
Building BioVision AI: A Microvascular-Guided Wound Intelligence System with AWS
BioVision AI combines computer vision and peri-wound microvascular analysis to support wound healing assessment. This article explains the system architecture, technology stack, Amazon S3 integration using Boto3, and EC2 deployment configuration.
Introduction
Wound assessment is an important part of healthcare, but evaluating a wound from an image involves more than simply identifying the wound boundary. The surrounding tissue may also contain useful visual information about the condition of the wound.
This led us to develop BioVision AI, a computer-vision-based wound intelligence system that analyzes both the wound and the surrounding peri-wound tissue.
The central idea behind the project is:
Can the vascular condition around a wound provide useful information for understanding its healing condition?
BioVision AI explores this question by combining wound-image analysis with microvascular feature analysis and cloud integration.
What is BioVision AI?
BioVision AI is designed to process a wound image and generate structured visual information that can support wound assessment and healing-risk evaluation.
The overall workflow is:
Wound Image → Image Enhancement → Microvascular Enhancement → Vessel Analysis → Wound Segmentation → Peri-Wound Analysis → Healing Assessment → Risk Score → Report
The system has two major analysis components.
1. Wound Intelligence
The wound region is analyzed using computer-vision techniques to identify characteristics such as:
- Wound region and area
- Wound segmentation
- Tissue-related visual characteristics
- Healing-related features
- Risk-related indicators
2. Microvascular Intelligence
Instead of stopping at the wound boundary, BioVision AI examines the surrounding peri-wound region.
The system focuses on vascular characteristics including:
- Vessel density
- Vessel width
- Branching patterns
- Symmetry
- Local perfusion-related features
The goal is to provide additional image-based information that can complement conventional wound assessment. BioVision AI is a prototype and is not intended to replace clinical diagnosis or professional medical judgment.
System Architecture
The BioVision AI architecture connects the user interface, computer-vision pipeline and cloud-storage layer.
USER | v STREAMLIT APPLICATION | v BIOVISION AI | +-------+-------+ | | v v IMAGE PROCESSING WOUND ANALYSIS | | v v MICROVASCULAR SEGMENTATION ANALYSIS | +-------+-------+ | v HEALING ASSESSMENT | v RESULTS | v AMAZON S3The application interface is built using Streamlit, while the image-processing pipeline is implemented in Python.
Technology Stack
The project combines several technologies:
- Python – core development and processing
- OpenCV – image processing and enhancement
- Computer Vision – wound and vascular analysis
- YOLO/Segmentation – segmentation components
- Streamlit – interactive application interface
- Boto3 – AWS integration
- Amazon S3 – cloud storage
- Amazon EC2 – deployment configuration
This combination allowed us to move from an image-processing concept toward an interactive application with cloud integration.
Why Amazon S3?
One of the requirements of the project was to introduce a genuine cloud component rather than keeping the entire workflow local.
For this, we integrated Amazon S3 as the cloud-storage layer.
The application communicates with S3 using Boto3, the AWS SDK for Python.
The integration is separated into a dedicated module:
aws_s3.pyThis keeps the AWS-related functionality separate from the main application logic.
The simplified architecture is:
BioVision AI | v Boto3 | vAmazon S3 | vCloud StorageWhen the required AWS configuration is available, the application can upload wound-image data to an S3 bucket.
This approach allows the computer-vision pipeline and cloud-storage functionality to remain modular.
Amazon EC2
We also prepared the project for deployment using Amazon EC2.
The repository contains:
deploy_aws.shbiovision.serviceThe deployment script is intended to automate the application setup, while the service configuration allows the application to be managed as a system service on an EC2 environment.
This gave us practical experience with the additional considerations involved in taking a Python application toward cloud deployment.
The current live demonstration is hosted on Streamlit. The project repository contains the AWS S3 integration and EC2 deployment configuration.
AWS Integration in the Application
The AWS-related project structure is organized around the following components:
app.py | +---- aws_s3.py | | | +---- Boto3 | | | +---- Amazon S3 | +---- requirements.txt
deploy_aws.sh | +---- EC2 deployment
biovision.service | +---- Application service configurationThis structure helped us keep the cloud functionality independent from the main computer-vision workflow.
Building the Application
The user interacts with BioVision AI through a Streamlit interface.
The workflow begins when an image is uploaded.
The image then passes through the processing pipeline:
Input Image
↓
Image Enhancement
↓
Microvascular Enhancement
↓
Vessel Analysis
↓
Wound Segmentation
↓
Peri-Wound Analysis
↓
Healing Assessment
↓
Risk Evaluation
↓
Results
The interface presents the processed information in a way that makes the computer-vision pipeline easier to understand.
What We Learned
Developing BioVision AI taught us that creating an AI project involves much more than developing an algorithm.
We had to consider:
- Image preprocessing
- Computer-vision pipelines
- Segmentation
- Feature extraction
- Application development
- Cloud storage
- AWS authentication and configuration
- Deployment preparation
- Project architecture
- User interaction
Integrating Boto3 with Amazon S3 was particularly useful because it demonstrated how a Python application can communicate with a cloud service instead of keeping all data handling local.
Preparing the project for Amazon EC2 also helped us understand the additional steps required when moving an application toward cloud deployment.
Challenges
One of the main challenges was connecting different parts of the system into one workflow.
The project combines image processing, computer vision, an interactive interface and cloud functionality. Each component has different requirements, so keeping the architecture modular was important.
AWS integration also introduced concepts such as:
- IAM permissions
- S3 buckets
- AWS configuration
- Boto3
- EC2 setup
- Deployment scripts
- System services
Working through these components gave us practical experience with cloud infrastructure alongside AI development.
Future Improvements
BioVision AI can be extended in several directions.
Future work could include:
- Larger and more diverse wound-image datasets
- Improved wound segmentation
- More robust vessel detection
- Better microvascular feature extraction
- Stronger validation of healing-related predictions
- Clinical evaluation with appropriate datasets and expert input
- Expanded AWS-based deployment
- Scalable cloud processing
These improvements would help evaluate the approach more rigorously.
Conclusion
BioVision AI combines wound intelligence, microvascular analysis, computer vision and cloud integration into a single application.
The project began with a simple research question about whether the tissue surrounding a wound could provide useful information for healing assessment.
Through the project, we developed a working computer-vision application and integrated Amazon S3 through Boto3, while also preparing Amazon EC2 deployment configuration.
For us, the most important learning was understanding how an AI idea can be transformed into a complete application by combining AI, software engineering and cloud technologies.
BioVision AI — AI + Computer Vision + Cloud for Wound Intelligence.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article