Scribd
Upload
6 views

Progress Presentation

The document presents a progress report on CardioAI, an AI surgical assistance project focusing on cardiac surgery. It details the requirement analysis, system architecture, data collection, model selection, and user interface design, highlighting the use of various AI models for text extraction from medical reports. The project aims to enhance surgical assistance through functionalities like voice commands and real-time data integration.

Uploaded by

javeed janan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
6 views

Progress Presentation

The document presents a progress report on CardioAI, an AI surgical assistance project focusing on cardiac surgery. It details the requirement analysis, system architecture, data collection, model selection, and user interface design, highlighting the use of various AI models for text extraction from medical reports. The project aims to enhance surgical assistance through functionalities like voice commands and real-time data integration.

Uploaded by

javeed janan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 15

CardioAI

AI Surgical Assistance
(Progress Presentation)

M.BILAL IJAZ -12577


JAVEED ULLAH -12575

Supervisor
MUhammad Wasim 1
2
Requirement Analysis and Research

Identifying Stakeholders:

Stakeholders: Identify key stakeholders such as doctors, surgeons, patients, and healthcare
administrators.

Defining Project Scope

● Functional Requirements: Determine what functionalities the system must have (e.g. user
authentication, voice assistance,pre surgery assistance, intra surgery and post surgery assistance).
● Non-Functional Requirements: Identify performance, scalability, security, and usability needs.

Data Requirements

● Data Sources: Identify necessary data sources, such as patient records, test reports, medical
history, and surgery details.

Research on Existing Solutions

Literature Review: Study existing research papers, articles, and case studies related to AI in cardiac
surgery.

3
System Architecture Design

Key Components of System Architecture Design for CardioAI:

Data Sources and Input Layer

● Patient Data: This includes medical records, test reports, and surgery details.
● Image Data: Medical imaging such as CT scans, MRI scans, and X-rays.

Data Ingestion Layer

● Data Acquisition: Use APIs, ETL processes, or direct connections to gather data from multiple sources.

Application Layer

● User Interface (UI/UX): Design a user-friendly mobile app using Flutter that allows doctors to interact with
the system seamlessly.
● Voice Assistance: Integrate voice commands and responses to assist surgeons during surgery without the
need for manual interaction.

Integration Layer

● APIs: Develop APIs to connect the mobile app with the backend services, enabling real-time data exchange
and communication.

4
Data Collection and Preprocessing:

Patient Medical Records: Includes patient demographics, medical history, diagnoses, alergies
and medication.

Medical Imaging: Data from X-rays, CT scans, MRI scans, and other relevant imaging.

Surgery Reports: Detailed notes on past surgeries, including outcomes, procedures followed,
and any complications.

Laboratory Test Results: Blood tests, ECG results, and other diagnostic reports.

Data Transformation:

Image Preprocessing: Resize, crop, and enhance medical images to standardize their formats
and improve model performance.

5
Model Selection and Training:

1. Initial Model Selection: OCR Model

● Objective: The first step in the model selection process for CardioAI was to extract text from medical test report
images.
● Model Used: An OCR (Optical Character Recognition) model.
● Outcome: While the OCR model successfully extracted text from the images, the results were not optimal. It
struggled with medical terminology, varying fonts, and poor image quality in some reports, leading to inaccuracies
and incomplete data extraction.

2. Gemini Model Implementation

● Objective: To improve data extraction accuracy while ensuring data privacy.


● Model Used: The Gemini model, known for its privacy-focused features.
● Outcome: The Gemini model performed well in terms of data privacy, as it effectively masked personal data in the
reports. However, this feature also became a limitation, as it sometimes hid essential details needed for accurate
medical data analysis. As a result, critical information required for downstream processes was missing, impacting
the overall system performance.

6
3. Optimal Solution: OpenAI Model

● Objective: To achieve the best possible accuracy in text extraction from medical test
report images without compromising essential details.
● Model Used: OpenAI’s advanced language model.
● Outcome: The OpenAI model provided the most accurate and comprehensive text
extraction results. It handled the complexity of medical terminologies, various formats, and
poor-quality images more effectively than the previous models. The model's ability to
maintain context and extract relevant information without masking essential data made it
the optimal choice for CardioAI.

7
8
App User Interface:
SplashScreen Onboarding Screens

9
App User Interface:
Login Screen Create Account Screen

10
App User Interface:
Home Screen

11
App User Interface:

Create Pre Surgery Report Screen

12
App User Interface:

Pre Surgery, Intra Surgery and Post Surgery Screens

13
Screen Recording

14
Thank You

15

You might also like