Rapid and Label-Free Histopathology of Oral Lesions Using Deep Learning Applied to Optical and Infrared Spectroscopic Imaging Data.

Oral potentially malignant disorders (OPMDs) are precursors to over 80% of oral cancers. Hematoxylin and eosin (H&E) staining, followed by pathologist interpretation of tissue and cellular morphology, is the current gold standard for diagnosis. However, this method is qualitative, can result in errors during the multi-step diagnostic process, and results may have significant inter-observer variability. Chemical imaging (CI) offers a promising alternative, wherein label-free imaging is used to record both the morphology and the composition of tissue and artificial intelligence (AI) is used to objectively assign histologic information. Here, we employ quantum cascade laser (QCL)-based discrete frequency infrared (DFIR) chemical imaging to record data from oral tissues. In this proof-of-concept study, we focused on achieving tissue segmentation into three classes (connective tissue, dysplastic epithelium, and normal epithelium) using a convolutional neural network (CNN) applied to three bands of label-free DFIR data with paired darkfield visible imaging. Using pathologist-annotated H&E images as the ground truth, we demonstrate results that are 94.5% accurate with the ground truth using combined information from IR and darkfield microscopy in a deep learning framework. This chemical-imaging-based workflow for OPMD classification has the potential to enhance the efficiency and accuracy of clinical oral precancer diagnosis.

Community cancer screening at primary care level in Northern India: determinants and policy implications for cancer prevention.

OBJECTIVE: Despite the established cancer screening programme for oral, breast and cervical cancer by the Government of India, the screening coverage remains inadequate. This study aimed to describe the determinants for oral, breast and cervical cancer prevention in a rural community at the primary care level of Northern India and its policy implications. DESIGN: This was a camp-based project conducted for 1 year, using oral visual examination, clinical breast examination and visual inspection of cervix by application of 5% acetic acid according to primary healthcare operational guidelines. During the project, screen-positive participants were followed through reverse navigation. Information about socio-demographic profile, clinical and behavioural history and screening were collected. Predictors for screen-positivity and follow-up compliance were identified through multivariable analysis. SETTINGS: Based on the aim of project, one of the remotely located and low socioeconomic rural blocks, having 148 villages (estimated population of 254 285) in Varanasi district, India was selected as the service site. There is an established healthcare delivery and referral system as per the National Health Mission of Government of India. Oral, breast, gallbladder and cervical cancers are the leading cancers in the district. PARTICIPANTS: We invited all men and women aged 30-65 years residing in the selected block for the last 6 months for the screening camps. Unmarried women, women with active vaginal bleeding, those currently pregnant and those who have undergone hysterectomy were excluded from cervical cancer screening. RESULTS: A total of 14 338 participants were screened through 190 camps and the majority (61.9%) were women. Hindu religion, tobacco use, intention to quit tobacco and presence of symptoms were significantly associated with screen-positivity. Nearly one-third (220; 30.1%) of the screened-positives complied with follow-up. Young age and illiteracy were significantly associated with lower compliance. CONCLUSION: Poor follow-up compliance, despite the availability of tertiary cancer care, patient navigation, free transportation and diagnostic services, calls for research to explore the role of contextual factors and develop pragmatic interventions to justify 'close the care gap'. Community cancer screening needs strengthening through cancer awareness, establishing referral system and integration with the National Tobacco Control and Cancer Registry Programmes.

AI-Assisted Screening of Oral Potentially Malignant Disorders Using Smartphone-Based Photographic Images.

The prevalence of oral potentially malignant disorders (OPMDs) and oral cancer is surging in low- and middle-income countries. A lack of resources for population screening in remote locations delays the detection of these lesions in the early stages and contributes to higher mortality and a poor quality of life. Digital imaging and artificial intelligence (AI) are promising tools for cancer screening. This study aimed to evaluate the utility of AI-based techniques for detecting OPMDs in the Indian population using photographic images of oral cavities captured using a smartphone. A dataset comprising 1120 suspicious and 1058 non-suspicious oral cavity photographic images taken by trained front-line healthcare workers (FHWs) was used for evaluating the performance of different deep learning models based on convolution (DenseNets) and Transformer (Swin) architectures. The best-performing model was also tested on an additional independent test set comprising 440 photographic images taken by untrained FHWs (set I). DenseNet201 and Swin Transformer (base) models show high classification performance with an F1-score of 0.84 (CI 0.79-0.89) and 0.83 (CI 0.78-0.88) on the internal test set, respectively. However, the performance of models decreases on test set I, which has considerable variation in the image quality, with the best F1-score of 0.73 (CI 0.67-0.78) obtained using DenseNet201. The proposed AI model has the potential to identify suspicious and non-suspicious oral lesions using photographic images. This simplified image-based AI solution can assist in screening, early detection, and prompt referral for OPMDs.

Interpretable and Reliable Oral Cancer Classifier with Attention Mechanism and Expert Knowledge Embedding via Attention Map.

Convolutional neural networks have demonstrated excellent performance in oral cancer detection and classification. However, the end-to-end learning strategy makes CNNs hard to interpret, and it can be challenging to fully understand the decision-making procedure. Additionally, reliability is also a significant challenge for CNN based approaches. In this study, we proposed a neural network called the attention branch network (ABN), which combines the visual explanation and attention mechanisms to improve the recognition performance and interpret the decision-making simultaneously. We also embedded expert knowledge into the network by having human experts manually edit the attention maps for the attention mechanism. Our experiments have shown that ABN performs better than the original baseline network. By introducing the Squeeze-and-Excitation (SE) blocks to the network, the cross-validation accuracy increased further. Furthermore, we observed that some previously misclassified cases were correctly recognized after updating by manually editing the attention maps. The cross-validation accuracy increased from 0.846 to 0.875 with the ABN (Resnet18 as baseline), 0.877 with SE-ABN, and 0.903 after embedding expert knowledge. The proposed method provides an accurate, interpretable, and reliable oral cancer computer-aided diagnosis system through visual explanation, attention mechanisms, and expert knowledge embedding.

Status and Determinants of Early Detection of Oral Premalignant and Malignant Lesions in India.

It has been over four decades since the launch of the National Cancer Control Programme in India, yet the cancer screening rates for oral cancer remain unremarkable. Moreover, India is bracing a large burden of oral cancer with poor survival rates. An effective public health programme implementation relies on a multitude of factors related to cost-effective evidence-based interventions, the healthcare delivery system, public health human resource management, community behaviour, partnership with stakeholders, identifying opportunities and political commitment. In this context, we discuss the various challenges in the early detection of oral premalignant and malignant lesions and potential solutions.

Consensus guidelines on management of oral potentially malignant disorders.

Oral cancer is usually preceded by oral potentially malignant disorders (OPMDs) and early detection can downstage the disease. The majority of OPMDs are asymptomatic in early stages and can be detected on routine oral examination. Though only a proportion of OPMDs may transform to oral squamous cell carcinoma (OSCC), they may serve as a surrogate clinical lesion to identify individuals at risk of developing OSCC. Currently, there is a scarcity of scientific evidence on specific interventions and management of OPMDs and there is no consensus regarding their management. A consensus meeting with a panel of experts was convened to frame guidelines for clinical practices and recommendations for management strategies for OPMDs. A review of literature from medical databases was conducted to provide the best possible evidence and provide recommendations in management of OPMDs.

Exploring uncertainty measures in convolutional neural network for semantic segmentation of oral cancer images.

Significance: Oral cancer is one of the most prevalent cancers, especially in middle- and low-income countries such as India. Automatic segmentation of oral cancer images can improve the diagnostic workflow, which is a significant task in oral cancer image analysis. Despite the remarkable success of deep-learning networks in medical segmentation, they rarely provide uncertainty quantification for their output. Aim: We aim to estimate uncertainty in a deep-learning approach to semantic segmentation of oral cancer images and to improve the accuracy and reliability of predictions. Approach: This work introduced a UNet-based Bayesian deep-learning (BDL) model to segment potentially malignant and malignant lesion areas in the oral cavity. The model can quantify uncertainty in predictions. We also developed an efficient model that increased the inference speed, which is almost six times smaller and two times faster (inference speed) than the original UNet. The dataset in this study was collected using our customized screening platform and was annotated by oral oncology specialists. Results: The proposed approach achieved good segmentation performance as well as good uncertainty estimation performance. In the experiments, we observed an improvement in pixel accuracy and mean intersection over union by removing uncertain pixels. This result reflects that the model provided less accurate predictions in uncertain areas that may need more attention and further inspection. The experiments also showed that with some performance compromises, the efficient model reduced computation time and model size, which expands the potential for implementation on portable devices used in resource-limited settings. Conclusions: Our study demonstrates the UNet-based BDL model not only can perform potentially malignant and malignant oral lesion segmentation, but also can provide informative pixel-level uncertainty estimation. With this extra uncertainty information, the accuracy and reliability of the model's prediction can be improved.

Feasibility of Training Community Health Workers in the Detection of Oral Cancer.

Importance: Visual screening for oral cancer has been found to be useful in a large randomized clinical trial in Kerala, India, showing substantial reduction in mortality. To address the shortage of medical personnel in resource-deficient regions, using the services of community health workers has been proposed as a strategy to fill the gap in human resources in health care. Objective: To assess the feasibility of community health workers in screening and early detection of oral cancer using a mobile application capturing system. Design, Setting, and Participants: A cross-sectional study using a household sample was conducted in 10 areas of Gautam Budhnagar district, Uttar Pradesh, India, from January 31, 2020, to March 31, 2021, to assess the feasibility of identification of oral lesions by community health workers using a mobile phone application compared with diagnosis by trained dentists in a screening clinic. Men and women aged 30 years or older as well as tobacco users younger than 30 years were eligible for screening. Interventions: Screening by trained community health workers vs dentists. Results: A total of 1200 participants were screened by the community health workers during their home visits; of these, 1018 participants (526 [51.7%] men; mean [SD] age, 35 [16] years) were also referred and screened by the dentists a clinic. There was near-perfect agreement (κ = 0.9) between the findings of the community health workers and the dentists in identifying the positive or negative cases with overall sensitivity of 96.69% (95% CI, 94.15%-98.33%) and specificity of identification of 98.69% (95% CI, 97.52%-99.40%). Conclusions and Relevance: In this cross-sectional study, trained community health workers were able after initial supervision by qualified dentists to perform oral cancer screening programs. These findings suggest that community health workers can perform this screening in resource-constrained settings.

Interpretable deep learning approach for oral cancer classification using guided attention inference network.

SIGNIFICANCE: Convolutional neural networks (CNNs) show the potential for automated classification of different cancer lesions. However, their lack of interpretability and explainability makes CNNs less than understandable. Furthermore, CNNs may incorrectly concentrate on other areas surrounding the salient object, rather than the network's attention focusing directly on the object to be recognized, as the network has no incentive to focus solely on the correct subjects to be detected. This inhibits the reliability of CNNs, especially for biomedical applications. AIM: Develop a deep learning training approach that could provide understandability to its predictions and directly guide the network to concentrate its attention and accurately delineate cancerous regions of the image. APPROACH: We utilized Selvaraju et al.'s gradient-weighted class activation mapping to inject interpretability and explainability into CNNs. We adopted a two-stage training process with data augmentation techniques and Li et al.'s guided attention inference network (GAIN) to train images captured using our customized mobile oral screening devices. The GAIN architecture consists of three streams of network training: classification stream, attention mining stream, and bounding box stream. By adopting the GAIN training architecture, we jointly optimized the classification and segmentation accuracy of our CNN by treating these attention maps as reliable priors to develop attention maps with more complete and accurate segmentation. RESULTS: The network's attention map will help us to actively understand what the network is focusing on and looking at during its decision-making process. The results also show that the proposed method could guide the trained neural network to highlight and focus its attention on the correct lesion areas in the images when making a decision, rather than focusing its attention on relevant yet incorrect regions. CONCLUSIONS: We demonstrate the effectiveness of our approach for more interpretable and reliable oral potentially malignant lesion and malignant lesion classification.

Bayesian deep learning for reliable oral cancer image classification.

In medical imaging, deep learning-based solutions have achieved state-of-the-art performance. However, reliability restricts the integration of deep learning into practical medical workflows since conventional deep learning frameworks cannot quantitatively assess model uncertainty. In this work, we propose to address this shortcoming by utilizing a Bayesian deep network capable of estimating uncertainty to assess oral cancer image classification reliability. We evaluate the model using a large intraoral cheek mucosa image dataset captured using our customized device from high-risk population to show that meaningful uncertainty information can be produced. In addition, our experiments show improved accuracy by uncertainty-informed referral. The accuracy of retained data reaches roughly 90% when referring either 10% of all cases or referring cases whose uncertainty value is greater than 0.3. The performance can be further improved by referring more patients. The experiments show the model is capable of identifying difficult cases needing further inspection.

Classification of imbalanced oral cancer image data from high-risk population.

SIGNIFICANCE: Early detection of oral cancer is vital for high-risk patients, and machine learning-based automatic classification is ideal for disease screening. However, current datasets collected from high-risk populations are unbalanced and often have detrimental effects on the performance of classification. AIM: To reduce the class bias caused by data imbalance. APPROACH: We collected 3851 polarized white light cheek mucosa images using our customized oral cancer screening device. We use weight balancing, data augmentation, undersampling, focal loss, and ensemble methods to improve the neural network performance of oral cancer image classification with the imbalanced multi-class datasets captured from high-risk populations during oral cancer screening in low-resource settings. RESULTS: By applying both data-level and algorithm-level approaches to the deep learning training process, the performance of the minority classes, which were difficult to distinguish at the beginning, has been improved. The accuracy of "premalignancy" class is also increased, which is ideal for screening applications. CONCLUSIONS: Experimental results show that the class bias induced by imbalanced oral cancer image datasets could be reduced using both data- and algorithm-level methods. Our study may provide an important basis for helping understand the influence of unbalanced datasets on oral cancer deep learning classifiers and how to mitigate.

Mobile-based oral cancer classification for point-of-care screening.

SIGNIFICANCE: Oral cancer is among the most common cancers globally, especially in low- and middle-income countries. Early detection is the most effective way to reduce the mortality rate. Deep learning-based cancer image classification models usually need to be hosted on a computing server. However, internet connection is unreliable for screening in low-resource settings. AIM: To develop a mobile-based dual-mode image classification method and customized Android application for point-of-care oral cancer detection. APPROACH: The dataset used in our study was captured among 5025 patients with our customized dual-modality mobile oral screening devices. We trained an efficient network MobileNet with focal loss and converted the model into TensorFlow Lite format. The finalized lite format model is ∼16.3 MB and ideal for smartphone platform operation. We have developed an Android smartphone application in an easy-to-use format that implements the mobile-based dual-modality image classification approach to distinguish oral potentially malignant and malignant images from normal/benign images. RESULTS: We investigated the accuracy and running speed on a cost-effective smartphone computing platform. It takes ∼300 ms to process one image pair with the Moto G5 Android smartphone. We tested the proposed method on a standalone dataset and achieved 81% accuracy for distinguishing normal/benign lesions from clinically suspicious lesions, using a gold standard of clinical impression based on the review of images by oral specialists. CONCLUSIONS: Our study demonstrates the effectiveness of a mobile-based approach for oral cancer screening in low-resource settings.

Oral cancer screening using mobile phone-based(mHealth) approach versus conventional oral examination approach, protocol of a cluster randomized study with cost-effectiveness analysis.

INTRODUCTION: Oral cancer is a significant health problem in India. Diagnosis is often delayed. The effectiveness of conventional oral screening has been shown in the Trivandrum oral cancer screening study. The present study will be a step forward to test a mobile phone-based (the mHealth approach) comparing it with the conventional approach. The purpose of this paper is to report the protocol for this study. The primary objective will be to compare both methods in diagnosing oral potentially malignant disorders and cancers. The secondary objective would be to study the cost-effectiveness. METHODS AND ANALYSIS: This will be a cluster-randomized clinical trial of the population in Ernakulam district of Kerala state in India. They will undergo oral cancer screening by community health workers, who will be pre-assigned to the randomly allotted intervention (mHealth) or control (conventional method) clusters. We will enrol a minimum of 9696 subjects from all 6 clusters over 18 months. The cost-effectiveness of the two strategies for oral screening will be determined using data from this randomized controlled trial. The incremental cost per oral cancer/high-risk dysplasia detected, and the incremental cost per life saved will be reported. We will conduct sensitivity and scenario analysis to evaluate the robustness of the findings. ETHICS AND DISSEMINATION: When completed, this will be the first cluster randomized population-based study to test the technology-based approach in India. The knowledge from this study will indicate whether specialists can make a remote diagnosis of oral lesions accurately based on the information gathered using a mobile phone health application and whether the mHealth strategy will be cost-effective in Oral cancer screening. The study will follow the ethical guidelines and will be published in an indexed journal.

Keratinocyte dissociation (desmolysis/acantholysis) in ameloblastoma.

Ameloblastoma is the only odontogenic tumor that displays diversified histomorphological features with subtypes like follicular, plexiform, acanthomatous, granular cell, clear cell, desmoplastic etc. In this paper we presented an extremely unusual presentation of ameloblastoma, which is characterized by desmolysis or acantholysis of stellate reticulum-like cells caused due to keratinocyte dissociation. A 35-year-old male patient presented with a painless hard 3×3 cm swelling in the mandibular right posterior region in the past 4-5 months. Radiographic examination revealed a multilocular radiolucent lesion in the body of mandible with resorption of the roots. Histopathological examination revealed ameloblastic follicles with central cells showing keratinocyte dissociation leading to desmolysis/acantholysis. Desmolytic cells were seen as an isolated entity in the follicular space with round to polygonal shaped morphology. Future retrospective studies on archival samples of ameloblastoma are recommended to relook into identification of such rare phenomenon. This will help in better understanding of the incidence rate and biological behavior of this rare variant of ameloblastoma.

Validation of the use of a fluorescent PARP1 inhibitor for the detection of oral, oropharyngeal and oesophageal epithelial cancers.

For oral, oropharyngeal and oesophageal cancer, the early detection of tumours and of residual tumour after surgery are prognostic factors of recurrence rates and patient survival. Here, we report the validation, in animal models and a human, of the use of a previously described fluorescently labelled small-molecule inhibitor of the DNA repair enzyme poly(ADP-ribose) polymerase 1 (PARP1) for the detection of cancers of the oral cavity, pharynx and oesophagus. We show that the fluorescent contrast agent can be used to quantify the expression levels of PARP1 and to detect oral, oropharyngeal and oesophageal tumours in mice, pigs and fresh human biospecimens when delivered topically or intravenously. The fluorescent PARP1 inhibitor can also detect oral carcinoma in a patient when applied as a mouthwash, and discriminate between fresh biopsied samples of the oral tumour and the surgical resection margin with more than 95% sensitivity and specificity. The PARP1 inhibitor could serve as the basis of a rapid and sensitive assay for the early detection and for the surgical-margin assessment of epithelial cancers of the upper intestinal tract.

Indian clinical practice consensus guidelines for the management of squamous cell carcinoma of head and neck.

Head and neck cancers (HNCs) are malignant tumors of the upper aerodigestive tract and are the sixth most common cancer worldwide. In India, around 30-40% of all cancers are HNCs. Even though there are global guidelines or recommendations for the management of HNCs, these may not be appropriate for Indian scenarios. In an effort to discuss current practices, latest developments and to come to a consensus to recommend management strategies for different anatomical subsites of HNCs for Indian patients, a group of experts (medical, surgical and radiation oncologists and dentists) was formed. A review of literature from medical databases was conducted to provide the best possible evidence base, which was reviewed by experts during a consensus group meeting (January, 2019) to provide recommendations.