ABSTRACT
PURPOSE: To compare the calibre of the cochlear (CN), superior vestibular (SVN) and inferior vestibular (IVN) nerves on magnetic resonance imaging (MRI), both between Ménière's Disease (MD) ears and clinical controls, and between inner ears with and without endolymphatic hydrops (EH) on MRI. METHODS: A retrospective case-control study evaluated patients undergoing MRI for suspected hydropic ear disease from 9/2017 to 8/2022. The CN, SVN, IVN and facial nerve (FN) diameters and cross-sectional areas (CSA) were measured on T2-weighted sequences whilst EH was evaluated on delayed post-gadolinium MRI. Absolute nerve calibre (and that relative to the FN) in unilateral definite MD ears (2015 Barany criteria) was compared to that in both asymptomatic contralateral ears and clinical control ears. Nerve calibre in ears with severe cochlear and vestibular EH was compared to ears without EH. t tests or Wilcoxon signed-rank test/Mann-Whitney U test were applied (p < 0.001). RESULTS: 173 patients (mean age 51.3 ± 15.1, 65 men) with 84 MD (62 unilateral) and 62 clinical control ears were studied. Absolute and relative CN dimensions were decreased in both MD ears (CSA and diameter) and the contralateral asymptomatic ears (CSA) when compared to clinical controls (p < 0.001). Absolute nerve dimensions were reduced in both severe vestibular EH (CN, IVN and SVN) and severe cochlear EH (CN) (p < 0.001), however this was not evident when adjusted according to facial nerve calibre. CONCLUSION: There is decreased absolute CN calibre in both symptomatic and asymptomatic MD ears as well as ears with severe cochlear and vestibular EH on MRI.
ABSTRACT
Automatic segmentation of vestibular schwannoma (VS) from routine clinical MRI has potential to improve clinical workflow, facilitate treatment decisions, and assist patient management. Previous work demonstrated reliable automatic segmentation performance on datasets of standardized MRI images acquired for stereotactic surgery planning. However, diagnostic clinical datasets are generally more diverse and pose a larger challenge to automatic segmentation algorithms, especially when post-operative images are included. In this work, we show for the first time that automatic segmentation of VS on routine MRI datasets is also possible with high accuracy. We acquired and publicly release a curated multi-center routine clinical (MC-RC) dataset of 160 patients with a single sporadic VS. For each patient up to three longitudinal MRI exams with contrast-enhanced T1-weighted (ceT1w) (n = 124) and T2-weighted (T2w) (n = 363) images were included and the VS manually annotated. Segmentations were produced and verified in an iterative process: (1) initial segmentations by a specialized company; (2) review by one of three trained radiologists; and (3) validation by an expert team. Inter- and intra-observer reliability experiments were performed on a subset of the dataset. A state-of-the-art deep learning framework was used to train segmentation models for VS. Model performance was evaluated on a MC-RC hold-out testing set, another public VS datasets, and a partially public dataset. The generalizability and robustness of the VS deep learning segmentation models increased significantly when trained on the MC-RC dataset. Dice similarity coefficients (DSC) achieved by our model are comparable to those achieved by trained radiologists in the inter-observer experiment. On the MC-RC testing set, median DSCs were 86.2(9.5) for ceT1w, 89.4(7.0) for T2w, and 86.4(8.6) for combined ceT1w+T2w input images. On another public dataset acquired for Gamma Knife stereotactic radiosurgery our model achieved median DSCs of 95.3(2.9), 92.8(3.8), and 95.5(3.3), respectively. In contrast, models trained on the Gamma Knife dataset did not generalize well as illustrated by significant underperformance on the MC-RC routine MRI dataset, highlighting the importance of data variability in the development of robust VS segmentation models. The MC-RC dataset and all trained deep learning models were made available online.
ABSTRACT
OBJECTIVES: To evaluate the diagnostic performance and reliability of MRI descriptors used for the detection of Ménière's disease (MD) on delayed post-gadolinium MRI. To determine which combination of descriptors should be optimally applied and whether analysis of the vestibular aqueduct (VA) contributes to the diagnosis. MATERIALS AND METHODS: This retrospective single centre case-control study evaluated delayed post-gadolinium MRI of patients with Ménièriform symptoms examined consecutively between Dec 2017 and March 2023. Two observers evaluated 17 MRI descriptors of MD and quantified perilymphatic enhancement (PLE) in the cochlea. Definite MD ears according to the 2015 Barany Society criteria were compared to control ears. Cohen's kappa and diagnostic odds ratio (DORs) were calculated for each descriptor. Forward stepwise logistic regression determined which combination of MRI descriptors would best predict MD ears, and the area under the receiver operating characteristic curve for this model was measured. RESULTS: A total of 227 patients (mean age 48.3 ± 14.6, 99 men) with 96 definite MD and 78 control ears were evaluated. The presence of saccular abnormality (absent, as large as or confluent with the utricle) performed best with a DOR of 292.6 (95% confidence interval (CI), 38.305-2235.058). All VA descriptors demonstrated excellent reliability and with DORs of 7.761 (95% CI, 3.517-17.125) to 18.1 (95% CI, 8.445-39.170). Combining these saccular abnormalities with asymmetric cochlear PLE and an incompletely visualised VA correctly classified 90.2% of cases (sensitivity 84.4%, specificity 97.4%, AUC 0.938). CONCLUSION: Either absent, enlarged or confluent saccules are the best predictors of MD. Incomplete visualisation of the VA adds value to the diagnosis. CLINICAL RELEVANCE STATEMENT: A number of different MRI descriptors have been proposed for the diagnosis of Ménière's disease, but by establishing the optimally performing MRI features and highlighting new useful descriptors, there is an opportunity to improve the diagnostic performance of Ménière's disease imaging. KEY POINTS: ⢠A comprehensive range of existing and novel vestibular aqueduct delayed post-gadolinium MRI descriptors were compared for their diagnostic performance in Ménière's disease. ⢠Saccular abnormality (absent, confluent with or larger than the utricle) is a reliable descriptor and is the optimal individual MRI predictor of Ménière's disease. ⢠The presence of this saccule descriptor or asymmetric perilymphatic enhancement and incomplete vestibular aqueduct visualisation will optimise the MRI diagnosis of Ménière's disease.
ABSTRACT
Scalar translocation is a severe form of intra-cochlear trauma during cochlear implant (CI) electrode insertion. This study explored the hypothesis that the dimensions of the cochlear basal turn and orientation of its inferior segment relative to surgically relevant anatomical structures influence the scalar translocation rates of a pre-curved CI electrode. In a cohort of 40 patients implanted with the Advanced Bionics Mid-Scala electrode array, the scalar translocation group (40%) had a significantly smaller mean distance A of the cochlear basal turn (p < 0.001) and wider horizontal angle between the inferior segment of the cochlear basal turn and the mastoid facial nerve (p = 0.040). A logistic regression model incorporating distance A (p = 0.003) and horizontal facial nerve angle (p = 0.017) explained 44.0-59.9% of the variance in scalar translocation and correctly classified 82.5% of cases. Every 1mm decrease in distance A was associated with a 99.2% increase in odds of translocation [95% confidence interval 80.3%, 100%], whilst every 1-degree increase in the horizontal facial nerve angle was associated with an 18.1% increase in odds of translocation [95% CI 3.0%, 35.5%]. The study findings provide an evidence-based argument for the development of a navigation system for optimal angulation of electrode insertion during CI surgery to reduce intra-cochlear trauma.
Subject(s)
Cochlear Implantation , Cochlear Implants , Craniocerebral Trauma , Humans , Cochlea/surgery , Electrodes, Implanted , Bionics , Translocation, GeneticABSTRACT
OBJECTIVES: To evaluate ancillary MRI features which may aid the identification of lateral temporal bone cephaloceles (LTBCs). METHODS: A retrospective cohort study analysed patients with MRI evidence of surgically confirmed spontaneous LTBCs as defined by intracranial contents traversing the tegmen tympani or mastoideum. Cases were identified from radiology and surgical databases. Two observers analysed three-dimensional T 2W temporal bone and whole brain imaging according to a priori criteria by consensus, with emphasis on the relationship of any adjacent cerebrospinal fluid (CSF) cleft to the defect. The contents, location, and clinical features of the LTBCs were recorded. RESULTS: Eighteen patients (11 female, 7 male; mean age 59.3 years, age range 42-86 years) with 20 surgically confirmed spontaneous LTBCs (2 bilateral;16 unilateral) were evaluated. A temporal lobe sulcus or other CSF cleft extending to or traversing the defect was identified in 19/20 (95%) cases. Isointense CSF tympanomastoid signal was present in 41.2% cases, whilst superior semi-circular canal dehiscence was found in 40% of cephaloceles. At least two MRI features of idiopathic intracranial hypertension were seen in 38.9% patients. Cephaloceles were most commonly centred on the tegmen tympani (55%). Meningoencephaloceles were present in 95% cases. CONCLUSION: A temporal lobe sulcus or CSF cleft extending to or traversing the defect may aid the identification of LTBCs. Isointense CSF tympanomastoid signal, superior semi-circular canal dehiscence and MRI features of idiopathic intracranial hypertension are only present in under half of LTBCs. ADVANCES IN KNOWLEDGE: The study details novel ancillary MRI features of LTBCs which may aid their identification.
Subject(s)
Encephalocele , Pseudotumor Cerebri , Humans , Male , Female , Adult , Middle Aged , Aged , Aged, 80 and over , Encephalocele/diagnostic imaging , Encephalocele/surgery , Retrospective Studies , Temporal Bone/diagnostic imaging , Temporal Bone/surgery , Magnetic Resonance ImagingABSTRACT
Cutaneous lesions are derived from the epidermis, dermis and cutaneous appendages. Whilst imaging may occasionally be performed to evaluate such lesions, they may be undiagnosed and demonstrated for the first time on head and neck imaging studies. Although usually amenable to clinical examination and biopsy, CT or MRI studies may also demonstrate characteristic imaging features which aid the radiological differential diagnosis. In addition, imaging studies define the extent and staging of malignant lesions, as well as the complications of benign lesions. It is important for the radiologist to understanding the clinical significance and associations of these cutaneous conditions. This pictorial review will describe and depict the imaging appearances of benign, malignant, overgrowth, blistering, appendage and syndromic cutaneous lesions. An increasing awareness of the imaging characteristics of cutaneous lesions and related conditions will help the framing of a clinically relevant report.
ABSTRACT
OBJECTIVES: Delayed post-gadolinium magnetic resonance imaging (MRI) detects changes of endolymphatic hydrops (EH) within the inner ear in Meniere's disease (MD). A systematic review with meta-analysis was conducted to summarise the diagnostic performance of MRI descriptors across the range of MD clinical classifications. MATERIALS AND METHODS: Case-controlled studies documenting the diagnostic performance of MRI descriptors in distinguishing MD ears from asymptomatic ears or ears with other audio-vestibular conditions were identified (MEDLINE, EMBASE, Web of Science, Scopus databases: updated 17/2/2022). Methodological quality was evaluated with Quality Assessment of Diagnostic Accuracy Studies version 2. Results were pooled using a bivariate random-effects model for evaluation of sensitivity, specificity and diagnostic odds ratio (DOR). Meta-regression evaluated sources of heterogeneity, and subgroup analysis for individual clinical classifications was performed. RESULTS: The meta-analysis included 66 unique studies and 3073 ears with MD (mean age 40.2-67.2 years), evaluating 11 MRI descriptors. The combination of increased perilymphatic enhancement (PLE) and EH (3 studies, 122 MD ears) achieved the highest sensitivity (87% (95% CI: 79.92%)) whilst maintaining high specificity (91% (95% CI: 85.95%)). The diagnostic performance of "high grade cochlear EH" and "any EH" descriptors did not significantly differ between monosymptomatic cochlear MD and the latest reference standard for definite MD (p = 0.3; p = 0.09). Potential sources of bias were case-controlled design, unblinded observers and variable reference standard, whilst differing MRI techniques introduced heterogeneity. CONCLUSIONS: The combination of increased PLE and EH optimised sensitivity and specificity for MD, whilst some MRI descriptors also performed well in diagnosing monosymptomatic cochlear MD. KEY POINTS: ⢠A meta-analysis of delayed post-gadolinium magnetic resonance imaging (MRI) for the diagnosis of Meniere's disease is reported for the first time and comprised 66 studies (3073 ears). ⢠Increased enhancement of the perilymphatic space of the inner ear is shown to be a key MRI feature for the diagnosis of Meniere's disease. ⢠MRI diagnosis of Meniere's disease can be usefully applied across a range of clinical classifications including patients with cochlear symptoms alone.
Subject(s)
Ear, Inner , Endolymphatic Hydrops , Meniere Disease , Humans , Adult , Middle Aged , Aged , Meniere Disease/diagnostic imaging , Gadolinium/pharmacology , Endolymphatic Hydrops/diagnosis , Ear, Inner/diagnostic imaging , Magnetic Resonance Imaging/methodsSubject(s)
Lichen Planus , Humans , Lichen Planus/diagnosis , Lichen Planus/surgery , United KingdomABSTRACT
Introduction: Craniopharyngiomas are benign tumours mainly confined to the cranial cavity in the suprasellar region. Research Question and Case Description: We present a rare case of an aggressive papillary craniopharyngioma with disseminated spinal intradural disease. A 67-year-old woman presented with a 4-month history of headache, visual disturbance, acute confusion and radicular leg pain. Previous history of breast carcinoma (ER â+ âPR â+ âHER2-) was noted. The importance of histological diagnosis prior to treatment of sellar or suprasellar lesions with atypical or aggressive features is explored. Materials and methods: MRI demonstrated a partly solid and partly cystic pituitary mass lesion in the sellar and suprasellar region with chiasmal compression and hypothalamic involvement. The sella was mildly enlarged and there were no calcifications. Whole neuraxis MRI revealed intradural deposits involving the ventricular system, spinal cord and conus. Within a month, the lesion rapidly increased in size. The patient underwent a craniotomy and transventricular resection of the sellar and suprasellar mass. Cranial lesion histology favoured papillary craniopharyngioma, confirmed by BRAF V600 mutation. Lumbar puncture CSF cytology confirmed craniopharyngioma with BRAF mutation and no evidence of metastatic breast cancer. Results: The patient remained confused postoperatively without focal neurological deficit and underwent palliative whole brain radiotherapy. She died 4 months later. A review of the literature identified 29 reports of ruptured craniopharyngioma. Discussion and Conclusion: Ruptured craniopharyngioma presents with a suprasellar mass and drop lesions in the spinal canal, characteristics radiologically indistinguishable from metastatic disease. The importance of histological diagnoses in directing the management of these cases is highlighted.
ABSTRACT
We describe the first ever-reported occurrence of a post-operative tension pneumosyrinx occurring after a resection of an intradural intramedullary spinal tumour in a 40-year-old patient. Post-operatively, he developed sudden onset paraplegia and imaging revealed a tension pneumosyrinx which was subsequently surgically decompressed. He made a gradual neurological recovery. This is an extremely rare complication with potentially long-lasting deleterious effects on patients' neurological status if not recognized. We aim to bring this pathology to the attention of our neurosurgical colleagues and share our surgical approach and management to assist those who may encounter this pathology in the future.
Subject(s)
Neurosurgical Procedures , Spinal Cord Neoplasms , Adult , Humans , Iatrogenic Disease , Male , Neurosurgical Procedures/adverse effects , Paraplegia/etiology , Spinal Cord Neoplasms/surgeryABSTRACT
The clinical behaviour and outcomes of patients with oropharyngeal cancer (OPC) may be dichotomised according to their association with human papilloma virus (HPV) infection. Patients with HPV-associated disease (HPV+OPC) have a distinct demographic profile, clinical phenotype and demonstrate considerably better responses to chemoradiotherapy. This has led to a reappraisal of staging and treatment strategies for HPV+OPC, which are underpinned by radiological data. Structural modalities, such as CT and MRI can provide accurate staging information. These can be combined with ultrasound-guided tissue sampling and functional techniques (such as diffusion-weighted MRI and 18F-fludeoxyglucose positron emission tomography-CT) to monitor response to treatment, derive prognostic information, and to identify individuals who might benefit from intensification or deintensification strategies. Furthermore, advanced MRI techniques, such as intravoxel incoherent motion and perfusion MRI as well as application of artificial intelligence and radiomic techniques, have shown promise in treatment response monitoring and prognostication. The following review will consider the contemporary role and knowledge on imaging in HPV+OPC.
Subject(s)
Alphapapillomavirus , Head and Neck Neoplasms , Oropharyngeal Neoplasms , Papillomavirus Infections , Artificial Intelligence , Humans , Oropharyngeal Neoplasms/diagnostic imaging , Oropharyngeal Neoplasms/pathology , Oropharyngeal Neoplasms/therapy , Papillomaviridae , Papillomavirus Infections/complications , Papillomavirus Infections/diagnostic imaging , Prognosis , Squamous Cell Carcinoma of Head and Neck , Tomography, X-Ray ComputedABSTRACT
OBJECTIVES: The appearance of low-frequency air-bone gaps (LFABGs) in Meniere's disease (MD) is a recognized but relatively unexplored phenomenon. Two theories have been proposed to explain their etiology: increased perilymphatic pressure resulting in either reduced stapedial mobility or dampened transmission of acoustic energy, and direct contact between the dilated saccule and the stapes footplate. The aim of this study was to evaluate these two hypotheses by comparing delayed postgadolinium magnetic resonance imaging (MRI) features of two groups of patients with unilateral definite MD, those with and without LFABGs. DESIGN: This retrospective case-control study was conducted at a tertiary otolaryngology unit in the United Kingdom. The study included 35 patients who satisfied the 2015 Barany criteria for unilateral definite MD. The cohort was divided into two groups, those with LFABGs (LFABG+ group) and those without (LFABG- group), according to the pure-tone audiometry performed within 6 months of MRI. Alternative potential causes for the LFABGs were excluded on the basis of otologic history, otoscopy, tympanometry, and/or imaging. Using a 4-hr delayed postgadolinium 3-dimensional fluid-attenuated inversion recovery sequence, two observers evaluated the severity of cochlear and vestibular endolymphatic hydrops (EH) and the presence of vestibular endolymphatic space contacting the oval window (VESCO). The air and bone conduction thresholds, ABGs and MRI features were compared between the LFABG+ and LFABG- groups. Where any of the variables were found to be significantly associated with the presence of ABGs, further analysis was performed to determine whether or not they were independent predictors. Continuous variables were compared using the independent t test if normally distributed, and the Mann-Whitney U test or Kruskall-Wallis test if not normally distributed. Categorical variables were compared with Pearson's Chi-squared test or Fishers/Fisher-Freeman-Halton exact tests. RESULTS: There were 10 patients in the LFABG+ group (28.6%) and 25 patients in the LFABG- group (71.4%). The mean ABGs in the symptomatic ear at 500 Hz, 1 kHz, and 2 kHz were 15.1 dB ± 6.4, 10.5 dB ± 9.0, and 4.0 dB ± 7.7, respectively, in the LFABG+ group and 2.0 ± 5.8, 2.4 ± 4.4, and -0.8 ± 4.7 dB in the LFABG- group. The differences in ABGs between the two groups were statistically significant at all three test frequencies ( p < 0.001 at 500 Hz, p = 0.007 at 1 kHz, and p = 0.041 at 2 kHz). The presence of ABGs was significantly associated with both the grade of vestibular EH ( p = 0.049) and VESCO ( p = 0.009). Further analysis showed a statistically significant association between the grade of vestibular EH and VESCO ( p = 0.007), and only VESCO was an independent variable associated with the presence of LFABGs ( p = 0.045). CONCLUSIONS: The study findings add to the existing body of evidence that LFABGs are a true audiological finding in MD and allow us to propose a mechanism. Analysis of delayed gadolinium-enhanced MRI suggests that direct contact between the distended saccule and the inner surface of the stapes footplate is the more likely underlying pathophysiological mechanism for this audiometric phenomenon.
Subject(s)
Endolymphatic Hydrops , Meniere Disease , Vestibule, Labyrinth , Humans , Gadolinium , Retrospective Studies , Case-Control Studies , Endolymphatic Hydrops/diagnostic imaging , Magnetic Resonance Imaging/methods , Audiometry, Pure-ToneABSTRACT
BACKGROUND: The vidian canal (VC) is normally a reliable anatomical landmark for locating the petrous internal carotid artery (pICA). This study determined the influence of petroclival chondrosarcoma on the relationship between the VC and pICA. METHODS: Nine patients (3 males, 6 females; median age 49) with petroclival chondrosarcoma, and depiction of the pICA on contrast-enhanced CT, were retrospectively studied. CT-based measurements were performed by two observers, both in the presence of the petroclival chondrosarcoma (case) and on the contralateral control side. The antero-posterior (AP) and craniocaudal (CC) measurements from the posterior VC to the pICA, whether the pICA was in the trajectory of the VC, and the coronal relationship of the pICA anterior genu with the VC were recorded. RESULTS: Chondrosarcoma usually displaced the pICA anteriorly (8/9 cases) and superiorly (6/9 cases) relative to the normal side with mean AP and CC measurements of 3.9 mm v 7.2 mm (p = 0.054) and 4.4 mm v 1.4 mm (p = 0.061). The VC trajectory less frequently intersected the pICA cross-section in the presence of chondrosarcoma however it was in the line of the eroded dorsal VC in one case. The anterior genu of the pICA was displaced more laterally by chondrosarcoma but usually remained superior to the VC. CONCLUSION: Petroclival chondrosarcoma variably influences the anatomical relationship between the VC and the pICA, hence requiring an individualised approach. The pICA is usually anterosuperiorly displaced, and the anterior genu remains superior to the VC, however it may be located in the line of the canal.
Subject(s)
Carotid Artery, Internal , Chondrosarcoma , Carotid Artery, Internal/diagnostic imaging , Case-Control Studies , Chondrosarcoma/diagnostic imaging , Chondrosarcoma/surgery , Female , Humans , Male , Middle Aged , Petrous Bone/diagnostic imaging , Pica , Retrospective Studies , Tomography, X-Ray ComputedABSTRACT
The use of standardized imaging protocols is paramount in order to facilitate comparable, reproducible images and, consequently, to optimize patient care. Standardized MR protocols are lacking when studying head and neck pathologies in the pediatric population. We propose an international, multicenter consensus paper focused on providing the best combination of acquisition time/technical requirements and image quality. Distinct protocols for different regions of the head and neck and, in some cases, for specific pathologies or clinical indications are recommended. This white paper is endorsed by several international scientific societies and it is the result of discussion, in consensus, among experts in pediatric head and neck imaging.
Subject(s)
Head and Neck Neoplasms , Head , Child , Consensus , Head/diagnostic imaging , Humans , Magnetic Resonance Imaging/methods , Neck/diagnostic imagingABSTRACT
OBJECTIVES: To evaluate interval changes in heterogeneity on diffusion-weighted apparent diffusion coefficient (ADC) maps and T1-weighted post-gadolinium (T1w post gad) MRI in head and neck carcinoma (HNSCC), with and without chemo-radiotherapy (CRT) response. METHODS: This prospective observational cohort study included 24 participants (20 men, age 62.9 ± 8.8 years) with stage III and IV HNSCC. The primary tumour (n = 23) and largest lymph node (n = 22) dimensions, histogram parameters and grey-level co-occurrence matrix (GLCM) parameters were measured on ADC maps and T1w post gad sequences, performed pretreatment and 6 and 12 weeks post CRT. The 2-year treatment response at primary and nodal sites was recorded. The Wilcoxon signed-rank test was used to compare interval changes in parameters after stratifying for treatment response and failure (p < 0.001 statistical significance). RESULTS: 23/23 primary tumours and 18/22 nodes responded to CRT at 2 years. Responding HNSCC demonstrated a significant interval change in ADC histogram parameters (kurtosis, coefficient of variation, entropy, energy for primary tumour; kurtosis for nodes) and T1w post gad GLCM (entropy and contrast in the primary tumour and nodes) by 6 weeks post CRT (p < 0.001). Lymph nodes with treatment failure did not demonstrate an interval alteration in heterogeneity parameters. CONCLUSIONS: ADC maps and T1w post gad MRI demonstrate the evolution of heterogeneity parameters in successfully treated HNSCC by 6 weeks post CRT; however, this is not observed in lymph nodes failing treatment. ADVANCES IN KNOWLEDGE: Early reduction in heterogeneity is demonstrated on MRI when HNSCC responds to CRT.
ABSTRACT
Objective: The Koos grading scale is a frequently used classification system for vestibular schwannoma (VS) that accounts for extrameatal tumor dimension and compression of the brain stem. We propose an artificial intelligence (AI) pipeline to fully automate the segmentation and Koos classification of VS from MRI to improve clinical workflow and facilitate patient management. Methods: We propose a method for Koos classification that does not only rely on available images but also on automatically generated segmentations. Artificial neural networks were trained and tested based on manual tumor segmentations and ground truth Koos grades of contrast-enhanced T1-weighted (ceT1) and high-resolution T2-weighted (hrT2) MR images from subjects with a single sporadic VS, acquired on a single scanner and with a standardized protocol. The first stage of the pipeline comprises a convolutional neural network (CNN) which can segment the VS and 7 adjacent structures. For the second stage, we propose two complementary approaches that are combined in an ensemble. The first approach applies a second CNN to the segmentation output to predict the Koos grade, the other approach extracts handcrafted features which are passed to a Random Forest classifier. The pipeline results were compared to those achieved by two neurosurgeons. Results: Eligible patients (n = 308) were pseudo-randomly split into 5 groups to evaluate the model performance with 5-fold cross-validation. The weighted macro-averaged mean absolute error (MA-MAE), weighted macro-averaged F1 score (F1), and accuracy score of the ensemble model were assessed on the testing sets as follows: MA-MAE = 0.11 ± 0.05, F1 = 89.3 ± 3.0%, accuracy = 89.3 ± 2.9%, which was comparable to the average performance of two neurosurgeons: MA-MAE = 0.11 ± 0.08, F1 = 89.1 ± 5.2, accuracy = 88.6 ± 5.8%. Inter-rater reliability was assessed by calculating Fleiss' generalized kappa (k = 0.68) based on all 308 cases, and intra-rater reliabilities of annotator 1 (k = 0.95) and annotator 2 (k = 0.82) were calculated according to the weighted kappa metric with quadratic (Fleiss-Cohen) weights based on 15 randomly selected cases. Conclusions: We developed the first AI framework to automatically classify VS according to the Koos scale. The excellent results show that the accuracy of the framework is comparable to that of neurosurgeons and may therefore facilitate management of patients with VS. The models, code, and ground truth Koos grades for a subset of publicly available images (n = 188) will be released upon publication.
ABSTRACT
OBJECTIVE: Management of recurrent laryngeal cancer presents a major challenge, and salvage laryngectomy is complicated by previous oncologic treatments. Thyroidectomy as part of salvage laryngectomy adds a nonnegligible degree of morbidity. The purpose of this study is to assess the rate of thyroid gland invasion in patients undergoing salvage laryngectomy to determine relevant predictive factors. STUDY DESIGN: Case series with chart review. SETTING: Department of Otorhinolaryngology, Head and Neck Surgery, Guy's Hospital, London, United Kingdom. METHODS: A retrospective review of patients undergoing salvage laryngectomy between 2009 and 2019 was undertaken. Preoperative cross-sectional imaging and histopathological analysis were performed to define evidence and predictors of thyroid gland invasion (TGI). RESULTS: Fifty-one patients had salvage laryngectomy. Histological evidence of TGI was found in 4 patients (7.8%). No significant relationship was found between histological TGI and subsite of primary carcinoma, degree of differentiation, T staging, or radiological TGI. Preoperative computed tomography had a high negative predictive value for TGI. CONCLUSION: Thyroidectomy should be carefully considered in patients undergoing salvage laryngectomy, and its extent should be defined on an individual basis. Total thyroidectomy should not routinely be performed in salvage laryngectomy or pharyngolaryngectomy in patients with no preoperative radiological evidence of TGI on cross-sectional imaging, unless there is intraoperative evidence of TGI.