Lungs cancer dataset

About Dataset editEdit LungCanC2024 DatasetA Comprehensive Multi-Modal Dataset for Lung Cancer AnalysisOverviewLung cancer is one of the leading causes of cancer-related deaths worldwide, requiring precise diagnostic and predictive models for improved patient outcomes. The LungCanC2024 Dataset is a large-scale dataset containing 289,010 patient records with imaging, clinical, and genomic features, designed to support research in machine learning, deep learning, federated learning, and personalized medicine. The dataset includes detailed radiomic features extracted from imaging, patient demographics, smoking history, tumor staging, treatment records, and biomarker expression levels. Additionally, it contains multi-label classification targets, making it suitable for research in cancer detection, subtype classification, stage prediction, and survival analysis. Dataset HighlightsMulti-modal features: Includes radiological, clinical, and genomic data.Large-scale cohort: Comprises 289,010 patient records, enabling deep learning and statistical modeling.Multi-label learning: Supports cancer presence detection, subtype classification, staging, and survival analysis.Federated Learning Ready: Can be used in privacy-preserving distributed training approaches.Suitable for AI-based Precision Medicine: Helps in developing personalized cancer treatment models.Feature DescriptionsThis dataset consists of three main categories: Imaging Features (Radiomics)Feature Name Descriptionnodule_size_mm Size of detected lung nodules (measured in mm).nodule_texture Texture-based feature derived from radiological analysis.HU_mean Mean Hounsfield Unit (HU) value from CT scans.HU_std Standard deviation of HU values indicating nodule density variations.GLCM_contrast Gray Level Co-occurrence Matrix (GLCM) contrast, measuring texture heterogeneity.GLCM_correlation GLCM correlation metric assessing pattern consistency.PET_SUVmax Maximum Standardized Uptake Value (SUV) from PET scans, indicating metabolic activity.PET_SUVmean Mean SUV value across the tumor region. Clinical & Metadata FeaturesFeature Name Descriptionpatient_age Age of the patient (30-90 years).patient_gender Male (70%) / Female (30%).smoking_history Smoking status: Never, Former, or Current smoker.family_history Binary (1 = Family history of lung cancer, 0 = No family history).tumor_location Left Lung / Right Lung (40% vs. 60%).tumor_stage Stage I-IV classification, with imbalanced distribution.radiation_therapy Binary (1 = Received therapy, 0 = No therapy).chemotherapy_received Whether the patient received chemotherapy (Binary).immunotherapy_received Whether the patient received immunotherapy (Binary).targeted_therapy_received Whether the patient received targeted therapy (Binary). Genomic & Biomarker FeaturesFeature Name DescriptionEGFR_mutation_status Binary (1 = EGFR mutation detected, 0 = No mutation).KRAS_mutation_status Binary (1 = KRAS mutation detected, 0 = No mutation).ALK_fusion_status Binary (1 = ALK gene fusion present, 0 = No fusion).PD-L1_expression_level PD-L1 biomarker expression level (0-100%).tumor_mutational_burden Tumor Mutational Burden (TMB), an indicator of genomic instability. Multi-Label Target VariablesFeature Name Descriptioncancer_presence Binary (1 = Malignant tumor detected, 0 = No cancer).cancer_subtype Categorical: No Cancer, Adenocarcinoma, Squamous Cell, Small Cell Lung Cancer (SCLC), Other.cancer_stage Categorical: No Cancer, Stage I, Stage II, Stage III, Stage IV.survival_time_months Estimated survival duration (in months).