Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CC-DCNet: Dynamic Convolutional Neural Network with Contrastive Constraints for Identifying Lung Cancer Subtypes on Multi-modality Images

Yuan Jin, Gege Ma, Geng Chen, Tianling Lyu, Jan Egger, Junhui Lyu, Shaoting Zhang, Wentao Zhu

TL;DR

This work tackles lung cancer subtype identification by leveraging multi-modality imaging (CT and pathology) with a dynamic convolutional network. The proposed CC-DCNet adapts to paired or CT-only inputs and uses a contrastive loss to distill pathology priors into radiology feature extraction, improving discrimination of subtypes. Through extensive multi-center experiments and ablations, the method demonstrates superior ACC, AUC, and F1-score, and shows robust generalization across institutions. The approach offers a practical pathway for better CT-based diagnosis when pathology is unavailable and highlights broader potential for multi-modality integration in medical imaging.

Abstract

The accurate diagnosis of pathological subtypes of lung cancer is of paramount importance for follow-up treatments and prognosis managements. Assessment methods utilizing deep learning technologies have introduced novel approaches for clinical diagnosis. However, the majority of existing models rely solely on single-modality image input, leading to limited diagnostic accuracy. To this end, we propose a novel deep learning network designed to accurately classify lung cancer subtype with multi-dimensional and multi-modality images, i.e., CT and pathological images. The strength of the proposed model lies in its ability to dynamically process both paired CT-pathological image sets as well as independent CT image sets, and consequently optimize the pathology-related feature extractions from CT images. This adaptive learning approach enhances the flexibility in processing multi-dimensional and multi-modality datasets and results in performance elevating in the model testing phase. We also develop a contrastive constraint module, which quantitatively maps the cross-modality associations through network training, and thereby helps to explore the "gold standard" pathological information from the corresponding CT scans. To evaluate the effectiveness, adaptability, and generalization ability of our model, we conducted extensive experiments on a large-scale multi-center dataset and compared our model with a series of state-of-the-art classification models. The experimental results demonstrated the superiority of our model for lung cancer subtype classification, showcasing significant improvements in accuracy metrics such as ACC, AUC, and F1-score.

CC-DCNet: Dynamic Convolutional Neural Network with Contrastive Constraints for Identifying Lung Cancer Subtypes on Multi-modality Images

TL;DR

This work tackles lung cancer subtype identification by leveraging multi-modality imaging (CT and pathology) with a dynamic convolutional network. The proposed CC-DCNet adapts to paired or CT-only inputs and uses a contrastive loss to distill pathology priors into radiology feature extraction, improving discrimination of subtypes. Through extensive multi-center experiments and ablations, the method demonstrates superior ACC, AUC, and F1-score, and shows robust generalization across institutions. The approach offers a practical pathway for better CT-based diagnosis when pathology is unavailable and highlights broader potential for multi-modality integration in medical imaging.

Abstract

The accurate diagnosis of pathological subtypes of lung cancer is of paramount importance for follow-up treatments and prognosis managements. Assessment methods utilizing deep learning technologies have introduced novel approaches for clinical diagnosis. However, the majority of existing models rely solely on single-modality image input, leading to limited diagnostic accuracy. To this end, we propose a novel deep learning network designed to accurately classify lung cancer subtype with multi-dimensional and multi-modality images, i.e., CT and pathological images. The strength of the proposed model lies in its ability to dynamically process both paired CT-pathological image sets as well as independent CT image sets, and consequently optimize the pathology-related feature extractions from CT images. This adaptive learning approach enhances the flexibility in processing multi-dimensional and multi-modality datasets and results in performance elevating in the model testing phase. We also develop a contrastive constraint module, which quantitatively maps the cross-modality associations through network training, and thereby helps to explore the "gold standard" pathological information from the corresponding CT scans. To evaluate the effectiveness, adaptability, and generalization ability of our model, we conducted extensive experiments on a large-scale multi-center dataset and compared our model with a series of state-of-the-art classification models. The experimental results demonstrated the superiority of our model for lung cancer subtype classification, showcasing significant improvements in accuracy metrics such as ACC, AUC, and F1-score.
Paper Structure (23 sections, 6 equations, 6 figures, 4 tables)

This paper contains 23 sections, 6 equations, 6 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Dynamic Convolutional Learning
  4. Feature Extraction
  5. Feature Concatenation
  6. Dynamic Convolution
  7. Contrastive Loss Function
  8. The Overall Loss Function
  9. Evaluation Metrics
  10. Experimental Setup
  11. Datasets and Preprocessing
  12. Pathological Image Preprocessing
  13. CT Image Preprocessing
  14. Implementation Details
  15. Experimental Results
  16. ...and 8 more sections

Figures (6)

  • Figure 1: The pipeline of the proposed novel lung cancer subtypes classification model, CC-DCNet.
  • Figure 2: The diagram of contrastive Loss.
  • Figure 3: Key part of SwitchNet (Our proposed additional benchmark model).
  • Figure 4: P-values obtained from paired-sample t-tests that were conducted between proposed CC-DCNet and other SOTAs for ACC, AUC and F1 SCore: (a) validated with testing CT inputs, (b) validated with testing paired CT/pathological inputs (the tests of RAFENet and CARL were skipped). All p-values are less than the significance level (0.05).
  • Figure 5: The ablation test of additional independent CT dataset and contrastive constraints in terms of ACC(%), AUC(%) and F1-Score(%), validated with testing CT input (a) and paired CT/pathological input (b).
  • ...and 1 more figures