Multimodal artificial intelligence-based pathogenomics improves survival prediction in oral squamous cell carcinoma.

In this study, we aimed to develop a novel prognostic algorithm for oral squamous cell carcinoma (OSCC) using a combination of pathogenomics and AI-based techniques. We collected comprehensive clinical, genomic, and pathology data from a cohort of OSCC patients in the TCGA dataset and used machine learning and deep learning algorithms to identify relevant features that are predictive of survival outcomes. Our analyses included 406 OSCC patients. Initial analyses involved gene expression analyses, principal component analyses, gene enrichment analyses, and feature importance analyses. These insights were foundational for subsequent model development. Furthermore, we applied five machine learning/deep learning algorithms (Random Survival Forest, Gradient Boosting Survival Analysis, Cox PH, Fast Survival SVM, and DeepSurv) for survival prediction. Our initial analyses revealed relevant gene expression variations and biological pathways, laying the groundwork for robust feature selection in model building. The results showed that the multimodal model outperformed the unimodal models across all methods, with c-index values of 0.722 for RSF, 0.633 for GBSA, 0.625 for FastSVM, 0.633 for CoxPH, and 0.515 for DeepSurv. When considering only important features, the multimodal model continued to outperform the unimodal models, with c-index values of 0.834 for RSF, 0.747 for GBSA, 0.718 for FastSVM, 0.742 for CoxPH, and 0.635 for DeepSurv. Our results demonstrate the potential of pathogenomics and AI-based techniques in improving the accuracy of prognostic prediction in OSCC, which may ultimately aid in the development of personalized treatment strategies for patients with this devastating disease.

Performance of artificial intelligence-based algorithms to predict prolonged length of stay after head and neck cancer surgery.

Background: Medical resource management can be improved by assessing the likelihood of prolonged length of stay (LOS) for head and neck cancer surgery patients. The objective of this study was to develop predictive models that could be used to determine whether a patient's LOS after cancer surgery falls within the normal range of the cohort. Methods: We conducted a retrospective analysis of a dataset consisting of 300 consecutive patients who underwent head and neck cancer surgery between 2017 and 2022 at a single university medical center. Prolonged LOS was defined as LOS exceeding the 75th percentile of the cohort. Feature importance analysis was performed to evaluate the most important predictors for prolonged LOS. We then constructed 7 machine learning and deep learning algorithms for the prediction modeling of prolonged LOS. Results: The algorithms reached accuracy values of 75.40 (radial basis function neural network) to 97.92 (Random Trees) for the training set and 64.90 (multilayer perceptron neural network) to 84.14 (Random Trees) for the testing set. The leading parameters predicting prolonged LOS were operation time, ischemia time, the graft used, the ASA score, the intensive care stay, and the pathological stages. The results revealed that patients who had a higher number of harvested lymph nodes (LN) had a lower probability of recurrence but also a greater LOS. However, patients with prolonged LOS were also at greater risk of recurrence, particularly when fewer (LN) were extracted. Further, LOS was more strongly correlated with the overall number of extracted lymph nodes than with the number of positive lymph nodes or the ratio of positive to overall extracted lymph nodes, indicating that particularly unnecessary lymph node extraction might be associated with prolonged LOS. Conclusions: The results emphasize the need for a closer follow-up of patients who experience prolonged LOS. Prospective trials are warranted to validate the present results.

Finetuning of GLIDE stable diffusion model for AI-based text-conditional image synthesis of dermoscopic images.

Background: The development of artificial intelligence (AI)-based algorithms and advances in medical domains rely on large datasets. A recent advancement in text-to-image generative AI is GLIDE (Guided Language to Image Diffusion for Generation and Editing). There are a number of representations available in the GLIDE model, but it has not been refined for medical applications. Methods: For text-conditional image synthesis with classifier-free guidance, we have fine-tuned GLIDE using 10,015 dermoscopic images of seven diagnostic entities, including melanoma and melanocytic nevi. Photorealistic synthetic samples of each diagnostic entity were created by the algorithm. Following this, an experienced dermatologist reviewed 140 images (20 of each entity), with 10 samples originating from artificial intelligence and 10 from original images from the dataset. The dermatologist classified the provided images according to the seven diagnostic entities. Additionally, the dermatologist was asked to indicate whether or not a particular image was created by AI. Further, we trained a deep learning model to compare the diagnostic results of dermatologist versus machine for entity classification. Results: The results indicate that the generated images possess varying degrees of quality and realism, with melanocytic nevi and melanoma having higher similarity to real images than other classes. The integration of synthetic images improved the classification performance of the model, resulting in higher accuracy and precision. The AI assessment showed superior classification performance compared to dermatologist. Conclusion: Overall, the results highlight the potential of synthetic images for training and improving AI models in dermatology to overcome data scarcity.

Realizing in-house algorithm-driven free fibula flap set up within 24 hours: a pilot study evaluating accuracy with open-source tools.

Objective: This study aims to critically evaluate the effectiveness and accuracy of a time safing and cost-efficient open-source algorithm for in-house planning of mandibular reconstructions using the free osteocutaneous fibula graft. The evaluation focuses on quantifying anatomical accuracy and assessing the impact on ischemia time. Methods: A pilot study was conducted, including patients who underwent in-house planned computer-aided design and manufacturing (CAD/CAM) of free fibula flaps between 2021 and 2023. Out of all patient cases, we included all with postoperative 3D imaging in the study. The study utilized open-source software tools for the planning step, and three-dimensional (3D) printing techniques. The Hausdorff distance and Dice coefficient metrics were used to evaluate the accuracy of the planning procedure. Results: The study assessed eight patients (five males and three females, mean age 61.75 ± 3.69 years) with different diagnoses such as osteoradionecrosis and oral squamous cell carcinoma. The average ischemia time was 68.38 ± 27.95 min. For the evaluation of preoperative planning vs. the postoperative outcome, the mean Hausdorff Distance was 1.22 ± 0.40. The Dice Coefficients yielded a mean of 0.77 ± 0.07, suggesting a satisfactory concordance between the planned and postoperative states. Dice Coefficient and Hausdorff Distance revealed significant correlations with ischemia time (Spearman's rho = -0.810, p = 0.015 and Spearman's rho = 0.762, p = 0.028, respectively). Linear regression models adjusting for disease type further substantiated these findings. Conclusions: The in-house planning algorithm not only achieved high anatomical accuracy, as reflected by the Dice Coefficients and Hausdorff Distance metrics, but this accuracy also exhibited a significant correlation with reduced ischemia time. This underlines the critical role of meticulous planning in surgical outcomes. Additionally, the algorithm's open-source nature renders it cost-efficient, easy to learn, and broadly applicable, offering promising avenues for enhancing both healthcare affordability and accessibility.

Associations between Periodontitis and COPD: An Artificial Intelligence-Based Analysis of NHANES III.

A number of cross-sectional epidemiological studies suggest that poor oral health is associated with respiratory diseases. However, the number of cases within the studies was limited, and the studies had different measurement conditions. By analyzing data from the National Health and Nutrition Examination Survey III (NHANES III), this study aimed to investigate possible associations between chronic obstructive pulmonary disease (COPD) and periodontitis in the general population. COPD was diagnosed in cases where FEV (1)/FVC ratio was below 70% (non-COPD versus COPD; binary classification task). We used unsupervised learning utilizing k-means clustering to identify clusters in the data. COPD classes were predicted with logistic regression, a random forest classifier, a stochastic gradient descent (SGD) classifier, k-nearest neighbors, a decision tree classifier, Gaussian naive Bayes (GaussianNB), support vector machines (SVM), a custom-made convolutional neural network (CNN), a multilayer perceptron artificial neural network (MLP), and a radial basis function neural network (RBNN) in Python. We calculated the accuracy of the prediction and the area under the curve (AUC). The most important predictors were determined using feature importance analysis. Results: Overall, 15,868 participants and 19 feature variables were included. Based on k-means clustering, the data were separated into two clusters that identified two risk characteristic groups of patients. The algorithms reached AUCs between 0.608 (DTC) and 0.953% (CNN) for the classification of COPD classes. Feature importance analysis of deep learning algorithms indicated that age and mean attachment loss were the most important features in predicting COPD. Conclusions: Data analysis of a large population showed that machine learning and deep learning algorithms could predict COPD cases based on demographics and oral health feature variables. This study indicates that periodontitis might be an important predictor of COPD. Further prospective studies examining the association between periodontitis and COPD are warranted to validate the present results.

Deep Learning on Basal Cell Carcinoma In Vivo Reflectance Confocal Microscopy Data.

BACKGROUND: Extended skin malignancies of the head and neck region are among the most common cancer types and are associated with numerous diagnostic and therapeutical problems. The radical resection of skin cancer in the facial area often leads to severe functional and aesthetic impairment, and precise margin assessments can avoid the extensive safety margins. On the other hand, the complete removal of the cancer is essential to minimize the risk of recurrence. Reliable intraoperative assessments of the wound margins could overcome this discrepancy between minimal invasiveness and safety distance in the head and neck region. With the help of reflectance confocal laser microscopy (RCM), cells can be visualized in high resolution intraoperatively. The combination with deep learning and automated algorithms allows an investigator independent and objective interpretation of specific confocal imaging data. Therefore, we aimed to apply a deep learning algorithm to detect malignant areas in images obtained via in vivo confocal microscopy. We investigated basal cell carcinoma (BCC), as one of the most common entities with well-described in vivo RCM diagnostic criteria, within a preliminary feasibility study. PATIENTS AND METHODS: We included 62 patients with histologically confirmed BCC in the head and neck region. All patients underwent in vivo confocal laser microscope scanning. Approximately 382 images with BCC structures could be obtained, annotated, and proceeded for further deep learning model training. RESULTS: A sensitivity of 46% and a specificity of 85% in detecting BCC regions could be achieved using a convolutional neural network model ("MobileNet"). CONCLUSION: The preliminary results reveal the potential and limitations of the automated detection of BCC with in vivo RCM. Further studies with a larger number of cases are required to obtain better predictability.

Artificial Intelligence-Based Prediction of Oroantral Communication after Tooth Extraction Utilizing Preoperative Panoramic Radiography.

Oroantral communication (OAC) is a common complication after tooth extraction of upper molars. Profound preoperative panoramic radiography analysis might potentially help predict OAC following tooth extraction. In this exploratory study, we evaluated n = 300 consecutive cases (100 OAC and 200 controls) and trained five machine learning algorithms (VGG16, InceptionV3, MobileNetV2, EfficientNet, and ResNet50) to predict OAC versus non-OAC (binary classification task) from the input images. Further, four oral and maxillofacial experts evaluated the respective panoramic radiography and determined performance metrics (accuracy, area under the curve (AUC), precision, recall, F1-score, and receiver operating characteristics curve) of all diagnostic approaches. Cohen's kappa was used to evaluate the agreement between expert evaluations. The deep learning algorithms reached high specificity (highest specificity 100% for InceptionV3) but low sensitivity (highest sensitivity 42.86% for MobileNetV2). The AUCs from VGG16, InceptionV3, MobileNetV2, EfficientNet, and ResNet50 were 0.53, 0.60, 0.67, 0.51, and 0.56, respectively. Expert 1-4 reached an AUC of 0.550, 0.629, 0.500, and 0.579, respectively. The specificity of the expert evaluations ranged from 51.74% to 95.02%, whereas sensitivity ranged from 14.14% to 59.60%. Cohen's kappa revealed a poor agreement for the oral and maxillofacial expert evaluations (Cohen's kappa: 0.1285). Overall, present data indicate that OAC cannot be sufficiently predicted from preoperative panoramic radiography. The false-negative rate, i.e., the rate of positive cases (OAC) missed by the deep learning algorithms, ranged from 57.14% to 95.24%. Surgeons should not solely rely on panoramic radiography when evaluating the probability of OAC occurrence. Clinical testing of OAC is warranted after each upper-molar tooth extraction.

Ex vivo fluorescent confocal microscopy images of oral mucosa: Tissue atlas and evaluation of the learning curve.

Ex vivo fluorescence confocal microscopy (FCM) is a developing tool providing rapid digital imaging of fresh tissue utilizing high-resolution optical sectioning that highly corresponds with conventional hmatoxylin and eosin (H&E)-stained slides. A very little data on oral mucosa lesions exist currently. The present work aimed to create an image atlas of benign and malignant oral tissues and compare them to the corresponding histopathology. Furthermore, we aimed to evaluate the learning curve for confocal image interpretation. From 50 samples obtained from the oral mucosa, including oral squamous cell carcinoma (OSCC), dysplasia, and healthy oral tissue, ex vivo FCM images and corresponding H&E slides were created and collected into a tissue atlas. Additionally, two experts were asked to analyze the images to assess the learning curve. Ex vivo FCM images revealed high comparability with histopathological images. Tissues including OSCC, dysplasia, and normal oral mucosa were implemented in the image atlas to provide the diagnostic fundament for pathologists and surgeons; the learning curve was short. Future studies on this topic will be advantageous for the development of artificial intelligence-based diagnostic approaches. The current work provides a novel set of data that are structured as an atlas of common pathologies of the mucosa to enhance the existing knowledge and material on confocal images.

Deep Learning on Oral Squamous Cell Carcinoma Ex Vivo Fluorescent Confocal Microscopy Data: A Feasibility Study.

BACKGROUND: Ex vivo fluorescent confocal microscopy (FCM) is a novel and effective method for a fast-automatized histological tissue examination. In contrast, conventional diagnostic methods are primarily based on the skills of the histopathologist. In this study, we investigated the potential of convolutional neural networks (CNNs) for automatized classification of oral squamous cell carcinoma via ex vivo FCM imaging for the first time. MATERIAL AND METHODS: Tissue samples from 20 patients were collected, scanned with an ex vivo confocal microscope immediately after resection, and investigated histopathologically. A CNN architecture (MobileNet) was trained and tested for accuracy. RESULTS: The model achieved a sensitivity of 0.47 and specificity of 0.96 in the automated classification of cancerous tissue in our study. CONCLUSION: In this preliminary work, we trained a CNN model on a limited number of ex vivo FCM images and obtained promising results in the automated classification of cancerous tissue. Further studies using large sample sizes are warranted to introduce this technology into clinics.

Mandibular Brown Tumor as a Result of Secondary Hyperparathyroidism: A Case Report with 5 Years Follow-Up and Review of the Literature.

BACKGROUND: Brown tumor is a rare skeletal manifestation of secondary hyperparathyroidism. Although diagnosis of the disease is increasingly seen in early stages due to improved screening techniques, some patients still present in a progressed disease stage. The treatment depends on tumor mass and varies from a conservative approach with supportive parathyroidectomy to extensive surgical resection with subsequent reconstruction. CASE PRESENTATION: We report a case of extensive mandibular brown tumor in a patient with a history of systemic lupus erythematosus, chronic kidney disease, and secondary hyperparathyroidism. Following radical resection of the affected bone, reconstruction could be successfully performed using a free flap. CONCLUSIONS: There were no signs of recurrence during five years of close follow-up. Increased awareness and multidisciplinary follow-ups could allow early diagnosis and prevent the need for radical therapeutical approaches.

Feasibility and Implementation of Ex Vivo Fluorescence Confocal Microscopy for Diagnosis of Oral Leukoplakia: Preliminary Study.

BACKGROUND: Oral leukoplakia is a potentially malignant lesion with a clinical impression similar to different benign and malignant lesions. Ex vivo fluorescence confocal microscopy is a developing approach for a rapid "chairside" detection of oral lesions with a cellular-level resolution. A possible application of interest is a quick differentiation of benign oral pathology from normal or cancerous tissue. The aim of this study was to analyze the sensitivity and specificity of ex vivo fluorescence confocal microscopy (FCM) for detecting oral leukoplakia and to compare confocal images with gold-standard histopathology. METHODS: Imaging of 106 submosaics of 27 oral lesions was performed using an ex vivo fluorescence confocal microscope immediately after excision. Every confocal image was qualitatively assessed for presence or absence of leukoplakia by an expert reader of confocal images. The results were compared to conventional histopathology with H&E staining. RESULTS: Leukoplakia was detected with an overall sensitivity of 96.3%, specificity of 92.3%, positive predictive value of 93%, and negative predictive value of 96%. CONCLUSION: The results demonstrate the potential of ex vivo confocal microscopy in fresh tissue for rapid real-time assessment of oral pathologies.

Features of oral squamous cell carcinoma in ex vivo fluorescence confocal microscopy.

BACKGROUND: Real-time microscopic imaging of freshly excised oral squamous cell carcinomas (OSCCs) would be potentially supportive in rapid recognition of oral malignancy and an optimal and time-saving management of patients' surgical treatment. OBJECTIVES: The aim of this study was to examine oral squamous cell cancer tissue in regards to the commonly known and well-described histomorphologic criteria for the diagnosis of OSCC in ex vivo confocal fluorescent microscopy and to analyze its correlation with grade of differentiation and level of invasiveness. METHODS: Ex vivo confocal laser scanning microscopy (CLSM) images of 38 OSCCs were evaluated immediately after excision for presence or absence of various cytological and architectural features based on the histopathological background. Next, these features were compared to the grade of differentiation as elaborated via gold standard histologic examination. RESULTS: Of 38 invasive OSCCs, 14 were well differentiated, while three moderately and 19 were poorly differentiated. The presence of the commonly known cytologic and histopathologic criteria for the diagnosis of oral squamous cell carcinoma such as the destruction of the basal cell membrane, cellular and nuclear pleomorphism, anisocytosis, intraepithelial keratinization, nuclear hyperchromasia, atypical mitotic figures as well as the presence of necrosis, and mixed inflammation could be observed in ex vivo fluorescence confocal microscopy (FCM). In ex vivo fluorescence confocal microscopy pictures, cellular pleomorphism and anisocytosis were observed more often in poorly differentiated OSCCs. Intraepithelial keratinization was associated with well differentiated and moderately differentiated OSCCs. CONCLUSION: The results demonstrate the high potential of ex vivo fluorescence confocal microscopy in fresh tissue for rapid real-time diagnosis of OSCC.

In vivo reflectance confocal microscopy of wounds: feasibility of intraoperative basal cell carcinoma margin assessment.

BACKGROUND: In vivo reflectance confocal microscopy (RCM) is well established in non-melanoma skin cancer detection and screening. However, there is no sufficient validation regarding intraoperatively obtained images of wound margins. A reliable and fast resection margin detection is of high clinical relevance. Hence, we aimed to investigate feasibility and validity of in vivo RCM imaging for wound margins assessment compared with standard skin surface imaging and the gold standard histopathology. METHODS: A surgical incision through the center of a large basal cell carcinoma (BCC) affected area in the head and face region was performed. After removing half of the tumor, the wound margins of the remaining half as well as the corresponding skin surface were scanned with an in vivo RCM. A total of 50 wound margin images with BCC, 50 images of BCC-free margins and the corresponding skin surface images from 50 patients were compared with each other and with histopathological findings. Presence of confocal diagnostic criteria for BCC in images was analyzed. RESULTS: An overall sensitivity and specificity in detection of BCC in wound margins was 88.5%, and 91.7% compared to skin surface imaging and 97.8% and 90.7%, respectively, compared to histopathology. We identified all known confocal patterns of healthy skin and BCC in wound margin scans: damage of the epidermal layer above the lesion and cellular pleomorphism, elongated and monomorphic basaloid nuclei, nuclear polarization, an increased number of dilated blood vessels with high leukocyte traffic, inflammatory cells. CONCLUSIONS: The accuracy of in vivo RCM imaging of wound margins is comparable with a standard skin surface imaging. The intraoperative detection of BCC areas in wound margins is as precise as the standard skin imaging and may be supportive for surgical interventions.

Below the horizon-the physics of extreme visual ranges.

Visual ranges of up to 440 km have recently been documented by photographs of ground-based observers. A report from 1948 claimed a record visual range from a plane of more than 530 km and a similar recent observation from 2017 was documented by a photo. Such extreme visual ranges can in principle be explained by the interplay of refraction and light scattering. However, they require optimal atmospheric conditions, and cleverly chosen locations and times.

Detection of oral squamous cell carcinoma with ex vivo fluorescence confocal microscopy: Sensitivity and specificity compared to histopathology.

Real-time microscopic imaging of freshly excised tissue enables a rapid bedside-pathology. A possible application of interest is the detection of oral squamous cell carcinomas (OSCCs). The aim of this study was to analyze the sensitivity and specificity of ex vivo fluorescence confocal microscopy (FCM) for OSCCs and to compare confocal images visually and qualitatively with gold standard histopathology. Two hundred eighty ex vivo FCM images were prospectively collected and evaluated immediately after excision. Every confocal image was blindly assessed for the presence or absence of malignancy by two clinicians and one pathologist. The results were compared with conventional histopathology with hematoxylin and eosin staining. OSCCs were detected with a very high sensitivity of 0.991, specificity of 0.9527, positive predictive value of 0.9322 and negative predictive value of 0.9938. The results demonstrate the potential of ex vivo FCM in fresh tissue for rapid real-time surgical pathology.

Extended visual range during solar eclipses.

During the solar eclipse on 21 August 2017, observed from Rexburg, Idaho, the eastern horizon for a short time period close to totality showed the contours of the Grand Teton mountain range from distances between about 80 and 90 km. In most of the partial phase before and after the eclipse, the visual range was below 80 km and the mountains could not be seen by the naked eye because of wildfire smoke in the air. We present photographic evidence of extension of the visual range during the eclipse and explain it in terms of a simple model for the daytime visual range. In this model, contrast between a black object and the sky increases whenever the ratio of nearby to distant air light reduces. As a consequence, we predict asymmetries with regard to totality and also that similar phenomena may be observed if part of the line of sight close to the observer is in deep shade of a thick cloud cover.

Blue sun glints on water viewed through a polarizer.

Sun glints are formed by specular reflections of the sun from capillary waves formed by wind blowing over water. These glints are normally colorless for a high sun or take on the color of the light source, such as orange-red during sunset or sunrise. However, when the glints are highly polarized by reflection near the Brewster angle, i.e., with relatively high sun they can change from colorless to a blue appearance caused by blue light leakage through a polarizing filter oriented orthogonal to the plane of polarization of the reflected light. Measurements are shown of crossed-polarizer transmission spectra exhibiting blue and near infrared light leakage for photographic polarizing filters and polarized sunglasses. A variety of photographs is shown to confirm blue light leakage as the source of the blue glint color.

Atmospheric optics in the near infrared.

Digital near-infrared photography opens up new observation possibilities and applications for atmospheric optics. We discuss necessary conditions and requirements for observing a variety of atmospheric optical phenomena in the infrared spectral range and report for the first time near-infrared photographs of 22° ring halos and inferior mirages. Our emphasis is on optical phenomena observable in the troposphere, excluding the large body of work addressing near-infrared airglow and aurora.

A Balamuthia survivor.

INTRODUCTION: This case report describes a human survivor of Balamuthia mandrillaris infection. This is a free-living amoeba that can cause infection with the devastating consequence of near universally fatal encephalitis. We report this case to demonstrate the possibility of recovery. CASE PRESENTATION: A 26-year-old Hispanic male, a landscape gardener, presented to the hospital in March 2010 with a two month history of headache, visual disturbances and new-onset seizures. Brain imaging identified two enhancing central lesions and Balamuthia mandrillaris was later identified by brain biopsy. He received several months of various antimicrobials including miltefosine, a novel use of the drug in this disease at the time. Seven weeks into therapy, considerations were made to switch him to 'comfort care' because of worsening clinical status and seemingly lack of response to treatment. The patient finally demonstrated clinical and radiological improvement after eight weeks with modified therapy, despite experiencing some debilitating toxic effects likely to be related to antibiotics. Two years after his initial presentation he made a complete recovery. CONCLUSION: Balamuthia mandrillaris amoebic encephalitis is considered an almost universally fatal disease; this case demonstrates the possibility of recovery. This report outlines his treatment, drug toxicities and includes additional information regarding the therapeutic use of the drug miltefosine. Whether his survival is related to the specific antimicrobials used in this case is unknown and further investigation is warranted.