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
Dentin matrix protein 1 (Dmp1) is a highly phosphorylated, extracellular matrix protein that is extensively expressed in bone and teeth but also found in soft tissues, including brain and muscle. However, the functions of Dmp1 in the mice cochlea are unknown. Our study showed that Dmp1 was expressed in auditory hair cells (HCs), with the role of Dmp1 in those cells identified using Dmp1 cKD mice. Immunostaining and scanning electron microscopy of the cochlea at P1 revealed that Dmp1 deficiency in mice resulted in an abnormal stereociliary bundle morphology and the mispositioning of the kinocilium. The following experiments further demonstrated that the cell-intrinsic polarity of HCs was affected without apparent effect on the tissue planer polarity, based on the observation that the asymmetric distribution of Vangl2 was unchanged whereas the Gαi3 expression domain was enlarged and Par6b expression was slightly altered. Then, the possible molecular mechanisms of Dmp1 involvement in inner ear development were explored via RNA-seq analysis. The study suggested that the Fgf23–Klotho endocrine axis may play a novel role in the inner ear and Dmp1 may regulate the kinocilium–stereocilia interaction via Fgf23–Klotho signaling. Together, our results proved the critical role of Dmp1 in the precise regulation of hair bundle morphogenesis in the early development of HCs.
Planar cell polarity (PCP) signalling specifies the orientation of epithelial cells and regulates directional beating of motile cilia of multiciliated epithelial cells. Clinically, defects in cilia function are associated with nasopharyngeal symptoms. The polarity of the nasopharyngeal epithelium is poorly understood. Here, we demonstrated PCP in the nasopharyngeal epithelium. Multiciliated cells were uniformly aligned with their long axis parallel to the tissue axis of the nasopharynx (NP). In addition, PCP proteins exhibited an asymmetrical localisation between adjacent cells. Motile cilia were uniformly aligned in the same direction within both individual cells and neighbouring cells, which manifested as cilial polarity in multiciliated cells. Mutation of Vangl2, a mammalian homologue of the Drosophila PCP gene, resulted in significant disruption of the orientation of epithelial cells. Finally, keratin‐5‐positive basal cells constantly replenished the luminal ciliated cells; the new dynamic ciliated cells were also oriented parallel to the tissue axis. These results indicate a role for the PCP pathway in the uniform orientation of dynamically replenished epithelial cells in the NP.
Objective: The objective of this study is to describe the clinical features, managements and outcomes of a rare coexistence of congenital ossicular anomaly and localized cholesteatoma. A literature review on these cases and each congenital disorder is also presented. Methods: A retrospective chart review was performed on patients diagnosed with congenital ossicular anomaly with concurrent localized cholesteatoma from 2008 to 2017. Clinical data of these patients were collected. Results: A total of 10 patients were identified. All patients presented with unilateral hearing loss. Pure-tone audiometry showed conductive hearing loss in all affected ears with an average air conduction (AC) threshold of 59 dB. High-resolution computed tomography scans of the temporal bone diagnosed ossicular anomaly for 90% (9/10); however, only 50% (5/10) had a diagnosis of localized cholesteatoma. A transcanal exploratory tympanotomy under the microscope was performed to discover whether the localized tiny-sized cholesteatoma around the ossicular chain did not have direct contact with the ossicular chain, which could be diagnosed as congenital cholesteatoma. We removed the localized cholesteatoma and reconstructed the ossicular chain in each patient. All localized cholesteatomas were found in the posterior-superior quadrant of the middle ear. Ossicular chain anomalies were associated with the incus and/or the stapes in all cases. Hearing improvement was achieved in each of the 6 patients who were followed up postoperatively, with an average AC threshold of 35 dB. The clinical features of congenital ossicular anomaly with concurrent congenital cholesteatoma were compared with those of each congenital disorder. The pathogenesis of each condition was also discussed. Conclusions: Congenital ossicular anomaly with concurrent congenital cholesteatoma is rare. It shares similar clinical features with congenital ossicular anomaly occurring alone, therefore awareness should be raised for a possible concurrent congenital cholesteatoma which was easy to miss in the diagnosis (50%) by the radiologist. A patient’s hearing level can be improved by removal of the cholesteatoma and reconstruction of the ossicular chain. Localized cholesteatoma does not usually show residuals or recurrence.
As part of the inner ear, the vestibular system is responsible for sense of balance, which consists of three semicircular canals, the utricle, and the saccule. Increasing evidence has indicated that the noncanonical Wnt/PCP signaling pathway plays a significant role in the development of the polarity of the inner ear. However, the role of canonical Wnt signaling in the polarity of the vestibule is still not completely clear. In this study, we found that canonical Wnt pathway-related genes are expressed in the early stage of development of the utricle and change dynamically. We conditionally knocked out β-catenin, a canonical Wnt signaling core protein, and found that the cilia orientation of hair cells was disordered with reduced number of hair cells in the utricle. Moreover, regulating the canonical Wnt pathway (Licl and IWP2) in vitro also affected hair cell polarity and indicated that Axin2 may be important in this process. In conclusion, our results not only confirm that the regulation of canonical Wnt signaling affects the number of hair cells in the utricle but also provide evidence for its role in polarity development.
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. CLINICALTRIAL Chinese Clinical Trial Registry ChiCTR2000036300; https://www.chictr.org.cn/showprojEN.html?proj=58685