How to Use DeepMammo

A step-by-step guide to making predictions

1

Prepare your image

Use a standard mammogram in PNG or JPG format. The image should be a full mammogram view (CC or MLO). Maximum file size is 20 MB. DICOM files should be exported to PNG before uploading.

2

Upload on the Dashboard

Navigate to the Dashboard page. Click the upload area or drag and drop your image directly. The filename will appear once selected.

3

Run the analysis

Click Run Analysis. The system will preprocess your image (CLAHE enhancement, breast tissue isolation), run it through the EfficientNet-B4 model, and generate segmentation and classification outputs. This takes 5 โ€“ 15 seconds.

4

Review the results

You will see: Abnormality type (mass or calcification), Pathology (benign or malignant), a segmentation mask showing the detected region, a saliency map highlighting model attention, and a structured clinical report.

5

Download the PDF report

Click Download PDF to save a formatted report containing all predictions, visual outputs, and the AI-generated clinical narrative. Share with your clinical team for review.

Important: DeepMammo is an AI-assisted research tool. All results must be reviewed and verified by a qualified radiologist before any clinical decision is made. This tool does not replace professional medical diagnosis.