Designing a Convolutional Neural Network for High-Accuracy Oral Cavity Squamous Cell Carcinoma (OCSCC) Detection

TL;DR

This work tackles late detection of Oral Cavity Squamous Cell Carcinoma (OCSCC) by combining a Convolutional Neural Network (CNN) with a purpose-built image-capturing hardware to improve diagnostic accuracy from mouth images. The CNN is trained on 4,293 labeled images (cancerous, non-cancerous, negative) and evaluated across five input resolutions, revealing a logarithmic increase in accuracy with resolution and a significant uplift when using the hardware accessory. Key contributions include a multi-class CNN for three-way classification, a hardware-assisted image capture system, and an open-access application for testing predictions. The findings suggest practical potential for early, scalable OCSCC screening, particularly in resource-limited settings, while noting dataset diversity as a limitation and outlining directions for larger datasets and cross-platform deployment. $R^2 = 0.996$ is reported for the resolution–accuracy relationship, indicating a strong but diminishing return at higher resolutions.

Abstract

Oral Cavity Squamous Cell Carcinoma (OCSCC) is the most common type of head and neck cancer. Due to the subtle nature of its early stages, deep and hidden areas of development, and slow growth, OCSCC often goes undetected, leading to preventable deaths. However, properly trained Convolutional Neural Networks (CNNs), with their precise image segmentation techniques and ability to apply kernel matrices to modify the RGB values of images for accurate image pattern recognition, would be an effective means for early detection of OCSCC. Pairing this neural network with image capturing and processing hardware would allow increased efficacy in OCSCC detection. The aim of our project is to develop a Convolutional Neural Network trained to recognize OCSCC, as well as to design a physical hardware system to capture and process detailed images, in order to determine the image quality required for accurate predictions. A CNN was trained on 4293 training images consisting of benign and malignant tumors, as well as negative samples, and was evaluated for its precision, recall, and Mean Average Precision (mAP) in its predictions of OCSCC. A testing dataset of randomly assorted images of cancerous, non-cancerous, and negative images was chosen, and each image was altered to represent 5 common resolutions. This test data set was thoroughly analyzed by the CNN and predictions were scored on the basis of accuracy. The designed enhancement hardware was used to capture detailed images, and its impact was scored. An application was developed to facilitate the testing process and bring open access to the CNN. Images of increasing resolution resulted in higher-accuracy predictions on a logarithmic scale, demonstrating the diminishing returns of higher pixel counts.

Figures (8)

• Figure 1: Apple iOS Application displaying an example (non-cancerous) scan with its accompanying result and model confidence.
• Figure 2: Rendering of the image enhancing hardware attachment used to aid with the scanning of the oral cavity.
• Figure 3: The fit of the hardware around an iPhone SE model, demonstrating its ease of attachment and removal.
• Figure 4: Processing of the model with the supports necessary to print in the Cura printer.
• Figure 5: Precision signifies the proportion of correctly identified positive instances against all predicted positives. Recall measures the proportion of actual positive instances correctly detected by the model. Mean Average Precision (AP) represents the area under the precision-recall curve, which is a comprehensive assessment of the model’s performance across varying thresholds.