synthcxr/constants.py

"""Shared constants for SynthCXR: disease labels, condition maps, severity modifiers."""

from __future__ import annotations

# CheXpert label column names -> natural-language descriptions used in prompts.
# Used by both dataset preparation and inference scripts.
LABEL_TEXT: dict[str, str] = {
    "Enlarged Cardiomediastinum": "enlarged cardiomediastinum",
    "Cardiomegaly": "cardiomegaly",
    "Lung Opacity": "diffuse lung opacity",
    "Lung Lesion": "discrete lung lesion",
    "Edema": "pulmonary edema",
    "Consolidation": "parenchymal consolidation",
    "Pneumonia": "findings compatible with pneumonia",
    "Atelectasis": "atelectasis",
    "Pneumothorax": "pneumothorax",
    "Pleural Effusion": "pleural effusion",
    "Pleural Other": "other pleural abnormality",
    "Fracture": "possible fracture",
    "Support Devices": "support devices in place",
}

# Snake_case keys for config files -> natural-language descriptions.
KNOWN_CONDITIONS: dict[str, str] = {
    "enlarged_cardiomediastinum": "enlarged cardiomediastinum",
    "cardiomegaly": "cardiomegaly",
    "lung_opacity": "diffuse lung opacity",
    "lung_lesion": "discrete lung lesion",
    "edema": "pulmonary edema",
    "consolidation": "parenchymal consolidation",
    "pneumonia": "findings compatible with pneumonia",
    "atelectasis": "atelectasis",
    "pneumothorax": "pneumothorax",
    "pleural_effusion": "pleural effusion",
    "pleural_other": "other pleural abnormality",
    "fracture": "possible fracture",
    "support_devices": "support devices in place",
}

SEVERITY_MODIFIERS: dict[str, str] = {
    "mild": "mild",
    "moderate": "moderate",
    "severe": "severe",
    "small": "small",
    "large": "large",
    "very_small": "very small",
    "very_large": "very large",
    "minimal": "minimal",
    "significant": "significant",
}

DEFAULT_MODEL_ID = "gradientguild/SynthCXR-Qwen-Weights"
TEXT_ENCODER_MODEL_ID = "gradientguild/SynthCXR-Qwen-Weights"
PROCESSOR_MODEL_ID = "gradientguild/SynthCXR-Qwen-Weights"