Multi-Scale Image Segmentation for Early Lung Cancer Detection from CT Scans: A Review

Vinay Vastrakar; Aakanksha Sahu; Nidhi Sharma

Multi-Scale Image Segmentation for Early Lung Cancer Detection from CT Scans: A Review

Author(s)	Mr. Vinay Vastrakar, Ms. Aakanksha Sahu, Ms. Nidhi Sharma
Country	India
Abstract	Lung cancer remains the leading cause of cancer-related mortality worldwide, responsible for approximately 1.8 million deaths annually. Five-year survival exceeds 80% at Stage I but collapses below 10% at Stage IV, making early detection through low-dose CT (LDCT) screening critically important. This paper reviews the evolution of automated pulmonary nodule segmentation across four methodological eras: classical intensity-based methods, CNN-based detection, encoder-decoder architectures (U-Net family), and multi-scale context-aggregation techniques. We then present and critically evaluate a proposed architecture — a Multi-Scale U-Net integrating an Atrous Spatial Pyramid Pooling (ASPP) module with a pre-trained MobileNetV2 encoder — trained and tested on LUNA16 subset0 (~89 CT scans). The proposed model achieves a test Dice of 0.762, IoU of 0.622, Sensitivity of 0.815, and Specificity of 0.9992, outperforming a baseline U-Net without ASPP by 6.8 Dice points overall and by 17 Dice points on small nodules (3–6 mm). Ablation experiments confirm that ImageNet pre-training contributes the single highest-impact improvement (+15 Dice points). The system is designed to be fully reproducible on a CPU-only student laptop. Research limitations and future directions are discussed.
Keywords	Lung Cancer Detection, Pulmonary Nodule Segmentation, CT Scans, U-Net, Atrous Spatial Pyramid Pooling, MobileNetV2, Transfer Learning, LUNA16, Computer-Aided Detection, Deep Learning
Field	Computer > Artificial Intelligence / Simulation / Virtual Reality
Published In	Volume 8, Issue 4, July-August 2026
Published On	2026-07-13

View / Download PDF File

About IJFMR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+91-9687-828-838	editor@ijfmr.com

International Journal For Multidisciplinary Research

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Multi-Scale Image Segmentation for Early Lung Cancer Detection from CT Scans: A Review

Share this