Non-invasive scoring of cellular atypia in keratinocyte cancers in 3D LC-OCT images using Deep Learning

Sébastien Fischman,Pietro Rubegni,Susana Puig,Jilliana Monnier,Javiera Pérez-Anker,Véronique Del Marmol,Jean-Luc Perrot,Arnaud Dubois,Angelo Di Naro,Josep Malvehy,Théo Viel,Linda Tognetti,Elisa Cinotti,Mariano Suppa

doi:10.1038/s41598-021-04395-1

Abstract

Diagnosis based on histopathology for skin cancer detection is today’s gold standard and relies on the presence or absence of biomarkers and cellular atypia. However it suffers drawbacks: it requires a strong expertise and is time-consuming. Moreover the notion of atypia or dysplasia of the visible cells used for diagnosis is very subjective, with poor inter-rater agreement reported in the literature. Lastly, histology requires a biopsy which is an invasive procedure and only captures a small sample of the lesion, which is insufficient in the context of large fields of cancerization. Here we demonstrate that the notion of cellular atypia can be objectively defined and quantified with a non-invasive in-vivo approach in three dimensions (3D). A Deep Learning (DL) algorithm is trained to segment keratinocyte (KC) nuclei from Line-field Confocal Optical Coherence Tomography (LC-OCT) 3D images. Based on these segmentations, a series of quantitative, reproducible and biologically relevant metrics is derived to describe KC nuclei individually. We show that, using those metrics, simple and more complex definitions of atypia can be derived to discriminate between healthy and pathological skins, achieving Area Under the ROC Curve (AUC) scores superior than 0.965, largely outperforming medical experts on the same task with an AUC of 0.766. All together, our approach and findings open the door to a precise quantitative monitoring of skin lesions and treatments, offering a promising non-invasive tool for clinical studies to demonstrate the effects of a treatment and for clinicians to assess the severity of a lesion and follow the evolution of pre-cancerous lesions over time.

Highlights

Histopathology is the gold standard to confirm a diagnosis in all tissues
This study proposes a novel automated approach based on deep learning segmentation applied to 3D Line-field Confocal Optical Coherence Tomography (LC-Optical Coherence Tomography (OCT)) images, capable of accurately assessing the amount of atypia in keratinocyte cancers
To define atypia at the cellular level, we considered four approaches based on the quantitative metrics we obtained from deep learning segmentation (Table 2), extending previous qualitative studies on grading KC atypia with Reflectance Confocal Microscopy 2D images[14]

Summary

Introduction

Histopathology is the gold standard to confirm a diagnosis in all tissues. The advent of numerical technologies facilitate the access of physicists to digital imaging. Zhou et al.[8] applied a graph neural network approach from nuclei segmentation in order to automatically grade colorectal cancer histology images. All these studies tried to capture the complexity of the spatial distribution of cell nuclei from histology slides into a few quantitative metrics to demonstrate predictive or discriminative power. LC-OCT is a new in-vivo non-invasive medical imaging technology that combines deep penetration and cellular resolution in 3D12 It allows to study cell nuclei distributions without the sliding procedure deformation and renders information in 3D. Such an approach does not allow to objectively define atypia nor systematically reproduce the results, and no absolute atypia score is generated, only a relative score among a fixed set of images

Methods

Results

Conclusion