[Paper Review] Hybrid deep convolution model for lung cancer detection with transfer learning

Sugandha Saxena, S N Prasad+2 more|arXiv (Cornell University)|Jan 6, 2025

Lung Cancer Diagnosis and Treatment3 citations

TL;DR

The paper proposes a hybrid deep convolution model named Maximum Sensitivity Neural Network (MSNN) that uses transfer learning to improve lung cancer detection, aiming to enhance sensitivity and specificity on CT scans.

ABSTRACT

Advances in healthcare research have significantly enhanced our understanding of disease mechanisms, diagnostic precision, and therapeutic options. Yet, lung cancer remains one of the leading causes of cancer-related mortality worldwide due to challenges in early and accurate diagnosis. While current lung cancer detection models show promise, there is considerable potential for further improving the accuracy for timely intervention. To address this challenge, we introduce a hybrid deep convolution model leveraging transfer learning, named the Maximum Sensitivity Neural Network (MSNN). MSNN is designed to improve the precision of lung cancer detection by refining sensitivity and specificity. This model has surpassed existing deep learning approaches through experimental validation, achieving an accuracy of 98% and a sensitivity of 97%. By overlaying sensitivity maps onto lung Computed Tomography (CT) scans, it enables the visualization of regions most indicative of malignant or benign classifications. This innovative method demonstrates exceptional performance in distinguishing lung cancer with minimal false positives, thereby enhancing the accuracy of medical diagnoses.

Motivation & Objective

• Address the need for more accurate and timely lung cancer diagnosis.
• Develop a hybrid deep convolution architecture leveraging transfer learning to improve detection performance.
• Enhance sensitivity and reduce false positives in lung cancer classification from CT scans.

Proposed method

• Introduce a hybrid deep convolution model called Maximum Sensitivity Neural Network (MSNN).
• Utilize transfer learning to refine sensitivity and specificity in lung cancer detection.
• Overlay sensitivity maps on lung CT scans to visualize regions driving malignant/benign classifications.
• Report experimental validation showing high accuracy and sensitivity.
• Address reported mistakes and data interpretation issues noted in the paper.
• Acknowledge withdrawal of the version and updated context.

Experimental results

Research questions

• RQ1 Can a hybrid deep convolution model with transfer learning achieve high accuracy in lung cancer detection from CT scans?
• RQ2 Does MSNN provide improved sensitivity and reduced false positives compared to existing deep learning approaches?
• RQ3 Do sensitivity maps effectively visualize regions indicative of malignant or benign classifications on CT scans?

Key findings

• The model claims accuracy of 98% and sensitivity of 97% based on experimental validation.
• MSNN overlays sensitivity maps on CT scans to highlight regions influencing the decision.
• The authors acknowledge mistakes and data misinterpretations in the reported version.
• The paper has been withdrawn (v2) by one author, indicating issues with the initial claims.
• The work is positioned within computer vision and AI for medical imaging.
• No license or formal citation details provided due to withdrawal.