Predicting HPV association using deep learning and regular H&E stains allows granular stratification of oropharyngeal cancer patients

Analysis of AI foundation model features decodes the histopathologic landscape of HPV-positive head and neck squamous cell carcinomas

OBJECTIVES

Human papillomavirus (HPV) influences the pathobiology of Head and Neck Squamous Cell Carcinomas (HSNCCs). While deep learning shows promise in detecting HPV from hematoxylin and eosin (H&E) stained slides, the histologic features utilized remain unclear. This study leverages artificial intelligence (AI) foundation models to characterize histopathologic features associated with HPV presence and objectively describe patterns of variability in the HPV-positive space.

MATERIALS AND METHODS

H&E images from 981 HNSCC patients across public and institutional datasets were analyzed. We used UNI, a foundation model based on self-supervised learning (SSL), to map the landscape of HNSCC histology and identify the axes of SSL features that best separate HPV-positive and HPV-negative tumors. To interpret the histologic features that vary across different regions of this landscape, we used HistoXGAN, a pretrained generative adversarial network (GAN), to generate synthetic histology images from SSL features, which a pathologist rigorously assessed.

RESULTS

Analyzing AI-generated synthetic images found distinctive features of HPV-positive histology, such as smaller, paler, more monomorphic nuclei; purpler, amphophilic cytoplasm; and indistinct cell borders with rounded tumor contours. The SSL feature axes we identified enabled accurate prediction of HPV status from histology, achieving validation sensitivity and specificity of 0.81 and 0.92, respectively. Our analysis subdivided image tiles from HPV-positive histology into three overlapping subtypes: border, inflamed, and stroma.

CONCLUSION

Foundation-model-derived synthetic pathology images effectively capture HPV-related histology. Our analysis identifies distinct subtypes within HPV-positive HNSCCs and enables accurate, explainable detection of HPV presence directly from histology, offering a valuable approach for low-resource clinical settings.

Ligand-receptor interactions combined with histopathology for improved prognostic modeling in HPV-negative head and neck squamous cell carcinoma

Head and neck squamous cell carcinoma (HNSC) is a prevalent malignancy, with HPV-negative tumors exhibiting aggressive behavior and poor prognosis. Understanding the intricate interactions within the tumor microenvironment (TME) is crucial for improving prognostic models and identifying therapeutic targets. Using BulkSignalR, we identified ligand-receptor interactions in HPV-negative TCGA-HNSC cohort (n = 395). A prognostic model incorporating 14 ligand-receptor pairs was developed using random forest survival analysis and LASSO-penalized Cox regression based on overall survival and progression-free interval of HPV-negative tumors from TCGA-HNSC. Multi-omics analysis revealed distinct molecular features between risk groups, including differences in extracellular matrix remodeling, angiogenesis, immune infiltration, and APOBEC enzyme activity. Deep learning-based tissue morphology analysis on HE-stained whole slide images further improved risk stratification, with region selection via Silicon enhancing accuracy. The integration of routine histopathology with deep learning and multi-omics data offers a clinically accessible tool for precise risk stratification, facilitating personalized treatment strategies in HPV-negative HNSC.

RL-Cervix.Net: A Hybrid Lightweight Model Integrating Reinforcement Learning for Cervical Cell Classification

Background: Reinforcement learning (RL) represents a significant advancement in artificial intelligence (AI), particularly for complex sequential decision-making challenges. Its capability to iteratively refine decisions makes it ideal for applications in medicine, such as the detection of cervical cancer; a major cause of mortality among women globally. The Pap smear test, a crucial diagnostic tool for cervical cancer, benefits from enhancements in AI, facilitating the development of automated diagnostic systems that improve screening effectiveness. This research introduces RL-Cervix.Net, a hybrid model integrating RL with convolutional neural network (CNN) technologies, aimed at elevating the precision and efficiency of cervical cancer screenings. Methods: RL-Cervix.Net combines the robust ResNet-50 architecture with a reinforcement learning module tailored for the unique challenges of cytological image analysis. The model was trained and validated using three extensive public datasets to ensure its effectiveness under realistic conditions. A novel application of RL for dynamic feature refinement and adjustment based on reward functions was employed to optimize the detection capabilities of the model. Results: The innovative integration of RL into the CNN framework allowed RL-Cervix.Net to achieve an unprecedented classification accuracy of 99.98% in identifying atypical cells indicative of cervical lesions. The model demonstrated superior accuracy and interpretability compared to existing methods, addressing variability and complexities inherent in cytological images. Conclusions: The RL-Cervix.Net model marks a significant breakthrough in the application of AI for medical diagnostics, particularly in the early detection of cervical cancer. By significantly improving diagnostic accuracy and efficiency, RL-Cervix.Net has the potential to enhance patient outcomes through earlier and more precise identification of the disease, ultimately contributing to reduced mortality rates and improved healthcare delivery.

Deep learning for predicting prognostic consensus molecular subtypes in cervical cancer from histology images

Cervical cancer remains the fourth most common cancer among women worldwide. This study proposes an end-to-end deep learning framework to predict consensus molecular subtypes (CMS) in HPV-positive cervical squamous cell carcinoma (CSCC) from H&E-stained histology slides. Analysing three CSCC cohorts (n = 545), we show our Digital-CMS scores significantly stratify patients by both disease-specific (TCGA p = 0.0022, Oslo p = 0.0495) and disease-free (TCGA p = 0.0495, Oslo p = 0.0282) survival. In addition, our extensive tumour microenvironment analysis reveals differences between the two CMS subtypes, with CMS-C1 tumours exhibit increased lymphocyte presence, while CMS-C2 tumours show high nuclear pleomorphism, elevated neutrophil-to-lymphocyte ratio, and higher malignancy, correlating with poor prognosis. This study introduces a potentially clinically advantageous Digital-CMS score derived from digitised WSIs of routine H&E-stained tissue sections, offers new insights into TME differences impacting patient prognosis and potential therapeutic targets, and identifies histological patterns serving as potential surrogate markers of the CMS subtypes for clinical application.

Computer Vision-Radiomics & Pathognomics

The role of computer vision in extracting radiographic (radiomics) and histopathologic (pathognomics) features is an extension of molecular biomarkers that have been foundational to our understanding across the spectrum of head and neck disorders. Especially within head and neck cancers, machine learning and deep learning applications have yielded advances in the characterization of tumor features, nodal features, and various outcomes. This review aims to overview the landscape of radiomic and pathognomic applications, informing future work to address gaps. Novel methodologies will be needed to potentially engineer ways of integrating multidimensional data inputs to examine disease features to guide prognosis comprehensively and ultimately clinical management.

Imaging Modalities for Head and Neck Cancer: Present and Future

Several imaging modalities are utilized in the diagnosis, treatment, and surveillance of head and neck cancer. First-line imaging remains computed tomography (CT); however, MRI, PET with CT (PET/CT), and ultrasound are often used. In the last decade, several new imaging modalities have been developed that have the potential to improve early detection, modify treatment, decrease treatment morbidity, and augment surveillance. Among these, molecular imaging, lymph node mapping, and adjustments to endoscopic techniques are promising. The present review focuses on existing imaging, novel techniques, and the recent changes to imaging practices within the field.

Genome composition-based deep learning predicts oncogenic potential of HPVs

Human papillomaviruses (HPVs) account for more than 30% of cancer cases, with definite identification of the oncogenic role of viral E6 and E7 genes. However, the identification of high-risk HPV genotypes has largely relied on lagged biological exploration and clinical observation, with types unclassified and oncogenicity unknown for many HPVs. In the present study, we retrieved and cleaned HPV sequence records with high quality and analyzed their genomic compositional traits of dinucleotide (DNT) and DNT representation (DCR) to overview the distribution difference among various types of HPVs. Then, a deep learning model was built to predict the oncogenic potential of all HPVs based on E6 and E7 genes. Our results showed that the main three groups of Alpha, Beta, and Gamma HPVs were clearly separated between/among types in the DCR trait for either E6 or E7 coding sequence (CDS) and were clustered within the same group. Moreover, the DCR data of either E6 or E7 were learnable with a convolutional neural network (CNN) model. Either CNN classifier predicted accurately the oncogenicity label of high and low oncogenic HPVs. In summary, the compositional traits of HPV oncogenicity-related genes E6 and E7 were much different between the high and low oncogenic HPVs, and the compositional trait of the DCR-based deep learning classifier predicted the oncogenic phenotype accurately of HPVs. The trained predictor in this study will facilitate the identification of HPV oncogenicity, particularly for those HPVs without clear genotype or phenotype.