Lung cancer remains one of the dominant causes of cancer-related deaths. Precision is needed in the classification of various types of lung cancers in order to properly diagnose and treat the disease. But certain difficulties are being faced by the conventional approaches in incorporating information from various modalities of medicine, including CT scans and other variables.

FusionNet-LC bridges the gap by developing a combination AI diagnostic tool using both image intelligence and clinical intelligence. It includes a Vision Transformer (ViT) module, which examines the CT scan image of the lungs, as well as an MLP module, which examines the clinical data. The two modules are integrated using an intelligent Hybrid Attention Module. In this way, the diagnostic tool is capable of flagging both interesting image and clinical features. This enables the tool to make a diagnosis with a certain level of accuracy.

By embracing the use of explainable AI and adaptive optimization, the value of the FusionNet-LC increases among oncologists with respect to speed, accuracy, and intensive data.The traditional approaches adopted for cancer diagnosis rely completely on imaging or clinical information. The imaging modality, though strong, could potentially ignore small details visible only after considering variables specific to patients like age, markers, or smokers vs. non-smokers. Similarly, statistical modeling based on clinicalinformation would not

be able to provide details similar to those provided by imaging modality approaches.

There have been major advancements in the area of deep learning and specifically in transformer models with regard to computer vision. There has been another method known as fusion models that have been quite productive in handling multimodal data in medical fields. However, the issue arises when incorporating this data while maintaining interpretability and results with no bias.

The FusionNet-LC bridges this gap through its fusion platform, enabled by artificial intelligence, to synchronize both visual and non-visual modalities to support explainable inference for lung cancer risk and type..