Data Model and Use Case Diagram

for

Automated Skin Care Detection

and Diagnosis Management
Version 2.0

Prepared by

TeamNumber: 7
VALLETI MAHI VIGNESH CB.EN.U4AIE22056
THATUKURU LAKSHMAN CB.EN.U4AIE22067
GOLI SURYA TEJA CB.EN.U4AIE22069
Y RAMA KRISHNA PRASAD CB.EN.U4AIE22070

Course: 22AIE311 SOFTWARE ENGINEERING

Date: 30/01/2025

1
 Table of Contents

1. Domain Objects 3
2. User Models 3
3. Use Case Diagram 5
Appendix A - Activity Log 5

Revisions
Version Primary Author(s) Description of Version Date Completed
V 2.0 MAHI VIGNESH Added relationships between entities, including 30/01/205
patient-image upload and dermatologist-image
LAKSHMAN review, with many-to-many associations
SURYA TEJA

KRISHNA PRASAD

2
1. Domain Objects

1. Skin Lesion
Represents a skin lesion image and its associated metadata.
Attributes
image_id: Unique identifier for the image (e.g., "ISIC_123456").
image_data: The actual image data (pixel values).
lesion_type: Type of skin lesion (e.g., melanoma, nevus, basal cell carcinoma).
location: Body location where the lesion was found (e.g., "back", "arm").
age: Age of the patient.
sex: Gender of the patient.

2. Diagnosis
Represents a diagnosis made by the model or a dermatologist.
Attributes
diagnosis_id: Unique identifier for the diagnosis.
skin_lesion: The SkinLesion object being diagnosed.
predicted_class: Predicted class of the skin lesion (e.g., "melanoma").
confidence_score: Confidence score of the prediction.
timestamp: Timestamp of the diagnosis.

3. SkinLesionDataset
Represents the collection of skin lesion images and metadata.
Attributes
dataset_id: Unique identifier for the dataset.
skin_lesions: List of SkinLesion objects in the dataset.
metadata: Additional metadata about the dataset (e.g., source, creation date).

2. User Models

1. Patient
o Attributes:
 PatientID (Unique identifier for the patient)
 Name (Name of the patient)
 Age (Age of the patient)
 Gender (Gender of the patient)
 MedicalHistory (Past medical history of the patient)
 SkinImages (Images of skin conditions uploaded by the patient)
o Functionalities:
 Upload skin images for analysis.

3
 View diagnosis results and recommendations.
Access historical records of previous diagnoses.
2. Dermatologist
o Attributes:
 DermatologistID (Unique identifier for the dermatologist)
 Name (Name of the dermatologist)
 Specialization (Area of specialization, e.g., pediatric dermatology, cosmetic
dermatology)
 LicenseNumber (Medical license number)
 YearsOfExperience (Number of years of experience in dermatology)
o Functionalities:
 Review and analyze skin images uploaded by patients.
 Provide diagnosis and treatment recommendations.
 Access patient medical history for better diagnosis.
3. Admin
o Attributes:
 AdminID (Unique identifier for the admin)
 Name (Name of the admin)
 Role (Role within the system, e.g., system administrator, data manager)
 AccessLevel (Level of access to the system, e.g., full access, limited
access)
o Functionalities:
 Manage user accounts (create, update, delete).
 Monitor system performance and security.
 Generate reports on system usage and diagnostics.
Many-to-Many relationships

1. Patient - Uploads - SkinImages

o Description: A patient can upload multiple skin images for analysis, but each skin
image is associated with only one patient.
o Relationship:
 One Patient can have many SkinImages.
 Each Skin Image belongs to only one Patient.
2. Dermatologist - Reviews - SkinImages
o Description: A dermatologist can review multiple skin images, but each skin image
is reviewed by only one dermatologist.
o Relationship:
 One Dermatologist can review many SkinImages.
 Each Skin Image is reviewed by only one Dermatologist.

4
 3. Use Case Diagram

Appendix A - Activity Log

1. Meeting Date: January 5, 2025
Duration: 2 hours
Agenda: Introduction to the skin disease detection project, task allocation, and initial
research on disease detection methods.

Contributions:
 Mahi Vignesh (CB.EN.U4AIE22056-Team Lead): Led the project discussion,
assigned roles to team members, and set up the project milestones. Suggested using
image classification and deep learning techniques for skin disease detection.

5
  Lakshman (CB.EN.U4AIE22067): Researched image preprocessing techniques for
skin disease images (e.g., noise reduction, image enhancement). Suggested using
CNNs (Convolutional Neural Networks) for the detection model.
 Surya Teja (CB.EN.U4AIE22069): Focused on researching medical datasets for skin
diseases and skin images. Proposed using the ISIC dataset for training the model.
 Krishna Prasad (CB.EN.U4AIE22070): Investigated deep learning models suitable
for skin disease classification and researched transfer learning to enhance model
accuracy.

2. Meeting Date: January 10, 2025

Duration: 1.5 hours
Agenda: Dataset finalization, model selection, and data preprocessing discussion.
Contributions:
 Mahi Vignesh (CB.EN.U4AIE22056-Team Lead): Reviewed the progress of team
members, ensured alignment with the timeline, and discussed methods for model
evaluation (e.g., accuracy, precision, recall).
 Lakshman (CB.EN.U4AIE22067): Worked on data augmentation techniques (rotation,
flipping, resizing) to increase dataset size for training. Preprocessed the dataset by
resizing and normalizing the images.
 Surya Teja (CB.EN.U4AIE22069): Finalized the ISIC skin disease dataset and
downloaded the training and validation data. Investigated the types of skin diseases in
the dataset.
 Krishna Prasad (CB.EN.U4AIE22070): Implemented the data pipeline and handled the
initial training of a CNN model. Experimented with different architectures to check for
performance improvements.