Related Experiment Video
Updated: Jun 26, 2025
05:00
Author Spotlight: Enhancing Multicolor Fluorescence Localization in Lung Carcinoma Sample
Published on: November 21, 2023
1.7K
ML3CNet: Non-local means-assisted automatic framework for lung cancer subtypes classification using histopathological
Anurodh Kumar1, Amit Vishwakarma1, Varun Bajaj2
1PDPM Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, 482005, India.
Computer Methods and Programs in Biomedicine
|May 9, 2024
Summary
This study introduces a novel deep learning model for accurate lung cancer classification from histopathological images. The proposed ML3CNet achieves high accuracy, aiding early detection and improving patient outcomes.
Area of Science:
- Oncology
- Medical Imaging
- Computational Pathology
Background:
- Lung cancer (LC) poses a significant global health challenge with high mortality rates.
- Early detection of LC is crucial for improving patient survival and enabling preventative measures.
- Traditional histopathological diagnosis is time-consuming and subjective, necessitating advanced diagnostic tools.
Purpose of the Study:
- To develop an effective convolutional neural network (CNN)-based architecture for classifying lung tissue subtypes using histopathological images.
- To address the limitations of manual inspection and traditional machine learning methods in lung cancer diagnosis.
Main Methods:
- Utilized a nonlocal mean (NLM) filter for noise reduction in histopathological images, preserving image edges.
- Proposed a multi-headed lung cancer classification convolutional neural network (ML3CNet) for image analysis.
- Implemented model quantization to reduce ML3CNet's size for efficient storage and faster processing.
Main Results:
- The ML3CNet model achieved high performance metrics, including 99.72% average classification accuracy, 99.66% sensitivity, and 99.64% precision.
- A quantized version of the model attained 98.92% accuracy.
- The model's applicability was validated on a colon cancer dataset.
Conclusions:
- The proposed approach offers a precise, automated method for classifying lung cancer subtypes, supporting clinical decision-making.
- The ML3CNet model demonstrates potential for practical implementation on hardware like Raspberry Pi for real-world applications.
