Abstract: Deep learning technologies such as convolutional neural networks (CNN) provide powerful methods for image recognition and have recently been employed in the field of automated carcinoma detection in confocal laser endomicroscopy (CLE) images. CLE is a (sub-)surface microscopic imaging technique that reaches magnifications of up to 1000x and is thus suitable for in vivo structural tissue analysis. In this work, we aim to evaluate the prospects of a priorly developed deep learning-based algorithm targeted at the identification of oral squamous cell carcinoma with regard to its generalization to further anatomic locations of squamous cell carcinomas in the area of head and neck. We applied the algorithm on images acquired from the vocal fold area of five patients with histologically verified squamous cell carcinoma and presumably healthy control images of the clinically normal contra-lateral vocal cord. We find that the network trained on the oral cavity data reaches an accuracy of 89.45% and an area-under-the-curve (AUC) value of 0.955, when applied on the vocal cords data. Compared to the state of the art, we achieve very similar results, yet with an algorithm that was trained on a completely disjunct data set. Concatenating both data sets yielded further improvements in cross-validation with an accuracy of 90.81% and AUC of 0.970. In this study, for the first time to our knowledge, a deep learning mechanism for the identification of oral carcinomas using CLE Images could be applied to other disciplines in the area of head and neck. This study shows the prospect of the algorithmic approach to generalize well on other malignant entities of the head and neck, regardless of the anatomical location and furthermore in an examiner-independent manner.
Introduction
Head and Neck Squamous Cell Carcinoma (HNSCC) is a major global health burden, often diagnosed at late stages when treatment is less effective. The current "gold standard" for diagnosis is an invasive physical biopsy followed by histopathology. While accurate, biopsies are slow, carry risks of infection, and can miss the most aggressive parts of a tumor due to limited sampling.
Confocal Laser Endomicroscopy (CLE) offers a solution: it’s a non-invasive imaging technique that provides 1000x magnification, allowing doctors to see cellular structures in real-time—essentially an "optical biopsy." However, CLE images are notoriously difficult to interpret, requiring extensive specialist training. This paper investigates whether a deep learning algorithm can not only automate this detection but also remain robust when moved from one anatomical site (the oral cavity) to another (the vocal cords).
Methods
The authors employed a patch-based Convolutional Neural Network (CNN) approach to handle the high-resolution, circular field of view characteristic of CLE.
- Patch Extraction: The algorithm breaks the circular CLE field of view into 80x80 pixel patches.
- Classification & Fusion: Each patch is classified independently by the CNN. The resulting probabilities are then "fused" to provide a single classification (cancerous vs. healthy) for the entire image.
- Training Strategy: The model utilized the Adam optimizer within the TensorFlow framework. Evaluation was conducted using Leave-One-Patient-Out (LOPO) cross-validation to ensure the model generalizes to unseen patients rather than memorizing specific cases.
Datasets
The study utilized two primary datasets acquired from different clinical departments:
- Oral Cavity (OC): 7,894 total images from 12 patients, including sub-sites like the inner lip, hard palate, and alveolar ridge.
- Vocal Cords (VC): 4,425 total images from 5 patients.
The OC dataset is considered more diverse and challenging due to a wider variety of tissue types and motion artifacts, as patients were often awake during acquisition.
Key Results and Insights
The study revealed significant findings regarding the robustness of deep learning in clinical settings:
- The "Superset" Advantage: A model trained on the oral cavity generalized remarkably well to the vocal cords, achieving 89.45% accuracy and an AUC of 0.955. The authors suggest the OC dataset acts as a "superset," containing a wider variety of noise and tissue structures that prepare the model for cleaner vocal cord images.
- Asymmetric Generalization: Interestingly, the reverse did not hold true. A model trained only on vocal cords struggled when applied to the oral cavity, reaching only 68.53% accuracy. This highlights that training on "messier," more diverse data is often superior for building robust medical AI.
- Combined Performance: When the datasets were concatenated (OC+VC), the model reached its peak performance with 90.81% accuracy and an AUC of 0.970. This performance is competitive with experienced CLE specialists, who typically score in the 82-92% range.
Discussion and Future Work
The results provide strong evidence that the cellular features of squamous cell carcinoma are consistent enough across different head and neck sites that a single specialized model could potentially serve multiple disciplines, such as ENT and oral surgery.
However, limitations remain. CLE has a low penetration depth, meaning it can detect surface-level epithelial changes but might miss deeper submucosal spread. Additionally, while the image count is high, the patient count (17) is relatively small. Future work will focus on automating quality control for image artifacts and conducting larger, multi-center trials to further validate these findings. Despite these constraints, this study demonstrates the potential for examiner-independent, real-time cancer detection across multiple anatomical sites.
Summary updated