The usefulness of narrow-band Imaging (NBI) in nasopharyngeal lesions-Validation of the Ni NBI classification dedicated for vascular pattern in the nasopharynx.

BACKGROUND: This study aims to explore the applicability of narrow-band imaging (NBI) involving the Ni classification for the diagnosis of nasopharyngeal mucosal lesions in order to distinguish malignant tumours (NPT) from non-malignant lesions. METHODS: Each patient (n = 53) with a suspected nasopharyngeal lesion underwent a trans-nasal flexible video endoscopy with an optical filter for NBI. We assessed the suspected area using white light imaging (WLI) in terms of location and morphology as well as the vascular pattern (using Ni classification of nasopharyngeal microvessels) and surrounding tissue by using NBI. Based on the results of the NBI and WLI, patients were classified into "positive" or "negative" groups. All lesions of the nasopharynx were biopsied and submitted for final histological evaluation. RESULTS: NBI showed higher sensitivity, specificity, and accuracy than WLI. There was a significant correlation between the final histological result and the NBI pattern of the NPT: Chi2(1) = 31.34; p = 0.000001 and the WLI assessment of the NPT: Chi2(1) = 14.78; p = 0.00012. CONCLUSIONS: The assessment of the NPT in NBI using Ni NBI classification proved valuable in suspected mucosa assessment. NBI not only confirms the suspicious areas in WLI, but it also shows microlesions beyond the scope of WLI and allows for proper sampling.

Effects of adenoid hypertrophy on nasopharyngeal airway ventilation: A computational fluid dynamics study.

OBJECTIVES: Adenoid hypertrophy causes impaired nasopharyngeal airways (NA) ventilation. However, it is difficult to evaluate the ventilatory conditions of NA. Therefore, this study aimed to analyze the nasopharyngeal airway resistance (NARES) based on computational fluid dynamics simulations and the nasopharyngeal airway depth (NAD) and adenoid hypertrophy grade measured on cephalometric cone-beam computed tomography images and determine the relationship between NAD and grade and NARES to ultimately assess using cephalometric measurements whether NA has airway obstruction defects. METHODS: Cephalogram images were generated from cone-beam computed tomography data of 102 children (41 boys; mean age: 9.14 ± 1.43 years) who received orthodontic examinations at an orthodontic clinic from September 2012 to March 2023, and NAD and adenoid grade and NARES values were measured based on computational fluid dynamics analyses using a 3D NA model. Nonlinear regression analyses were used to evaluate the relationship between NARES and NAD and correlation coefficients to evaluate the relationship between grade and NARES. RESULTS: NARES was inversely proportional to the cube of NAD (R2 = 0.786, P < 0.001), indicating a significant relationship between these variables. The resistance NARES increased substantially when the distance NAD was less than 5 mm. However, adenoid Grade 4 (75 % hypertrophy) was widely distributed. CONCLUSIONS: These study findings demonstrate that the ventilatory conditions of NA can be determined based on a simple evaluation of cephalogram images. An NAD of less than 5 mm on cephalometric images results in NA obstruction with substantially increased airflow resistance.

Comparative cephalometric study of the airways between different ethnic groups with normal occlusion.

OBJECTIVE: This study aimed to compare the nasopharynx and oropharynx airway dimensions of Caucasians, Blacks, Japanese, Japanese Brazilians, and Black Caucasians. METHODS: A sample of 216 lateral radiographs of untreated young Brazilian subjects (mean age of 12.94 years; SD 0.88) were divided into five groups: Black Caucasian, Black, Caucasian, Japanese, and Japanese Brazilian. Lateral radiographs were used to measure the oropharynx (from the midpoint on the soft palate to the closest point on the anterior pharyngeal wall) and the nasopharynx (from the intersection of the posterior border of the tongue and the inferior border of the mandible to the closest point on the posterior pharyngeal wall). Analyses of variance (ANOVA) and Tukey's test were performed (p< 0.05). RESULTS: The linear dimension of the oropharynx was similar among the different ethnic groups. Caucasian individuals presented a significantly greater linear dimension of the nasopharynx than Black Caucasian and Black individuals. CONCLUSIONS: All the groups had similar buccopharyngeal values. However, Caucasian individuals had significantly higher values when compared to Black Caucasians and Black individuals.

An unusual case of long-standing retained nasopharyngeal foreign body.

The nasopharynx is a rare anatomical location where a foreign body may become lodged after being ingested or inhaled. We are presenting a rare case of nasopharyngeal foreign body impaction in a two-and-a-half-year-old child that had been missed for almost a year. The child presented with a history of right-sided foul-smelling nasal discharge, snoring and mouth breathing. An X-Ray soft tissue lateral view of the post-nasal space showed an irregular partially radiopaque nasopharyngeal foreign body. The removal of the foreign body was performed under general anaesthesia successfully. Foreign body impaction in the nasopharynx can easily be missed and it is important to keep this region in mind when dealing with missing inhaled or ingested foreign bodies.

Non-brachycephalic dogs with stertor have higher nasopharyngeal collapsibility compared with dogs without stertor and foramen lacerum level is ideal for evaluating nasopharyngeal collapse on dynamic CT.

Stertor, a clinical sign associated with obstructive airway syndrome, is often observed in non-brachycephalic dogs. This prospective, case-control study aimed to compare soft palate dimensions, nasopharyngeal cross-sectional area (CSA), and nasopharyngeal collapsibility at various locations in non-brachycephalic dogs with and without stertor. A total of 50 dogs were recruited and stratified into control (n = 34) and stertor (n = 13) groups. Static and dynamic computed tomography was conducted without tracheal intubation, and the following variables were calculated: normalized soft palate length and thickness, normalized maximum and minimum nasopharyngeal CSAs (rCSAmax and rCSAmin), and nasopharyngeal collapsibility at the level of the cranial end of the soft palate, pterygoid hamulus, foramen lacerum, bony labyrinth, and caudal end of the soft palate. The stertor group demonstrated significantly lower rCSAmax and rCSAmin, as well as higher nasopharyngeal collapsibility compared with the control group, while no significant differences were noted in the soft palate dimension. Evaluating nasopharyngeal collapse at the foramen lacerum level was recommended due to the clear presence of identifiable bony landmarks and lower overlap in the nasopharyngeal collapsibility between dogs with and without stertor. Physical dimensions of the soft palate may not be the primary contributing factor to nasopharyngeal collapse and clinical signs in non-brachycephalic dogs.

3D analysis of upper airway morphology related to obstructive sleep apnea severity.

BACKGROUND: Obstructive sleep apnea (OSA) is the most common sleep-related breathing disorder characterized by repetitive collapse of the upper airway during sleep. However, little evidence is available on the differences between the sub-regions of the upper airway morphology and OSA severity. Since orthodontists frequently perform cone beam computed tomography (CBCT) in the neck area, we aimed to investigate the relationship and the differences between upper airway morphology and OSA severity using CBCT. MATERIALS AND METHODS: The medical records, CBCT imaging of 21 OSA patients diagnosed by polysomnography, and the apnea-hypopnea index (AHI) results were included to classify OSA severity as mild, moderate, or severe. The minimum cross-sectional areas (MCA) and volumes of the upper pharyngeal airway boundaries in four sub-regions: nasopharynx, retropalatal, retroglossal, and hypopharynx were measured. Dolphin Imaging software was used for upper airway segmentation. The correlation coefficient (r), one-way ANOVA, and the least significant difference post hoc multiple comparison test were applied to fulfill the objectives. RESULTS: A statistically significant relationship was found between the MCA of the nasopharynx and the AHI (r = -0.473, P < 0.05). Furthermore, a difference was found between mild and moderate and moderate and severe OSA severity in the MCA results of the retroglossal region (P < 0.05). However, no relationship was found between the upper airway volume and OSA severity. CONCLUSIONS: MCA was moderately negatively correlated to AHI only in the nasopharynx subregion. Moderate OSA presented significantly less MCA than mild and severe OSA only in the oropharynx and retroglossal subregions.

Large Nasopharyngeal Hairy Polyp Causing Immediate Respiratory Distress at Birth.

This article presents a rare case of a large hairy polyp, a developmental malformation causing a benign tumor, within the nasopharynx. The patient, born with the polyp obstructing the airway, required immediate intubation and a combined transnasal-transoral surgical approach for excision. The case underscores the challenges in diagnosing and managing such polyps, emphasizing the importance of imaging for surgical planning, and the consideration of multiple approaches to ensure complete resection and prevent recurrence. Laryngoscope, 134:4118-4121, 2024.

Medium-term effects of rapid maxillary expansion on nasal cavity and pharyngeal airway volumes considering age as a factor: A retrospective study.

OBJECTIVES: The medium-term effects of rapid maxillary expansion (RME) on nasal cavity (NC) and upper airway (UA) dimensions based on chronological age are still unclear. This retrospective study evaluated the medium-term changes occurring in the NC and pharyngeal airways (PA) after RME in two distinct age-based cohorts of patients. METHODS: This retrospective study included 48 subjects who underwent RME grouped in two cohorts: a 6-9-year-old group (EEG group: early expansion group - 25 subjects) and an 11-14-year-old group (LEG group: late expansion group - 23 subjects). NC and PA volumes were analyzed from CBCT imaging segmentation before RME (T0) and twelve months after RME (T1). The amount of maxillary expansion (PW) and minimal cross-sectional area (CSmin) were also considered. RESULTS: All PAs' volumetric sub-regions, CSmin and PW showed a significant volumetric increment (p < 0.05). Inter-group comparisons showed significant differences (p < 0.05) for nasopharynx and CSmin parameters (p < 0.05), while no significant changes were recorded for the other UA's sub-regions and PW (p > 0.05). According to a deviation analysis, part of the UA increase (more marked for the nasopharynx area) may have occurred due to reduced adenotonsillar tissues, which were larger in the EEG group. CONCLUSIONS: Twelve months after treatment, clinicians should not expect changes in the UAs dimensions to be solely related to treatment effects of RME; instead, normal craniofacial growth changes and spontaneous regression of the adenotonsillar tissue could represent the most significant factors influencing UAs changes. CLINICAL SIGNIFICANCE: From the clinical perspective, the results of the present study encourage caution when considering the therapeutic effects of RME on airways dimensions.

Magnetic resonance imaging based on radiomics for differentiating T1-category nasopharyngeal carcinoma from nasopharyngeal lymphoid hyperplasia: a multicenter study.

PURPOSE: To investigate the role of magnetic resonance imaging (MRI) based on radiomics using T2-weighted imaging fat suppression (T2WI-FS) and contrast enhanced T1-weighted imaging (CE-T1WI) sequences in differentiating T1-category nasopharyngeal carcinoma (NPC) from nasopharyngeal lymphoid hyperplasia (NPH). MATERIALS AND METHODS: This study enrolled 614 patients (training dataset: n = 390, internal validation dataset: n = 98, and external validation dataset: n = 126) of T1-category NPC and NPH. Three feature selection methods were used, including analysis of variance, recursive feature elimination, and relief. The logistic regression classifier was performed to construct the radiomics signatures of T2WI-FS, CE-T1WI, and T2WI-FS + CE-T1WI to differentiate T1-category NPC from NPH. The performance of the optimal radiomics signature (T2WI-FS + CE-T1WI) was compared with those of three radiologists in the internal and external validation datasets. RESULTS: Twelve, 15, and 15 radiomics features were selected from T2WI-FS, CE-T1WI, and T2WI-FS + CE-T1WI to develop the three radiomics signatures, respectively. The area under the curve (AUC) values for radiomics signatures of T2WI-FS + CE-T1WI and CE-T1WI were significantly higher than that of T2WI-FS (AUCs = 0.940, 0.935, and 0.905, respectively) for distinguishing T1-category NPC and NPH in the training dataset (Ps all < 0.05). In the internal and external validation datasets, the radiomics signatures based on T2WI-FS + CE-T1WI and CE-T1WI outperformed T2WI-FS with no significant difference (AUCs = 0.938, 0.925, and 0.874 for internal validation dataset and 0.932, 0.918, and 0.882 for external validation dataset; Ps > 0.05). The radiomics signature of T2WI-FS + CE-T1WI significantly performed better than three radiologists in the internal and external validation datasets. CONCLUSION: The MRI-based radiomics signature is meaningful in differentiating T1-category NPC from NPH and potentially helps clinicians select suitable therapy strategies.

Estimation of Nasal Airway Cross-sectional Area From Endoscopy Using Depth Maps: A Proof-of-Concept Study.

OBJECTIVE: Endoscopy is routinely used to diagnose obstructive airway diseases. Currently, endoscopy is only a visualization technique and does not allow quantification of airspace cross-sectional areas (CSAs). This pilot study tested the hypothesis that CSAs can be accurately estimated from depth maps created from virtual endoscopy videos. STUDY DESIGN: Cross-sectional. SETTING: Academic tertiary medical center. METHODS: Virtual endoscopy and depth map videos of the nasal cavity were digitally created based on anatomically accurate three-dimensional (3D) models built from computed tomography scans of 30 subjects. A software tool was developed to outline the airway perimeter and estimate the airspace CSA from the depth maps. Two otolaryngologists used the software tool to estimate the nasopharynx CSA and the nasal valve minimal CSA (mCSA) in the left and right nasal cavities. Model validation statistics were performed. RESULTS: Nasopharynx CSA had a median percent error of 3.7% to 4.6% when compared to the true values measured in the 3D models. Nasal valve mCSA had a median percent error of 22.7% to 33.6% relative to the true values. Raters successfully used the software tool to identify subjects with nasal valve stenosis (ie, mCSA < 0.20 cm2) with a sensitivity of 83.3%, specificity ≥ 90.7%, and classification accuracy ≥ 90.0%. Interrater and intrarater agreements were high. CONCLUSION: This study demonstrates that airway CSAs in 3D models can be accurately estimated from depth maps. The development of artificial intelligence algorithms to compute depth maps may soon allow the quantification of airspace CSAs from clinical endoscopies.

Reliability of fluoroscopic examination of nasopharyngeal dorsoventral dimension change in pugs and French bulldogs.

OBJECTIVE: To compare intra- and interobserver agreements in two-dimensional measurements of changes in nasopharyngeal dimensions during breathing in pugs and French bulldogs. STUDY DESIGN: Experimental randomized study. ANIMALS: A total of 20 French bulldogs and 16 pugs. METHODS: Four observers with different levels of experience measured the dorsoventral dimensions of the nasopharynx during inspiration and expiration on fluoroscopy videos. Measurements were performed at the maximal narrowing of the nasopharynx for the functional method and at the level of the tip of the epiglottis for the anatomically adjusted method. The intra- and interobserver agreements of the measurements, ratio of the dynamic nasopharyngeal change (ΔL), and grade of nasopharyngeal (NP) collapse (no, partial or complete) were evaluated. RESULTS: The functional method resulted in intraobserver correlation coefficients of 0.532 (p < .01) and 0.751 (p < .01) and interobserver correlation coefficients of 0.378 (p < .01) and 0.621 (p < .01) for NP collapse grade and ΔL, respectively. The anatomically adjusted method, 0.491 (p < .01) and 0.576 (p < .01) and 0.495 (p < .01) and 0.729 (p < .01) for NP collapse grade and ΔL, respectively, were being used. One observer (radiologist) achieved intraobserver correlation coefficients >0.9 for both methods. CONCLUSION: Fair interobserver agreement was found for NP collapse grade (functional method), moderate intra- and interobserver agreements were found for NP collapse grade and ΔL (both methods) while intraobserver agreement for ΔL was good (functional method). CLINICAL SIGNIFICANCE: Both methods seem repeatable and reproducible but only for experienced radiologists. The use of ΔL may offer higher repeatability and reproducibility than grade of NP collapse regardless of the method used.

A Case of Nasopharyngeal Pleomorphic Lipoma.

We describe the presentation and treatment of the first reported case of a nasopharyngeal pleomorphic lipoma. The mass was successfully treated with a combined endoscopic trans-oral surgical excision approach by using low temperature-controlled plasma technology, resulting in optimal patient outcomes. Laryngoscope, 134:2710-2712, 2024.

Evaluation of pharyngeal airway volume three-dimensionally in various sagittal skeletal patterns - Systematic review.

To investigate the pharyngeal airway volume in different anteroposterior skeletal malocclusions. This study was prepared according to the Cochrane criteria for creating a systematic review and meta-analysis and confirms the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement. This search was conducted to answer the patient/population, intervention, comparison and outcomes (PICO) question: To evaluate (outcome) and compare (comparison) whether there is a difference in airway volume in patients (participants) with different skeletal malocclusions diagnosed using 3D data (intervention).The first two authors extracted the data from the included studies and assessed the risk of bias in the individual studies using the Newcastle-Ottawa scale. Meta-analysis was done using STATA version 16, which compared various three-dimensional pharyngeal airway parameters in skeletal Class II and skeletal Class III malocclusions with that of skeletal class I malocclusion. Out of 370 articles from the initial search, 17 articles were included in the systematic review. Out of 17 studies, 12 eligible studies were included in the quantitative synthesis. The nasopharynx, oropharynx, hypopharynx, and total airway volume were increased in skeletal Class I malocclusion compared to that of skeletal Class II malocclusion and decreased in comparison to skeletal Class III malocclusion. The moderate quality of evidence indicates the total airway volume, oropharynx, and hypopharynx are largest in skeletal Class III compared to Class I and Class II skeletal malocclusion.

[Application of virtual endoscopy in the diagnosis of adenoid hypertrophy and the morphologic classification of adenoid].

Objective:To discuss the application of virtual endoscopy in the diagnosis of adenoid hypertrophy and the morphologic classification of adenoid. Methods:The clinical data of 97 children with adenoid hypertrophy admitted to Department of Otorhinolaryngology Head and Neck Surgery, Shenzhen University General Hospital from July 2022 to December 2022 were collected. The virtual endoscopic reconstruction of the nasopharynx was performed by cone beam computed tomography. The results of virtual endoscopic adenoid size measurement were compared with the results of nasopharyngeal CT median sagittal position and nasopharyngeal endoscopy. Virtual endoscopic classification of adenoid based on the size of the adenoids and their relationship with the torus tubarius. Results:The t-test results of the size of adenoids measured by virtual endoscopy and nasopharyngeal CT were t=1.699 and P=0.093, and the results of intra-group correlation coefficient（ICC） analysis were ICC=0.921 and P<0.01. The proportion of adenoids measured by virtual endoscopy and nasopharyngeal CT was highly consistent. The t-test results of the size of adenoids measured virtual endoscopy and nasopharyngeal endoscopy were t=1.543 and P=0.15, and the results of intra-group correlation coefficient（ICC） analysis were ICC=0.900 and P<0.01. The proportion of adenoids measured by virtual endoscopy and nasopharyngeal endoscopy was highly consistent. Among the 97 children, the morphological classification results of adenoids were 48 cases of overall hypertrophy type, 47 cases of central bulge type, and 2 cases of flat thickening type. Conclusion:The diagnosis of adenoid hypertrophy by virtual endoscopy has high accuracy, which not only avoids the invasive operation of traditional nasopharyngeal endoscopy, but also can observe the adenoid condition and its relationship with the torus tubarius from multiple angles. And, the morphological classification of adenoids using virtual endoscopy has guiding significance for perioperative preparation.

Multi-stage Unet segmentation and automatic measurement of pharyngeal airway based on lateral cephalograms.

OBJECTIVES: Orthodontic treatment profoundly impact the pharyngeal airway (PA) of patients. Airway examination is an integral part of daily orthodontic diagnosis, and lateral cephalograms (LC) are reliable to reveal PA structures. This study attempted to develop a simple method to help clinicians make a preliminary judgement of patients' PA conditions and assess the impact of orthodontic treatment on their airways. METHODS: LCs of 764 patients were used to train a multistage unit segmentation model. Another 130 images were used to validate the model and more 130 images were used to test the model. RESULTS: Unet was used as the backbone, with a mean dice value of 0.8180, precision of 0.8393, and recall of 0.8188. Furthermore, we identified seven key points and measured related indices. The length of the line separating the nasopharynx and oropharynx and the line separating the oropharynx and hypopharynx were manually measured thrice and the average values was compared. The intraclass correlation coefficient (ICC) for the two lines was 0.599 and 0.855. Then, we performed a single linear regression analysis, which indicated a strong correlation between the predictions and measurements for the two lines. CONCLUSIONS: This method is reliable for segmenting three regions (nasopharynx, oropharynx, and hypopharynx) of the PA and calculating related indices. However, the predictions obtained from this model still have errors, and it is necessary for clinical practitioners to assess and adjust the predictions. CLINICAL SIGNIFICANCE: Our model can help orthodontists formulate personalised treatment plans and evaluate the risk of airway stenosis during orthodontic treatment. This method may mark the beginning of a new and simpler approach for PA obstruction detection, specifically tailored to orthodontic patients.

Comparative study of optical coherence tomograph and histological images of eustachian tube nasopharyngeal region and adjacent structures in vivo and ex-vivo miniature pigs.

OBJECTIVES: Optical Coherence Tomograph (OCT) imaging technology can be used to examine, in vivo, the human ET. At present, it is impossible to achieve the OCT scanning vivo and ex vivo in the same individual human body, or study the consistency between OCT images and histological images of the eustachian tube nasopharyngeal region and adjacent structures. The aim of this study was to determine the consistency between OCT images and histological sections in vivo and ex vivo in miniature pigs. METHODS: OCT imaging was performed on five adult miniature pigs in vivo and ex vivo. The images of the eustachian tube OCT (ET-OCT), nasopharynx OCT (NP-OCT) and histological cross sections were further studied. RESULTS: All five miniature pigs achieved the OCT scan successfully, acquiring ET-OCT and NP-OCT images in vivo and ex vivo on both sides. The acquired ET OCT images closely matched the histological images, revealing details of the cartilage, submucosa, glands, and mucosa. The lower segment of the ET wall mucosa had an abundance of glands and submucosal tissues, with more low-signal areas appearing in the ex vivo images. The NP-OCT images of the nasopharynx matched the details of the mucosa and submucosal tissues. The ex-vivo OCT images showed thicker mucosa and more scattered slightly lower signal areas compared to the vivo OCT images. CONCLUSIONS: ET-OCT images and NP-OCT images matched the histological structure of eustachian tube nasopharyngeal region structures in miniature pigs both in vivo and ex vivo. OCT images may be sensitive to changes in edema and ischemia status. There is a great potential for morphological assessment of inflammation, edema, injure, mucus gland status.

Effect of different degrees of adenoid hypertrophy on pediatric upper airway aerodynamics: a computational fluid dynamics study.

To improve the diagnostic accuracy of adenoid hypertrophy (AH) in children and prevent further complications in time, it is important to study and quantify the effects of different degrees of AH on pediatric upper airway (UA) aerodynamics. In this study, based on computed tomography (CT) scans of a child with AH, UA models with different degrees of obstruction (adenoidal-nasopharyngeal (AN) ratio of 0.9, 0.8, 0.7, and 0.6) and no obstruction (AN ratio of 0.5) were constructed through virtual surgery to quantitatively analyze the aerodynamic characteristics of UA with different degrees of obstruction in terms of the peak velocity, pressure drop (△P), and maximum wall shear stress (WSS). We found that two obvious whirlpools are formed in the anterior upper part of the pediatric nasal cavity and in the oropharynx, which is caused by the sudden increase in the nasal cross-section area, resulting in local flow separation and counterflow. In addition, when the AN ratio was ≥ 0.7, the airflow velocity peaked at the protruding area in the nasopharynx, with an increase 1.1-2.7 times greater than that in the nasal valve area; the △P in the nasopharynx was significantly increased, with an increase 1.1-6.8 times greater than that in the nasal cavity; and the maximum WSS of the posterior wall of the nasopharynx was 1.1-4.4 times larger than that of the nasal cavity. The results showed that the size of the adenoid plays an important role in the patency of the pediatric UA.

Automatic detection and recognition of nasopharynx gross tumour volume (GTVnx) by deep learning for nasopharyngeal cancer radiotherapy through magnetic resonance imaging.

BACKGROUND: In this study, we propose the deep learning model-based framework to automatically delineate nasopharynx gross tumor volume (GTVnx) in MRI images. METHODS: MRI images from 200 patients were collected for training-validation and testing set. Three popular deep learning models (FCN, U-Net, Deeplabv3) are proposed to automatically delineate GTVnx. FCN was the first and simplest fully convolutional model. U-Net was proposed specifically for medical image segmentation. In Deeplabv3, the proposed Atrous Spatial Pyramid Pooling (ASPP) block, and fully connected Conditional Random Field(CRF) may improve the detection of the small scattered distributed tumor parts due to its different scale of spatial pyramid layers. The three models are compared under same fair criteria, except the learning rate set for the U-Net. Two widely applied evaluation standards, mIoU and mPA, are employed for the detection result evaluation. RESULTS: The extensive experiments show that the results of FCN and Deeplabv3 are promising as the benchmark of automatic nasopharyngeal cancer detection. Deeplabv3 performs best with the detection of mIoU 0.8529 ± 0.0017 and mPA 0.9103 ± 0.0039. FCN performs slightly worse in term of detection accuracy. However, both consume similar GPU memory and training time. U-Net performs obviously worst in both detection accuracy and memory consumption. Thus U-Net is not suggested for automatic GTVnx delineation. CONCLUSIONS: The proposed framework for automatic target delineation of GTVnx in nasopharynx bring us the desirable and promising results, which could not only be labor-saving, but also make the contour evaluation more objective. This preliminary results provide us with clear directions for further study.

MSU-Net: Multi-scale Sensitive U-Net based on pixel-edge-region level collaborative loss for nasopharyngeal MRI segmentation.

Radiotherapy is the traditional treatment of early nasopharyngeal carcinoma (NPC). Automatic accurate segmentation of risky lesions in the nasopharynx is crucial in radiotherapy. U-Net has been proved its effective medical image segmentation ability. However, the great difference in the structure and size of nasopharynx among different patients requires a network that pays more attention to multi-scale information. In this paper, we propose a multi-scale sensitive U-Net (MSU-Net) based on pixel-edge-region level collaborative loss (LCo-PER) for NPC segmentation task. A series of novel feature fusion modules based on spatial continuity and multi-scale semantic are proposed for extracting multi-level features while efficiently searching for all size lesions. A spatial continuity information extraction module (SCIEM) is proposed for effectively using the spatial continuity information of context slices to search small lesions. And a multi-scale semantic feature extraction module (MSFEM) is proposed for extracting features of different receptive fields. LCo-PER is proposed for the network training which makes network model could take into account the size of different lesions. The global Dice, Precision, Recall and IOU of the testing set are 84.50%, 97.48%, 84.33% and 82.41%, respectively. The results show that our method is better than the other state-of-the-art methods for NPC segmentation which obtain higher accuracy and effective segmentation performance.