DCSNet: A Lightweight Knowledge Distillation-Based Model with Explainable AI for Lung Cancer Diagnosis from Histopathological Images

TL;DR

The paper addresses the need for accurate, resource-efficient lung cancer diagnosis from histopathology by applying knowledge distillation (KD) from eight large CNNs to a lightweight Distilled Custom Student Network (DCSNet). ResNet50 is identified as the best teacher, and DCSNet achieves 0.92 accuracy with about 0.66M parameters, aided by Grad-CAM based explanations for transparency. The study systematically analyzes teacher–student configurations and hyperparameters, demonstrating that KD can preserve diagnostic performance while enabling deployment in limited-resource settings, with XAI enabling clinician trust. This work advances deployable, interpretable AI tools in pathology by combining KD with explainability and focusing on practical efficiency gains.

Abstract

Lung cancer is a leading cause of cancer-related deaths globally, where early detection and accurate diagnosis are critical for improving survival rates. While deep learning, particularly convolutional neural networks (CNNs), has revolutionized medical image analysis by detecting subtle patterns indicative of early-stage lung cancer, its adoption faces challenges. These models are often computationally expensive and require significant resources, making them unsuitable for resource constrained environments. Additionally, their lack of transparency hinders trust and broader adoption in sensitive fields like healthcare. Knowledge distillation addresses these challenges by transferring knowledge from large, complex models (teachers) to smaller, lightweight models (students). We propose a knowledge distillation-based approach for lung cancer detection, incorporating explainable AI (XAI) techniques to enhance model transparency. Eight CNNs, including ResNet50, EfficientNetB0, EfficientNetB3, and VGG16, are evaluated as teacher models. We developed and trained a lightweight student model, Distilled Custom Student Network (DCSNet) using ResNet50 as the teacher. This approach not only ensures high diagnostic performance in resource-constrained settings but also addresses transparency concerns, facilitating the adoption of AI-driven diagnostic tools in healthcare.

Figures (7)

• Figure 1: Workflow of the proposed system
• Figure 2: Images of three types of lung cancer
• Figure 3: Student model learning from teacher model and ground truth (True Label)
• Figure 4: DCSNet Architecture
• Figure 5: Confusion Matrices of eight teacher models