Automatic detection of head and neck squamous cell carcinoma on pathologic slides using polarized hyperspectral imaging and deep learning.

The study is to incorporate polarized hyperspectral imaging (PHSI) with deep learning for automatic detection of head and neck squamous cell carcinoma (SCC) on hematoxylin and eosin (H&E) stained tissue slides. A polarized hyperspectral imaging microscope had been developed in our group. In this paper, we firstly collected the Stokes vector data cubes (S0, S1, S2, and S3) of histologic slides from 17 patients with SCC by the PHSI microscope, under the wavelength range from 467 nm to 750 nm. Secondly, we generated the synthetic RGB images from the original Stokes vector data cubes. Thirdly, we cropped the synthetic RGB images into image patches at the image size of 96×96 pixels, and then set up a ResNet50-based convolutional neural network (CNN) to classify the image patches of the four Stokes vector parameters (S0, S1, S2, and S3) by application of transfer learning. To test the performances of the model, each time we trained the model based on the image patches (S0, S1, S2, and S3) of 16 patients out of 17 patients, and used the trained model to calculate the testing accuracy based on the image patches of the rest 1 patient (S0, S1, S2, and S3). We repeated the process for 6 times and obtained 24 testing accuracies (S0, S1, S2, and S3) from 6 different patients out of the 17 patients. The preliminary results showed that the average testing accuracy (84.2%) on S3 outperformed the average testing accuracy (83.5%) on S0. Furthermore, 4 of 6 testing accuracies of S3 (96.0%, 87.3%, 82.8%, and 86.7%) outperformed the testing accuracies of S0 (93.3%, 85.2%, 80.2%, and 79.0%). The study demonstrated the potential of using polarized hyperspectral imaging and deep learning for automatic detection of head and neck SCC on pathologic slides.

Oncologic outcomes in primary squamous cell carcinoma of the auricle: a retrospective cohort analysis.

PURPOSE: The prognostic significance of auricular location in cutaneous squamous cell carcinoma (cSCC) is controversial. We aimed to characterize risk factors for, and evaluate the cumulative incidence of, locoregional recurrence in a cohort of patients with primary auricular cSCC. METHODS: The study design was a single-institution retrospective cohort review from 1/2007 to 12/2016. RESULTS: Among 851 potentially eligible individuals, 178 patients with primary auricular cSCC met strict criteria for inclusion. Median follow-up was 32 months, 93% were AJCC 8th edition (AJCC8) stage I and 6% were AJCC8 stage II. Most underwent Mohs micrographic surgery (MMS; 97%) and the remainder underwent wide local excision ± parotidectomy ± neck dissection ± adjuvant therapy (3%). Recurrences occurred in seven patients (4%): six were local and one was regional. The 3-year cumulative incidence of local and regional recurrence for AJCC8 stage I-II tumors were 1% (95% CI 0-5%) and 0%, respectively. Among ten patients upstaged to pT3 disease who underwent MMS alone, none recurred locoregionally. Compared to their respective counterparts, advanced stage, PNI, and LVI associated with a significantly increased risk of locoregional recurrence. CONCLUSIONS: Our findings suggest that auricular location may not be a significant risk factor for cSCC staging systems. In the absence of other risk factors, unimodal therapy appears adequate for patients with primary, stage I-II auricular cSCC. The prognostic significance of pT3 auricular cSCC stage due to depth of invasion alone should be evaluated further.

Automatic detection of head and neck squamous cell carcinoma on pathologic slides using polarized hyperspectral imaging and machine learning.

The aim of this study is to incorporate polarized hyperspectral imaging (PHSI) with machine learning for automatic detection of head and neck squamous cell carcinoma (SCC) on hematoxylin and eosin (H&E) stained tissue slides. A polarized hyperspectral imaging microscope had been developed in our group. In this paper, we imaged 20 H&E stained tissue slides from 10 patients with SCC of the larynx by the PHSI microscope. Several machine learning algorithms, including support vector machine (SVM), random forest, Gaussian naive Bayes, and logistic regression, were applied to the collected image data for the automatic detection of SCC on the H&E stained tissue slides. The performance of these methods was compared among the collected PHSI data, the pseudo-RGB images generated from the PHSI data, and the PHSI data after applying the principal component analysis (PCA) transformation. The results suggest that SVM is a superior classifier for the classification task based on the PHSI data cubes compared to the other three classifiers. The incorporate of four Stokes vector parameters improved the classification accuracy. Finally, the PCA transformed image data did not improve the accuracy as it might lose some important information from the original PHSI data. The preliminary results show that polarized hyperspectral imaging can have many potential applications in digital pathology.

High-Risk Cutaneous Squamous Cell Carcinoma of the Head and Neck: A Clinical Review.

BACKGROUND: Given the rapidly evolving nature of the field, the current state of "high-risk" head and neck cutaneous squamous cell carcinoma (HNcSCC) is poorly characterized. METHODS: Narrative review of the epidemiology, diagnosis, workup, risk stratification, staging and treatment of high-risk HNcSCC. RESULTS: Clinical and pathologic risk factors for adverse HNcSCC outcomes are nuanced (e.g., immunosuppression and perineural invasion). Frequent changes in adverse prognosticators have outpaced population-based registries and the variables they track, restricting our understanding of the epidemiology of HNcSCC and inhibiting control of the disease. Current heterogeneous staging and risk stratification systems are largely derived from institutional data, compromising their external validity. In the absence of staging system consensus, tumor designations such as "high risk" and "advanced" are variably used and insufficiently precise to guide management. Evidence guiding treatment of high-risk HNcSCC with curative intent is also suboptimal. For patients with incurable disease, an array of trials are evaluating the impact of immunotherapy, targeted biologic therapy, and other novel agents. CONCLUSION: Population-based registries that broadly track updated, nuanced, adverse clinicopathologic risk factors, and outcomes are needed to guide development of improved staging systems. Design and development of randomized controlled trials (RCTs) in advanced-stage HNcSCC populations are needed to evaluate (1) observation, sentinel lymph node biopsy, or elective neck dissection for management of the cN0 neck, (2) indications for surgery plus adjuvant radiation versus adjuvant chemoradiation, and (3) the role of immunotherapy in treatment with curative intent. Considering these knowledge gaps, the authors explore a potential high-risk HNcSCC treatment framework.

Pixel-level Tumor Margin Assessment of Surgical Specimen with Hyperspectral Imaging and Deep Learning Classification.

Surgery is a major treatment method for squamous cell carcinoma (SCC). During surgery, insufficient tumor margin may lead to local recurrence of cancer. Hyperspectral imaging (HSI) is a promising optical imaging technique for in vivo cancer detection and tumor margin assessment. In this study, a fully convolutional network (FCN) was implemented for tumor classification and margin assessment on hyperspectral images of SCC. The FCN was trained and validated with hyperspectral images of 25 ex vivo SCC surgical specimens from 20 different patients. The network was evaluated per patient and achieved pixel-level tissue classification with an average area under the curve (AUC) of 0.88, as well as 0.83 accuracy, 0.84 sensitivity, and 0.70 specificity across all the 20 patients. The 95% Hausdorff distance of assessed tumor margin in 17 patients was less than 2 mm, and the classification time of each tissue specimen took less than 10 seconds. The proposed methods can potentially facilitate intraoperative tumor margin assessment and improve surgical outcomes.

Hyperspectral Microscopic Imaging for the Detection of Head and Neck Squamous Cell Carcinoma on Histologic Slides.

The purpose of this study is to investigate hyperspectral microscopic imaging and deep learning methods for automatic detection of head and neck squamous cell carcinoma (SCC) on histologic slides. Hyperspectral imaging (HSI) cubes were acquired from pathologic slides of 18 patients with SCC of the larynx, hypopharynx, and buccal mucosa. An Inception-based two-dimensional convolutional neural network (CNN) was trained and validated for the HSI data. The automatic deep learning method was tested with independent data of human patients. This study demonstrated the feasibility of using hyperspectral microscopic imaging and deep learning classification to aid pathologists in detecting SCC on histologic slides.

Using a 22-Layer U-Net to Perform Segmentation of Squamous Cell Carcinoma on Digitized Head and Neck Histological Images.

Squamous cell carcinoma (SCC) comprises over 90 percent of tumors in the head and neck. The diagnosis process involves performing surgical resection of tissue and creating histological slides from the removed tissue. Pathologists detect SCC in histology slides, and may fail to correctly identify tumor regions within the slides. In this study, a dataset of patches extracted from 200 digitized histological images from 84 head and neck SCC patients was used to train, validate and test the segmentation performance of a fully-convolutional U-Net architecture. The neural network achieved a pixel-level segmentation AUC of 0.89 on the testing group. The average segmentation time for whole slide images was 72 seconds. The training, validation, and testing process in this experiment produces a model that has the potential to help segment SCC images in histological images with improved speed and accuracy compared to the manual segmentation process performed by pathologists.

Head and neck oncology during the COVID-19 pandemic: Reconsidering traditional treatment paradigms in light of new surgical and other multilevel risks.

The COVID-19 pandemic demands reassessment of head and neck oncology treatment paradigms. Head and neck cancer (HNC) patients are generally at high-risk for COVID-19 infection and severe adverse outcomes. Further, there are new, multilevel COVID-19-specific risks to patients, surgeons, health care workers (HCWs), institutions and society. Urgent guidance in the delivery of safe, quality head and neck oncologic care is needed. Novel barriers to safe HNC surgery include: (1) imperfect presurgical screening for COVID-19; (2) prolonged SARS-CoV-2 aerosolization; (3) occurrence of multiple, potentially lengthy, aerosol generating procedures (AGPs) within a single surgery; (4) potential incompatibility of enhanced personal protective equipment (PPE) with routine operative equipment; (5) existential or anticipated PPE shortages. Additionally, novel, COVID-19-specific multilevel risks to HNC patients, HCWs and institutions, and society include: use of immunosuppressive therapy, nosocomial COVID-19 transmission, institutional COVID-19 outbreaks, and, at some locations, societal resource deficiencies requiring health care rationing. Traditional head and neck oncology doctrines require reassessment given the extraordinary COVID-19-specific risks of surgery. Emergent, comprehensive management of these novel, multilevel surgical risks are needed. Until these risks are managed, we temporarily favor nonsurgical therapy over surgery for most mucosal squamous cell carcinomas, wherein surgery and nonsurgical therapy are both first-line options. Where surgery is traditionally preferred, we recommend multidisciplinary evaluation of multilevel surgical-risks, discussion of possible alternative nonsurgical therapies and shared-decision-making with the patient. Where surgery remains indicated, we recommend judicious preoperative planning and development of COVID-19-specific perioperative protocols to maximize the safety and quality of surgical and oncologic care.

Unique Patterns of Distant Metastases in HPV-Positive Head and Neck Cancer.

BACKGROUND: HPV-positive head and neck squamous cell carcinoma (HPV+ HNSCC) demonstrates favorable outcomes compared to HPV-negative SCC, but distant metastases (DM) still occur. The pattern of DM in HPV+ HNSCC is unclear. METHODS: 1,494 HNSCC patients were treated from 2006 to 2012. Recurrence time and metastatic sites in HPV+ HNSCC (Group 1) were compared to patients with HPV-negative/unknown cancers arising in the hypopharynx, larynx, or glottis (Group 2) as well as to patients with HPV-negative/unknown cancers in theoral cavity, oropharynx, hard palate, or tonsil (Group 3). RESULTS: 7/109 (6.4%) patients with HPV+ HNSCC developed DM. The median time to metastases was 11 months. At a median follow-up of 18-25 months, there was no difference in the overall rate of DM for the HPV+ HNSCC group compared to Group 2 (HPV-/unknown) (p = 0.21) and Group 3 (HPV-/unknown) (p = 0.13). There was a significant difference in the rate of DM to the lung in the HPV+ HNSCC group compared to Group 2 (HPV-/unknown) (p = 0.012) and Group 3 (HPV-/unknown) (p = 0.002). CONCLUSIONS: There was no observed difference in the time to development of DM between the HPV-/unknown and HPV+ HNSCC groups. However, the HPV+ HNSCC group showed a higher rate of DM to the lung compared to the HPV-/unknown -HNSCC group (p = 0.002).

Head and Neck Cancer Detection in Digitized Whole-Slide Histology Using Convolutional Neural Networks.

Primary management for head and neck cancers, including squamous cell carcinoma (SCC), involves surgical resection with negative cancer margins. Pathologists guide surgeons during these operations by detecting cancer in histology slides made from the excised tissue. In this study, 381 digitized, histological whole-slide images (WSI) from 156 patients with head and neck cancer were used to train, validate, and test an inception-v4 convolutional neural network. The proposed method is able to detect and localize primary head and neck SCC on WSI with an AUC of 0.916 for patients in the SCC testing group and 0.954 for patients in the thyroid carcinoma testing group. Moreover, the proposed method is able to diagnose WSI with cancer versus normal slides with an AUC of 0.944 and 0.995 for the SCC and thyroid carcinoma testing groups, respectively. For comparison, we tested the proposed, diagnostic method on an open-source dataset of WSI from sentinel lymph nodes with breast cancer metastases, CAMELYON 2016, to obtain patch-based cancer localization and slide-level cancer diagnoses. The experimental design yields a robust method with potential to help create a tool to increase efficiency and accuracy of pathologists detecting head and neck cancers in histological images.

Spindle-cell variant of ameloblastic carcinoma: a report of 3 cases and demonstration of epithelial-mesenchymal transition in tumor progression.

Ameloblastic carcinoma is a rare odontogenic neoplasm that demonstrates the histologic characteristics of ameloblastoma, accompanied by the cytologic features of malignancy. The spindle-cell variant of ameloblastic carcinoma (SCAC) is exceptionally rare, with a total of 10 cases having been reported in the literature to date. Histologically, a prominent sarcomatoid cell population appears to originate from the epithelial (ameloblastic) component. Like conventional ameloblastic carcinoma, most cases of SCAC occur in individuals older than 40 years of age. Here, 3 additional cases of SCAC are reported, 2 of which occurred in young individuals. Diagnostic criteria to aid in the identification of SCAC are proposed. Finally, histologic and immunohistochemical evidence supporting the occurrence of epithelial-mesenchymal transition in SCAC is presented.

Cutaneous squamous cell carcinoma metastasis to the parotid region lymph nodes.

OBJECTIVE: To determine the effect on survival of periauricular region cutaneous squamous cell carcinomas (cSCC) metastasizing to parotid region and cervical lymph nodes. METHODS: From May 1995 to October 2013, consecutive patients with cSCC undergoing parotidectomy/neck dissection ≥ 18 years without distant metastasis were included. Demographic, clinical, and pathologic data were analyzed. Uni- and multivariate analyses of disease-specific survival and overall survival were performed. Statistical significance was set at P < 0.05. RESULTS: We evaluated 137 patients. Most patients were Caucasian males with an average age of 71.6 years. The median follow-up time was 29.4 months. Thirty-six percent of patients had pathological evidence of cervical metastasis (N+). Seven percent had metastasis to level I, 21.9% to level II, 14.6% to level III, 8% to level IV, and 9.5% to level V. Thirty-nine percent of patients had metastasis to parotid region (P+). The average number of positive:total nodes from parotidectomy was 1.9:4.6. The median overall and disease-specific survival times were 2.81 and 1.96 years, respectively. Patients with either neck or parotid metastasis (N + or P+) had significantly decreased survival, hazard ratio 2.298 (1.2739, 4.1445), compared to patients without metastasis (N0P0), P < 0.0057. CONCLUSION: Patients with periauricular region cSCC metastasizing to parotid and cervical regions lymph nodes have significantly decreased survival compared with patients without cervical and/or parotid metastasis. Metastasis to parotid region should be considered as a distinct head and neck level when staging periauricular cancers. This study provides a basis for further studies to validate these findings. LEVEL OF EVIDENCE: 4 Laryngoscope, 129:1579-1586, 2019.

Association between treatment delays and oncologic outcome in patients treated with surgery and radiotherapy for head and neck cancer.

BACKGROUND: This study sought to determine the oncologic impact of delays to surgery, radiotherapy, and completion of therapy in patients with head and neck squamous cell carcinoma. METHODS: The impact of biopsy to surgery (BTS) time, surgery to start of radiation time (STSR), and radiation treatment time (RTT) on locoregional recurrence (LRR), distant metastases (DMs), and cancer-specific mortality (CSM) was examined. The cumulative incidences (CI) of LRR, DMs, and CSM were examined using Fine-Gray testing. RESULTS: A total of 277 patients treated with surgery and adjuvant radiotherapy were analyzed. On multivariable testing, BTS >50 days was associated with DM (P = .03), whereas RTT and STSR were not. RTT >43 days was associated with LRR (P = .02) in patients with non-p16-positive-oropharynx cancer. CONCLUSIONS: An increase in DM appears to be the mechanism by which prolonged time to treatment initiation leads to worse overall survival. Prolonged RTT has the greatest impact on patients with non-p16 positive oropharynx cancers.

Detection of Squamous Cell Carcinoma in Digitized Histological Images from the Head and Neck Using Convolutional Neural Networks.

Primary management for head and neck squamous cell carcinoma (SCC) involves surgical resection with negative cancer margins. Pathologists guide surgeons during these operations by detecting SCC in histology slides made from the excised tissue. In this study, 192 digitized histological images from 84 head and neck SCC patients were used to train, validate, and test an inception-v4 convolutional neural network. The proposed method performs with an AUC of 0.91 and 0.92 for the validation and testing group. The careful experimental design yields a robust method with potential to help create a tool to increase efficiency and accuracy of pathologists for detecting SCC in histological images.

Cancer Detection Using Hyperspectral Imaging and Evaluation of the Superficial Tumor Margin Variance with Depth.

Head and neck squamous cell carcinoma (SCCa) is primarily managed by surgical resection. Recurrence rates after surgery can be as high as 55% if residual cancer is present. In this study, hyperspectral imaging (HSI) is evaluated for detection of SCCa in ex-vivo surgical specimens. Several methods are investigated, including convolutional neural networks (CNNs) and a spectral-spatial variant of support vector machines. Quantitative results demonstrate that additional processing and unsupervised filtering can improve CNN results to achieve optimal performance. Classifying regions that include specular glare, the average AUC is increased from 0.73 [0.71, 0.75 (95% confidence interval)] to 0.81 [0.80, 0.83] through an unsupervised filtering and majority voting method described. The wavelengths of light used in HSI can penetrate different depths into biological tissue, while the cancer margin may change with depth and create uncertainty in the ground-truth. Through serial histological sectioning, the variance in cancer-margin with depth is also investigated and paired with qualitative classification heat maps using the methods proposed for the testing group SCC patients.

MR Neurographic Evaluation of Facial and Neck Pain: Normal and Abnormal Craniospinal Nerves below the Skull Base.

Cranial nerve disease outside the skull base is a common cause of facial and/or neck pain, which causes significant disability for patients and frustration for clinicians. Neuropathy in this region can be traumatic, idiopathic, or iatrogenic secondary to dental and surgical procedures. MR neurography is a modification of conventional MRI techniques dedicated to evaluation of peripheral nerves and is being increasingly used for imaging of peripheral neuropathies at various sites in the body. MR neurography facilitates assessment of different causes of craniofacial pain and cranial nerves and allows elegant depiction of a multitude of regional neuropathies. This article discusses the anatomy, pathologic conditions, and imaging findings of the commonly implicated but difficult to image infratentorial nerves, such as the peripheral trigeminal nerve and its branches, facial nerve, glossopharyngeal nerve, vagus nerve, hypoglossal nerve, and greater and lesser occipital nerves. ©RSNA, 2018.

A Phase I/II Study of Nab-Paclitaxel, Cisplatin, and Cetuximab With Concurrent Radiation Therapy for Locally Advanced Squamous Cell Cancer of the Head and Neck.

Nab-paclitaxel might impact efficacy of radiation for head and neck (H&N) cancer. Nab-paclitaxel, cisplatin, cetuximab, and radiation were evaluated in patients with locally advanced head and neck cancer in this phase I/II trial. Median follow-up was 24 months for 34 patients. The maximum tolerated dose of nab-paclitaxel was 20 mg/m2 with 20 mg/m2 cisplatin and 250 mg/m2 cetuximab. The 2-year progression-free survival (PFS) was 60% (95% confidence interval (CI) 0.42, 0.78), local control 71% (95% CI 0.55, 0.87), and overall survival 68% (95% CI 0.50, 0.86). This is the first study evaluating these agents with radiation in humans, with similar 2-year PFS as historic control.

Patterns of Care and Comparative Effectiveness of Intensified Adjuvant Therapy for Resected Oropharyngeal Squamous Cell Carcinoma in the Human Papillomavirus Era.

IMPORTANCE: There is a growing debate on the relative benefits of adjuvant chemoradiotherapy (CRT) and boost doses of postoperative radiotherapy (B-PORT) in oropharyngeal squamous cell carcinoma (OPSCC) treated with primary surgery, especially for patients with human papillomavirus (HPV)-driven disease. OBJECTIVE: To characterize the recent patterns of care in and overall survival (OS) outcomes following the use of adjuvant CRT and B-PORT after primary surgery for OPSCC. DESIGN, SETTING, AND PARTICIPANTS: Retrospective analysis of patients in the National Cancer Database with stage III to IVA-B OPSCC treated with surgery and adjuvant radiotherapy between 2010 and 2012 at Commission on Cancer-accredited facilities. The data analysis was performed between June 15, 2015, and May 4, 2016. MAIN OUTCOMES AND MEASURES: The primary outcomes were prevalence of CRT and B-PORT, and OS. The primary predictors were HPV positivity and high-risk pathologic features (HRPFs) (extracapsular extension and positive surgical margins). RESULTS: Of the 1409 patients (1153 [82%] male; median age, 57 [interquartile range {IQR}, 51-63] years), 873 (62%) and 789 (56%) patients received CRT and B-PORT, respectively; most patients (n = 583 [79%]) with HRPFs received CRT, and many patients (n = 227 [40%]) without HRPFs received CRT. Multivariable predictors of CRT included adverse pathologic features (extracapsular extension [OR, 6.99; 95% CI, 5.22-9.35], positive surgical margins [OR, 2.07; 95% CI, 1.50-2.87], ≥6 involved nodes [OR, 2.34; 95% CI, 1.39-3.92], or low-neck disease [OR, 1.52; 95% CI, 1.01-2.28]), and treatment at a nonacademic institution (OR, 1.59 [95% CI, 1.21-2.10] for comprehensive community cancer center vs academic program). Patients with HPV-positive disease (OR, 0.47; 95% CI, 0.33-0.68) were less likely to receive CRT; this decrease was limited to absent HRPF treated at academic institutions (n = 173, 44 [25%] received CRT). With a median follow-up of surviving patients of 27 (IQR, 21-33) months, the 2-year OS probability was 92% (95% CI, 90%-94%). Multivariable analysis including age, sex, pathologic T stage, 6 or more positive nodes, and educational status confirmed the prognostic impact of HPV positivity (hazard ratio [HR], 0.41; 95% CI, 0.21-0.80) and HRPFs (positive surgical margins [HR, 2.15; 95% CI, 1.27-3.66] and ≥6 involved nodes [HR, 2.11; 95% CI, 1.13-3.93]), but neither CRT (HR, 1.27; 95% CI, 0.70-2.30) nor B-PORT (HR, 1.04; 95% CI, 0.63-1.73) was associated with improved OS. CONCLUSIONS AND RELEVANCE: Postoperative CRT and B-PORT following resection of OPSCC were dependent on factors beyond HRPFs, including HPV status and treatment at an academic institution. No benefit was seen with intensified adjuvant therapy, supporting enrollment of the HPV-positive population into deintensification trials.