FDA_Submission_Template.md

FDA Submission

Your Name:

Name of your Device:

Algorithm Description

1. General Information

Intended Use Statement:

Indications for Use:

Device Limitations:

Clinical Impact of Performance:

2. Algorithm Design and Function

<< Insert Algorithm Flowchart >>

DICOM Checking Steps:

Preprocessing Steps:

CNN Architecture:

3. Algorithm Training

Parameters:

• Types of augmentation used during training
• Batch size
• Optimizer learning rate
• Layers of pre-existing architecture that were frozen
• Layers of pre-existing architecture that were fine-tuned
• Layers added to pre-existing architecture

<< Insert algorithm training performance visualization >>

<< Insert P-R curve >>

Final Threshold and Explanation:

4. Databases

(For the below, include visualizations as they are useful and relevant)

Description of Training Dataset:

Description of Validation Dataset:

5. Ground Truth

6. FDA Validation Plan

Patient Population Description for FDA Validation Dataset:

Ground Truth Acquisition Methodology:

Algorithm Performance Standard: