ConvPath

ConvPath classifies cell types in lung adenocarcinoma pathology images using convolutional neural networks to generate spatial maps and extract tumor microenvironment features for prognostic modeling.

Key Features:

• Automated Cell Type Classification Pipeline: Nuclei segmentation followed by CNN-based classification of tumor cells, stromal cells, and lymphocytes.
• Spatial Mapping Capabilities: Identification and classification of cell types within pathology slides to produce spatial maps representing the distribution of tumor, stromal, and lymphocyte cells.
• Extraction of Tumor Microenvironment Features: Derivation of image-based features from spatial maps that characterize interactions between cancer cells and the surrounding microenvironment.
• Prognostic Model Development: Use of extracted image features to build an image feature-based prognostic model validated across two independent cohorts and shown to be an independent prognostic factor after adjustment for age, gender, smoking status, and cancer stage.
• High Classification Accuracy: Reported overall classification accuracy of 92.9% in training datasets and 90.1% in independent testing datasets.

Scientific Applications:

• Research Facilitation: Enables analysis of the spatial organization of cells within the tumor microenvironment to study tumor progression and metastasis.
• Prognostic Analysis: Provides image-derived features and a prognostic model to inform prediction of patient outcomes and stratification for personalized treatment strategies.

Methodology:

Nuclei segmentation followed by convolutional neural network-based classification of tumor cells, stromal cells, and lymphocytes.