Cloud-Based Lung Tumor Detection and Stage Classification Using Deep Learning Techniques.

Area of Science:

• Medical Imaging and Artificial Intelligence
• Oncology and Computer-Aided Diagnosis

Background:

• Artificial intelligence (AI), Internet of Things (IoT), and cloud computing are increasingly utilized in healthcare for enhanced decision-making.
• Positron emission tomography (PET) imaging is crucial for diagnosing various cancers, including lung tumors.

Purpose of the Study:

• To propose a modified computer-aided diagnosis system for challenging lung tumor stage classification.
• To develop a cloud-based system to assist radiologists by reducing workload and providing a second opinion.

Main Methods:

• A hybrid technique, Cloud-based Lung Tumor Detector and Stage Classifier (Cloud-LTDSC), was developed for PET/CT images.
• Lung tumor segmentation was performed using an active contour model, followed by classification using a multilayer convolutional neural network (M-CNN).
• The system was validated using the LIDC-IDRI dataset and lung CT DICOM images.

Main Results:

• The proposed Cloud-LTDSC system achieved high performance metrics, including accuracy, recall, and precision.
• The system demonstrated superior outcomes compared to existing techniques across applied datasets.
• An average lung tumor stage classification accuracy of 97%-99.1% was achieved, with an overall average of 98.6%.

Conclusions:

• The developed Cloud-LTDSC system offers a robust and accurate solution for lung tumor stage classification.
• The hybrid approach integrating AI, cloud computing, and advanced imaging techniques shows significant promise in improving cancer diagnosis.
• This AI-powered tool can effectively aid radiologists in diagnosing and staging pulmonary illnesses.