ageer/medgemma-1.5-mammogram-finetuned

MedGemma 1.5 Fine-Tuned for Mammogram Analysis

This model is a fine-tuned version of google/medgemma-1.5-4b-it specialized for mammogram analysis and breast imaging interpretation.

Model Description

• Base Model: google/medgemma-1.5-4b-it (4B parameter multimodal model)
• Fine-tuning Method: QLoRA (4-bit quantization with LoRA)
• Training Data: CDD-CESM (Contrast-Enhanced Spectral Mammography) dataset
• Specialization: Breast imaging, mammogram interpretation, radiology reports

Capabilities

This fine-tuned model can:

• 🔬 Analyze mammogram images (CC and MLO views)
• 📝 Generate detailed radiological reports
• 🎯 Detect and characterize breast lesions
• 📊 Assess breast density (BI-RADS A-D)
• 💡 Provide clinical decision support recommendations

Usage

from transformers import AutoProcessor, AutoModelForImageTextToText
from PIL import Image
import torch

# Load model
model_id = "ageer/medgemma-1.5-mammogram-finetuned"
model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)
processor = AutoProcessor.from_pretrained(model_id)

# Load and analyze mammogram
image = Image.open("mammogram.jpg")

messages = [
    {
        "role": "system",
        "content": [{
            "type": "text",
            "text": "You are an expert radiologist specializing in breast imaging."
        }]
    },
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Analyze this mammogram and provide a detailed assessment."},
            {"type": "image", "image": image}
        ]
    }
]

inputs = processor.apply_chat_template(
    messages,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt"
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=500)
response = processor.decode(outputs[0], skip_special_tokens=True)
print(response)

Training Details

Parameter	Value
Base Model	google/medgemma-1.5-4b-it
Fine-tuning Method	QLoRA
LoRA Rank	16
LoRA Alpha	32
Learning Rate	2e-4
Epochs	3
Training Data	CDD-CESM mammogram dataset (326 patients)

⚠️ Important Disclaimer

This model is for research and educational purposes only.

• NOT approved for clinical use or medical diagnosis
• Should NOT replace professional medical advice
• Requires validation before any clinical application
• Model outputs should be reviewed by qualified radiologists

License

This model inherits the Health AI Developer Foundations License from MedGemma.