PT from 2004 to 2009 and collected clinical and imaging data. Seven patients with PT and proved DAVF and 7 age- and sex-matched control patients with PT but no DAVF composed the study group. CTA images were blindly interpreted by 2 experienced neuroradiologists for the presence of 5 variables: asymmetric arterial feeding vessels, "shaggy" appearance of a dural venous sinus, transcalvarial venous channels, asymmetric venous collaterals, and abnormal size and number of cortical veins. Asymmetric attenuation of jugular veins was additionally assessed. RESULTS: The presence of arterial feeders showed good test characteristics for screening, with a sensitivity of 86% (95% CI, 42-99) and a specificity of 100% (95% CI, 52-100). A shaggy sinus or tentorium was highly specific: sensitivity of 42% (95% CI, 11-79) and specificity of 100% (95% CI, 56-100). The presence of transcalvarial venous channels demonstrated a poor sensitivity of 29% (95% CI, 5-70) but a high specificity 86% (95% CI, 42-99). CT attenuation of the jugular veins showed statistically significant asymmetry in the DAVF group versus the control group (P < .05). CONCLUSIONS: CTA can be used to screen for DAVF in patients with PT. The presence of asymmetrically visible and enlarged arterial feeding vessels has a high sensitivity and specificity for the diagnosis of DAVF.

Routine preoperative imaging in chronic ear surgery.

OBJECTIVE: This article provides an overview of the practice and utility of preoperative radiologic studies in chronic otitis media (COM). DATA SOURCES: A literature search of English language clinical and basic science publications was performed. Major otolaryngology texts were reviewed. Special attention was given to the clinical experience and recommendations of experienced otologic surgeons and radiologists regarding the use of radiologic studies in COM. CONCLUSIONS: There is no single accepted standard for the use of preoperative imaging in uncomplicated COM. Imaging studies, especially computed tomography (CT), can provide information regarding the nature and extent of disease, which may not be apparent on the basis of clinical findings alone. This information may impact the patient's operative management, especially in complex or revision cases. Each clinician must assess the benefits derived from these studies in his or her own practice.