HRCT imaging of acquired cholesteatoma: a pictorial review

A 3D and Explainable Artificial Intelligence Model for Evaluation of Chronic Otitis Media Based on Temporal Bone Computed Tomography: Model Development, Validation, and Clinical Application

BACKGROUND

Temporal bone computed tomography (CT) helps diagnose chronic otitis media (COM). However, its interpretation requires training and expertise. Artificial intelligence (AI) can help clinicians evaluate COM through CT scans, but existing models lack transparency and may not fully leverage multidimensional diagnostic information.

OBJECTIVE

We aimed to develop an explainable AI system based on 3D convolutional neural networks (CNNs) for automatic CT-based evaluation of COM.

METHODS

Temporal bone CT scans were retrospectively obtained from patients operated for COM between December 2015 and July 2021 at 2 independent institutes. A region of interest encompassing the middle ear was automatically segmented, and 3D CNNs were subsequently trained to identify pathological ears and cholesteatoma. An ablation study was performed to refine model architecture. Benchmark tests were conducted against a baseline 2D model and 7 clinical experts. Model performance was measured through cross-validation and external validation. Heat maps, generated using Gradient-Weighted Class Activation Mapping, were used to highlight critical decision-making regions. Finally, the AI system was assessed with a prospective cohort to aid clinicians in preoperative COM assessment.

RESULTS

Internal and external data sets contained 1661 and 108 patients (3153 and 211 eligible ears), respectively. The 3D model exhibited decent performance with mean areas under the receiver operating characteristic curves of 0.96 (SD 0.01) and 0.93 (SD 0.01), and mean accuracies of 0.878 (SD 0.017) and 0.843 (SD 0.015), respectively, for detecting pathological ears on the 2 data sets. Similar outcomes were observed for cholesteatoma identification (mean area under the receiver operating characteristic curve 0.85, SD 0.03 and 0.83, SD 0.05; mean accuracies 0.783, SD 0.04 and 0.813, SD 0.033, respectively). The proposed 3D model achieved a commendable balance between performance and network size relative to alternative models. It significantly outperformed the 2D approach in detecting COM (P≤.05) and exhibited a substantial gain in identifying cholesteatoma (P<.001). The model also demonstrated superior diagnostic capabilities over resident fellows and the attending otologist (P<.05), rivaling all senior clinicians in both tasks. The generated heat maps properly highlighted the middle ear and mastoid regions, aligning with human knowledge in interpreting temporal bone CT. The resulting AI system achieved an accuracy of 81.8% in generating preoperative diagnoses for 121 patients and contributed to clinical decision-making in 90.1% cases.

CONCLUSIONS

We present a 3D CNN model trained to detect pathological changes and identify cholesteatoma via temporal bone CT scans. In both tasks, this model significantly outperforms the baseline 2D approach, achieving levels comparable with or surpassing those of human experts. The model also exhibits decent generalizability and enhanced comprehensibility. This AI system facilitates automatic COM assessment and shows promising viability in real-world clinical settings. These findings underscore AI's potential as a valuable aid for clinicians in COM evaluation.

TRIAL REGISTRATION

Chinese Clinical Trial Registry ChiCTR2000036300; https://www.chictr.org.cn/showprojEN.html?proj=58685.

High-Resolution Computed Tomography and Intraoperative Correlation in Cholesteatoma: Enhancing Preoperative Evaluation and Surgical Management

INTRODUCTION

Cholesteatoma, a hazardous non-neoplastic lesion of the temporal bone, is prevalent in socio-economically disadvantaged groups in developing nations like India. Timely detection and surgical intervention are essential for effective management. High-resolution computed tomography (HRCT) has revolutionized the assessment of temporal bone pathology, though its role in preoperative evaluation remains debated. This study aimed to validate HRCT's utility in diagnosing cholesteatoma, compare its findings with intraoperative observations, and assess sensitivity and specificity.

METHODS

This diagnostic accuracy study was conducted at a tertiary care center in Western India, from March 2021 to November 2022. HRCT findings of 54 cholesteatoma patients were evaluated and compared with intraoperative findings. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and Cohen's kappa coefficient were calculated.

RESULTS

HRCT demonstrated a sensitivity exceeding 90% in identifying scutum erosion, mastoid sclerosis, and abnormalities in the tympanic membrane, along with a specificity surpassing 90% in detecting various conditions, including facial canal erosion, sinus plate erosion, lateral semicircular canal erosion, erosion of the posterior wall of the external auditory canal, and abnormalities in the tympanic membrane. Furthermore, HRCT exhibited an accuracy of over 90% in detecting most pathologies. There was a perfect or near-perfect agreement observed for abnormal tympanic membrane, sinus plate erosion, mastoid sclerosis, and erosion of the posterior wall of the external auditory canal (with kappa values > 0.8). Moderate to fair agreement was noted for other pathologies.

CONCLUSION

HRCT offered precise detection of the majority of pathologies, thereby facilitating surgical planning. However, the presence of limitations in distinguishing specific abnormalities highlights the significance of utilizing HRCT in tandem with other diagnostic modalities to ensure meticulous diagnosis and effective treatment planning.

CT and MRI Features of Hairy Polyps in Neonates and Infants: A Retrospective Study of 14 Patients

Background: The typical imaging findings of hairy polyps have been described mostly in case reports. This study was conducted to describe the CT and MRI features of hairy polyps and their common associated abnormalities. Methods: Medical records of 14 patients with pathological diagnosis of hairy polyps were collected for this study. For each patient, the medical records, including demographics, clinical manifestations, and imaging findings were reviewed. Results: The female-to-male ratio was 3.7:1. The age at first episode varied from birth to 2.7 years. The masses were derived from the back side of the soft palate in seven (50.0%) cases, from the lateral pharyngeal wall in four (28.6%) cases, from the soft palate in one (7.1%) case, from the nasal vestibule in one (7.1%) case, and from the parapharyngeal space in one (7.1%) case. A total of 11 (78.6%) cases presented with pedicled masses containing fat and a central core of soft tissue, there were 3 (21.4%) cases whose imaging findings were atypical, and there were 6 (42.9%) patients who had other pathologies. Conclusions: Hairy polyps typically presented as pedicled masses containing fat and a central core of soft tissue, but sometimes their imaging findings can be atypical and they can be associated with other congenital abnormalities. CT and MRI are reliable methods for the diagnosis of hairy polyps and their associated abnormalities.

Pearls of Temporal Bone Imaging in Children with Hearing Loss

Hearing loss is one of the most common indications for temporal bone imaging in children. Hearing loss may be congenital or acquired, and it may be conductive, sensorineural, or mixed audiologically. Temporal bone imaging plays an important role in the assessment and management of this condition. An understanding of the embryology of ear structures better enables the radiologist to interpret abnormalities on imaging of the temporal bone. Here, we provide a general review of ear development and a description of known genetic defects that contribute to congenital ear anomalies associated with hearing loss. We provide appropriate imaging techniques for the temporal bone depending on the clinical presentation and a systematic approach to imaging for children with hearing loss. Diagnostic imaging for developmental anomalies of the ear and cholesteatoma will be discussed.

Could computed tomography histogram analysis add value to the diagnosis of cholesteatoma?

BACKGROUND

To investigate the potential role of computed tomography (CT) histogram analysis in differentiating cholesteatoma (CHS) and non-cholesteatoma (NCHS).

METHODS

We evaluated 77 temporal bone CT images (from November 2016 to February 2020) that were obtained preoperatively (mean age, 37.10±17.27 years in CHS; 36.72±16.08 years in NCHS group). Histogram analyses of the resulting XML files were conducted using the R Project 3.3.2 program. ROC analysis was used to find threshold values, and the diagnostic efficiency of these values in differentiating CHS-NCHS was determined.

RESULTS

The CT images of 41 CHS (53.25%) and 36 NHCS cases (46.75%) were evaluated. There was a statistically significant difference between the CHS and NCHS group in terms of the mean, maximum, and median values (p = 0.036, p = 0.006, p = 0.043). When examining the ROC curve obtained from the mean of these parameters, area under the curve (AUC) is determined as 0.638, and when the threshold value is selected as 42.55, the mean value was determined to have a sensitivity of 86.50% and specificity of 56.10% in differentiating CHS-NCHS.

Computed-Tomography-Structured Reporting in Middle Ear Opacification: Surgical Results and Clinical Considerations From a Large Retrospective Analysis

Purpose: The aim of the study is to compare the accuracy of unstructured preoperative Computed Tomography (CT) reports from non-tertiary diagnostic centers with intraoperative findings in a large cohort of patients with Chronic Otitis Media (COM) undergone surgery.

Methods: From 2012 to 2019, a total number of 301 patients were considered for our purposes. All patients with clinical evidence of COM had preoperative non-contrast high resolution CT scan of the temporal bone in non-tertiary diagnostic centers, performed within 3 months before surgery.

Results: The accuracy of CT reports was analyzed in terms of nature, anatomical site, disease extension, bony erosion, vascular structures abnormalities relevant to surgical planning, and Eustachian tube patency. Compared to post-surgical findings, CT reporting critical analysis revealed a tendency to overestimation of bony erosion, coupled to underestimated description of facial canal/lateral semi-circular canal, vascular structures, and Eustachian tube.

Conclusion: Discrepancies between CT reports and surgical findings in middle ear opacification can be at least in part due to limited expertise of general radiologists in ENT neuroimaging. To limit this lack of information and the limited accuracy of middle ear structures depiction, here we propose a structured checklist to adopt in the case of a temporal bone CT scan for COM, in order to optimize the communication with surgeons and provide all the crucial information for an accurate surgical planning.

Eyes in Ears: A Miniature Steerable Digital Endoscope for Trans-Nasal Diagnosis of Middle Ear Disease

The aim of this work is to design, fabricate and experimentally validate a miniature steerable digital endoscope that can provide comprehensive, high-resolution imaging of the middle ear using a trans-nasal approach. The motivation for this work comes from the high incidence of middle ear diseases, and the current reliance on invasive surgery to diagnose and survey these diseases which typically consists of the eardrum being lifted surgically to directly visualize the middle ear using a trans-canal approach. To enable less-invasive diagnosis and surveillance of middle ear disease, we propose an endoscope that is small enough to pass into the middle ear through the Eustachian tube, with a steerable tip that carries a 1 Megapixel image sensor and fiber-optic illumination to provide high-resolution visualization of critical middle ear structures. The proposed endoscope would enable physicians to diagnose middle ear disease using a non-surgical trans-nasal approach instead, enabling such procedures to be performed in an office setting and greatly reducing invasiveness for the patient. In this work, the computational design of the steerable tip based on computed tomography models of real human middle ear anatomy is presented, and these results informed the fabrication of a clinical-scale steerable endoscope prototype. The prototype was used in a pilot study in three cadaveric temporal bone specimens, where high-quality middle ear visualization was achieved as determined by an unbiased cohort of otolaryngologists. This is the first paper to demonstrate cadaveric validation of a digital, steerable, clinical-scale endoscope for middle ear disease diagnosis, and the experimental results illustrate that the endoscope enables the visualization of critical middle ear structures (such as the epitympanum or sinus tympani) that were seldom or never visualized in prior published trans-Eustachian tube endoscopy feasibility studies.

Distribution of different morphological types of anterior epitympanic plate “cog” and Köerner’s septum in CT images of cholesteatomatous and non-cholesteatomatous CSOM: is it really significant?

Background

The anterior epitympanic plate (cog) and Köerner’s septum are gaining more importance since the introduction of transcanal mastoidectomy as these anatomical structures are serving landmarks for the attic. Moreover, different morphological types of cog and Köerner’s septum revealed embryological relation to the development of the isthmic membrane; the latter is linked to aeration of the attic and thus affects the pathological development of the cholesteatoma.

Results

A retrospective review of CT images of 86 patients proved by surgical biopsies revealed: 49 cholesteatomas and 37 non-cholesteatomatous CSOM. The type-I “cog” had a higher incidence (n = 40) and was statistically more prevalent in non-cholesteatomatous CSOM (n = 30, 81.1%). Type-II was the second commonest and was statistically more prevalent in cholesteatomatous CSOM (n = 22, 44.9%). Type-III was the third commonest, seen in (n = 18, 20.9%) and was statistically more prevalent in cholesteatomatous CSOM (n = 17, 34.7%). Köerner’s septum was more prevalent in non-cholesteatomatous CSOM (70.3%) with statistical significance (p value = 0.002).

Conclusions

The difference in the distribution of different morphological types of anterior epitympanic plate “cog,” as well as the difference in Köerner’s septum existence amongst cholesteatomatous and non-cholesteatomatous CSOM are suggested as risk factors for the development of cholesteatoma and may predict a cholesteatoma on CT images.

[Not Available]

Objectif

Fournir aux médecins de famille une approche pratique et fondée sur des données probantes pour la prise en charge de la perte auditive.

Sources d’information

Une recherche dans les bases de données MEDLINE et PubMed a relevé les revues de synthèse, recherches et lignes directrices publiées en anglais de 1980 à 2020. Les données probantes étaient de niveau II ou III dans la plupart des articles relevés.

Message principal

La perte auditive est l’une des déficiences sensorielles les plus fréquentes dans le monde, et elle est grandement préjudiciable au bien-être général du patient, affectant sa santé physique, ses finances, son inclusion sociale et sa santé mentale. Une solide évaluation clinique de la perte auditive comprend une anamnèse et un examen physique qui caractérisent efficacement la perte auditive comme étant de transmission, neurosensorielle ou mixte. Les patients qui présentent des signes alarmants (comme une perte auditive neurosensorielle unilatérale soudaine) doivent être aiguillés d’urgence en oto-rhino-laryngologie et chirurgie cervico-faciale ou être évalués immédiatement au service des urgences. Beaucoup de cas non urgents de perte auditive nécessitent également une évaluation audiologique plus poussée, un diagnostic et la prise en charge.

Conclusion

À titre de fournisseurs de soins de première ligne, les médecins de famille sont bien placés pour gérer les préoccupations psychologiques liées à la perte auditive et pour renforcer les stratégies thérapeutiques prudentes. Il est fréquemment nécessaire d’aiguiller le patient ou d’effectuer un bilan urgent, dont l’imagerie, pour confirmer le diagnostic et instaurer la prise en charge afin de prévenir d’autres complications.

Approach to hearing loss

OBJECTIVE

To provide family physicians with a practical evidence-based approach to the management of patients with hearing loss.

SOURCES OF INFORMATION

MEDLINE and PubMed databases were searched for English-language hearing loss research, review articles, and guidelines published between 1980 and 2020. Most of the retrieved articles provided level II or III evidence.

MAIN MESSAGE

Hearing loss is one of the most common sensory impairments worldwide and causes great detriment to a patient's overall well-being by affecting physical health, finances, social inclusion, and mental health. A robust clinical assessment of hearing loss includes a history and physical examination that effectively characterizes the deficit as conductive, sensorineural, or mixed. Patients presenting with red flags (such as sudden unilateral sensorineural hearing loss) must be urgently referred to otolaryngology-head and neck surgery or immediately assessed in the emergency department. Many nonurgent presentations of hearing loss will also require referral for further audiological assessment, diagnosis, and management.

CONCLUSION

As primary care providers, family physicians are well equipped to manage the psychological concerns associated with hearing loss and to reinforce conservative treatment strategies. Frequently, referral or urgent workup, including imaging, is necessary to confirm a patient's diagnosis and initiate management in order to prevent further complications.

MRI of the Internal Auditory Canal, Labyrinth, and Middle Ear: How We Do It

MRI is firmly established as an essential modality in the imaging of the temporal bone and lateral skull base. It is used to evaluate normal anatomic structures, evaluate for vestibular schwannomas, assess for inflammatory and/or infectious processes, and detect residual and/or recurrent cholesteatoma. It is also extensively used in pre- and postoperative evaluations, particularly in patients with vestibular schwannomas and candidates for cochlear implantation. Nevertheless, despite the widespread use of MRI for these purposes, many radiologists remain unfamiliar with the complex anatomy and expected imaging findings with such examinations. The purpose of this review is to provide an overview of the most useful MRI sequences for internal auditory canal and labyrinthine imaging, review the relevant anatomy, and discuss the expected appearances of the most commonly encountered pathologic entities. In addition, the features at pre- and postprocedural MRI will be discussed to help ensure that diagnostic radiologists may be of greatest use to the ordering physicians. © RSNA, 2020.

High-resolution computed tomography (HRCT) in pediatric and adult patients with unsafe chronic suppurative otitis media (CSOM) and its surgical correlation

Background and Aim:

Atticoantral, alias unsafe type of CSOM affects the posterosuperior part of the middle ear cleft and is frequently coupled with complications and bony erosions. This study aimed to correlate the high-resolution computed tomography (HRCT) temporal bone and intraoperative findings in the patients with the unsafe type of CSOM.

Methods:

This prospective study included 50 patients (28 males: 22 females; mean age 24 ± 14 years) who presented with clinically suspected unsafe CSOM. All patients underwent HRCT of the temporal bone and subsequent surgical procedure. The intraoperative and histopathological findings were compared with HRCT findings. Descriptive statistics, sensitivity, specificity, and positive and negative predictive value for HRCT were calculated. Student's t-test and Chi-square test were performed.

Results:

Out of 50 patients, left, right, and bilateral ear involvement were seen in 42% (21/50), 38% (19/50), and 20% (10/50) patients, respectively. Ear discharge was the most common symptom (100%) followed by earache (66%), vertigo (16%), and tinnitus (14%), respectively. Cholesteatoma was reported in 82% (49/60) of ears on HRCT while histopathological and intraoperative evaluation confirmed the diagnosis in 40 out of 49 ears. In 18% (11/60) ears, the cholesteatoma was not diagnosed on HRCT evaluation; however, the intraoperative and histopathological assessment revealed cholesteatoma in six patients while the rest had granulation tissue. For detection of ossicular erosions, tegmen erosions, erosions of facial nerve canal, erosions of sigmoid sinus plate, and erosions of lateral/posterior semicircular canals; HRCT had high sensitivity (86.44%–100%) and specificity (93.33%–100%).

Conclusion:

HRCT has a superb correlation with intraoperative findings and is a valuable tool for preoperative assessment of temporal bone pathologies.