Multimodal widefield fluorescence imaging with nonlinear optical microscopy workflow for noninvasive oral epithelial neoplasia detection: a preclinical study

Fluorescence images of skin lesions and automated diagnosis using convolutional neural networks

In recent years, interest in applying deep learning (DL) to medical diagnosis has rapidly increased, driven primarily by the development of Convolutional Neural Networks and Transformers. Despite advancements in DL, the automated diagnosis of skin cancer remains a significant challenge. Emulating dermatologists, deep learning approaches using clinical images acquired from smartphones and considering patient lesion information have achieved performance levels close to those of specialists. While including clinical information, such as whether the lesion bleeds, hurts, or itches, improves diagnostic metrics, it is insufficient for correctly differentiating some major skin cancer lesions. An alternate technology for diagnosing skin cancer is fluorescence widefield imaging, where the skin lesion is illuminated with excitation light, causing it to emit fluorescence. Since there is no public dataset of fluorescence images for skin lesions, to the best of our knowledge, an effort has been made and resulted in 1,563 fluorescence images of major skin lesions taken by smartphones using the handheld LED wieldfield fluorescence device. The collected images were annotated and analyzed, creating a new dataset named FLUO-SC. Convolutional neural networks were then applied to classify skin lesions using these fluorescence images. Experimental results indicate that fluorescence images are competitive with clinical images (baseline) for classifying major skin lesions and show promising potential for discrimination.

Let it shine: Autofluorescence of Papanicolaou-stain improves AI-based cytological oral cancer detection

BACKGROUND AND OBJECTIVES

Oral cancer is a global health challenge. The disease can be successfully treated if detected early, but the survival rate drops significantly for late stage cases. There is a growing interest in a shift from the current standard of invasive and time-consuming tissue sampling and histological examination, towards non-invasive brush biopsies and cytological examination, facilitating continued risk group monitoring. For cost effective and accurate cytological analysis there is a great need for reliable computer-assisted data-driven approaches. However, infeasibility of accurate cell-level annotation hinders model performance, and limits evaluation and interpretation of the results. This study aims to improve AI-based oral cancer detection by introducing additional information through multimodal imaging and deep multimodal information fusion.

METHODS

We combine brightfield and fluorescence whole slide microscopy imaging to analyze Papanicolaou-stained liquid-based cytology slides of brush biopsies collected from both healthy and cancer patients. Given the challenge of detailed cytological annotations, we utilize a weakly supervised deep learning approach only relying on patient-level labels. We evaluate various multimodal information fusion strategies, including early, late, and three recent intermediate fusion methods.

RESULTS

Our experiments demonstrate that: (i) there is substantial diagnostic information to gain from fluorescence imaging of Papanicolaou-stained cytological samples, (ii) multimodal information fusion improves classification performance and cancer detection accuracy, compared to single-modality approaches. Intermediate fusion emerges as the leading method among the studied approaches. Specifically, the Co-Attention Fusion Network (CAFNet) model achieves impressive results, with an F1 score of 83.34% and an accuracy of 91.79% at cell level, surpassing human performance on the task. Additional tests highlight the importance of accurate image registration to maximize the benefits of the multimodal analysis.

CONCLUSION

This study advances the field of cytopathology by integrating deep learning methods, multimodal imaging and information fusion to enhance non-invasive early detection of oral cancer. Our approach not only improves diagnostic accuracy, but also allows an efficient, yet uncomplicated, clinical workflow. The developed pipeline has potential applications in other cytological analysis settings. We provide a validated open-source analysis framework and share a unique multimodal oral cancer dataset to support further research and innovation.

Clinical, histopathological characteristics and malignant transformation of proliferative verrucous leukoplakia with 36 patients: a retrospective longitudinal study

BACKGROUND

Proliferative verrucous leukoplakia (PVL), distinguished by its malignant transformation rate of 43.87% to 65.8%, stands as the oral potentially malignant disorder with the highest propensity for malignancy. PVL is marked by distinctive heterogeneity regarding the clinical or histopathological characteristics as well as prognostic factors pertinent to this condition. The purpose of this study is to compile and assess the clinicopathological features, malignant transformation, and associated risk factors in patients diagnosed with PVL.

METHODS

This study is a hospital-based retrospective longitudinal study of 36 patients diagnosed with PVL from 2013 to 2023. We conducted complete clinical and histopathological evaluations of the patients.

RESULTS

The cohort comprised 16 males and 20 females, yielding a male-to-female ratio of 1:1.25. The follow-up period ranged from 8 to 125 months, with an average of 47.50 months. The most common clinical type of lesion was the verrucous form (58.33%), and the gingiva was the most common site (44.44%). Each patient had between 2 to 7 lesions, averaging 3.36 per patient. During the follow-up period, twelve patients (33.3%) developed oral cancer, with an average time to malignant transformation of 35.75 months. Kaplan-Meier survival analysis indicated that patients with complaints of pain, roughness, or a rough sensation, with diabetes, and the presence of cytologic atypia histologically showed a higher risk of malignant transformation (p < 0.05). In this study, the rate of malignant transformation in the treatment group (5/23) was lower than that in the untreated group (7/13), however, no statistically significant difference (p = 0.05).

CONCLUSION

The main complaints of pain, roughness, or foreign body sensation, coupled with cytologic atypia histologically are indicative of an increased risk of malignant transformation in PVL. Further research is needed to elucidate the influence of these clinicopathological parameters on the malignant progression of PVL.

Motionless volumetric structured light sheet microscopy

To meet the increasing need for low-cost, compact imaging technology with cellular resolution, we have developed a microLED-based structured light sheet microscope for three-dimensional ex vivo and in vivo imaging of biological tissue in multiple modalities. All the illumination structure is generated directly at the microLED panel-which serves as the source-so light sheet scanning and modulation is completely digital, yielding a system that is simpler and less prone to error than previously reported methods. Volumetric images with optical sectioning are thus achieved in an inexpensive, compact form factor without any moving parts. We demonstrate the unique properties and general applicability of our technique by ex vivo imaging of porcine and murine tissue from the gastrointestinal tract, kidney, and brain.

Multimodal optical imaging with real-time projection of cancer risk and biopsy guidance maps for early oral cancer diagnosis and treatment

SIGNIFICANCE

Despite recent advances in multimodal optical imaging, oral imaging systems often do not provide real-time actionable guidance to the clinician who is making biopsy and treatment decisions.

AIM

We demonstrate a low-cost, portable active biopsy guidance system (ABGS) that uses multimodal optical imaging with deep learning to directly project cancer risk and biopsy guidance maps onto oral mucosa in real time.

APPROACH

Cancer risk maps are generated based on widefield autofluorescence images and projected onto the at-risk tissue using a digital light projector. Microendoscopy images are obtained from at-risk areas, and multimodal image data are used to calculate a biopsy guidance map, which is projected onto tissue.

RESULTS

Representative patient examples highlight clinically actionable visualizations provided in real time during an imaging procedure. Results show multimodal imaging with cancer risk and biopsy guidance map projection offers a versatile, quantitative, and precise tool to guide biopsy site selection and improve early detection of oral cancers.

CONCLUSIONS

The ABGS provides direct visible guidance to identify early lesions and locate appropriate sites to biopsy within those lesions. This represents an opportunity to translate multimodal imaging into real-time clinically actionable visualizations to help improve patient outcomes.

Development of a multimodal mobile colposcope for real-time cervical cancer detection

Cervical cancer remains a leading cause of cancer death among women in low-and middle-income countries. Globally, cervical cancer prevention programs are hampered by a lack of resources, infrastructure, and personnel. We describe a multimodal mobile colposcope (MMC) designed to diagnose precancerous cervical lesions at the point-of-care without the need for biopsy. The MMC integrates two complementary imaging systems: 1) a commercially available colposcope and 2) a high speed, high-resolution, fiber-optic microendoscope (HRME). Combining these two image modalities allows, for the first time, the ability to locate suspicious cervical lesions using widefield imaging and then to obtain co-registered high-resolution images across an entire lesion. The MMC overcomes limitations of high-resolution imaging alone; widefield imaging can be used to guide the placement of the high-resolution imaging probe at clinically suspicious regions and co-registered, mosaicked high-resolution images effectively increase the field of view of high-resolution imaging. Representative data collected from patients referred for colposcopy at Barretos Cancer Hospital in Brazil, including 22,800 high resolution images and 9,900 colposcope images, illustrate the ability of the MMC to identify abnormal cervical regions, image suspicious areas with subcellular resolution, and distinguish between high-grade and low-grade dysplasia.

Optical Imaging in the Diagnosis of OPMDs Malignant Transformation

Oral potentially malignant disorders (OPMDs) are a heterogeneous group of oral lesions with a variable risk of malignant transformation to oral squamous cell carcinoma. The current OPMDs malignant transformation screening depends on conventional oral examination (COE) and is confirmed by biopsy and histologic examination. However, early malignant lesions with subtle mucosal changes are easily unnoticed by COE based on visual inspection and palpation. Optical techniques have been used to determine the biological structure, composition, and function of cells and tissues noninvasively by analyzing the changes in their optical properties. The oral epithelium and stroma undergo persistent structural, functional, and biochemical alterations during malignant transformation, leading to variations in optical tissue properties; optical techniques are thus powerful tools for detecting OPMDs malignant transformation. The optical imaging methods already used to detect OPMDs malignant transformation in vivo include autofluorescence imaging, narrowband imaging, confocal reflectance microscopy, and optical coherence tomography. They exhibit advantages over COE in detecting biochemical or morphologic changes at the molecular or cellular level in vivo; however, limitations also exist. This article comprehensively reviews the various real-time in vivo optical imaging methods used in the adjunctive diagnosis of OPMDs malignant transformation. We focus on the principles of these techniques, review their clinical application, and compare and summarize their advantages and disadvantages. Finally, we conclude with a discussion of current challenges and future directions of this field.