A Synopsis on

“Skin Cancer Detection using Dermoscopic Images”

Department of Computer Science & Engineering

Submitted to: Submitted by:

Ms. Khushbu Tamrakar Chanchal Ratnawat

Riya Yogi

Oriental University, Indore

Gate No.1, Sanwer Rd, opposite Revati Range, Jakhya, Indore, Madhya Pradesh 453555

1
 INDEX

S No. Topic Page No.

1. Team 3

4
2. Introduction

5
3. Aim & Objectives

6-7
4. Time Plan

8
5. Brief Description

9
6. Usefulness

10-11
7. Process Development Model Proposed

12
8. Advantages / Benefits of the Proposed System

13
9. Technology Used

14
10. References

15
11. Signature

2
 TEAM

i) Size:
2 Members

ii) Name with Enrollment Number:

1. Chanchal Ratnawat – ENR: OUI122BAM003
2. Riya Yogi – ENR: OUI122BCS053

iii) Mail-ID:
1. chanchalratnawat311@[Link]
2. riyayogi910@[Link]

iv) Phone Number:

1. +91-9109793315
2. +91-9981193174

3
 INTRODUCTION

Skin cancer is one of the most common types of cancer in the world. It happens when skin cells
start growing abnormally, usually because of too much exposure to sunlight. The harmful
ultraviolet (UV) rays from the sun damage the skin over time. If this damage is not treated
early, it can develop into cancer. There are different types of skin cancer, and some are more
dangerous than others. Among them, melanoma is the most serious type because it can spread
quickly to other parts of the body. The good news is that if skin cancer is found early, it can
often be treated successfully.

Doctors usually check for skin cancer by looking at unusual spots or moles on the skin. They
often use a special tool called a dermoscope, which helps them see the skin more clearly by
magnifying it. This process is called dermoscopy. While this method helps doctors examine
the skin in detail, it still depends on their experience and judgment. Sometimes, early signs of
cancer can look very similar to harmless skin spots, which makes diagnosis difficult. This can
lead to delays or mistakes.

With the growth of technology, especially Artificial Intelligence (AI), computers can now help
doctors in many medical tasks. AI systems can study large numbers of images and learn
patterns that humans might not easily notice. In this project, we use a type of AI called Deep
Learning to study dermoscopic images of skin. The system learns from many example images
and understands the difference between normal (benign) and cancerous (malignant) skin spots.
The main idea of this project is to create a smart system that can check a skin image and give
a prediction about whether it may be cancerous or not. The user uploads an image, and the
system analyzes it and shows the result. This tool is not meant to replace doctors but to support
them. It can help in early screening, especially in places where skin specialists are not easily
available. By using technology, we aim to make skin cancer detection faster, easier, and more
accurate.

4
 AIM & OBJECTIVES

Aim:
To design and develop an intelligent deep learning-based system capable of detecting and
classifying skin cancer from dermoscopic images with high accuracy and reliability.

Objectives:
1. To study various types of skin cancer and their visual characteristics.
2. To collect and preprocess dermoscopic image datasets for model training.
3. To implement a Convolutional Neural Network (CNN) architecture for image
classification.
4. To train and validate the model using appropriate performance metrics.
5. To develop a user-friendly web interface for uploading images and displaying results.
6. To evaluate system performance using accuracy, precision, recall, and F1-score.
7. To reduce false-negative cases, ensuring higher reliability in cancer detection.

5
 TIME PLAN

The project duration is approximately 12 weeks. The work will be divided into phases to ensure
systematic development and timely completion.

Week 1–2
Project Planning & Literature Review
 Understanding skin cancer types and dermoscopy
 Studying research papers and existing AI models
 Finalising tools, technologies, and dataset sources
Week 3–4
Dataset Collection & Preprocessing
 Collecting dermoscopic image dataset
 Cleaning and organizing data
 Image resizing, normalization, augmentation
 Splitting dataset into training and testing sets
Week 5–6
Model Design & Development
 Designing CNN architecture
 Implementing model using TensorFlow/PyTorch
 Initial model training
Week 7–8
Model Training & Optimization
 Fine-tuning hyperparameters
 Improving accuracy
 Evaluating using performance metrics
 Reducing overfitting
Week 9–10
Frontend & Backend Development
 Designing user interface
 Integrating trained model with backend
 Implementing image upload and result display
Week 11
System Integration & Testing
 Testing complete system
 Fixing bugs and improving performance
 Final validation
Week 12
Documentation & Final Submission Preparation
 Preparing project report
 Creating diagrams (architecture, flowchart)
 Preparing presentation (PPT)
 Final review and submission

6
Task / Activity W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12
Project Planning █ █
& Literature
Review
Dataset Collection █ █
Data █ █
Preprocessing
CNN Model █ █
Design
Model Training █ █ █ █
Model █ █
Optimization &
Evaluation
Frontend █ █
Development
Backend █ █
Integration
System Testing █
Documentation & █
PPT

Fig: Time Plan in Weeks

Fig: Gantt Chart – Skin Cancer Detection using Dermoscopic Images

7
 BRIEF DESCRIPTION

The project Skin Cancer Detection using Dermoscopic Images focuses on developing an
intelligent system that can automatically analyze skin lesion images and predict whether they
are benign (non-cancerous) or malignant (cancerous). The system uses Artificial Intelligence
and Deep Learning techniques, specifically Convolutional Neural Networks (CNN), to identify
patterns and features in dermoscopic images that indicate skin cancer.
The working of the system begins with data collection, where dermoscopic images are
obtained from publicly available medical datasets. These images are then passed through a
preprocessing stage, which includes resizing, normalization, and data augmentation
techniques such as rotation and flipping. This step improves image quality and helps the model
learn better.
After preprocessing, the images are used to train the CNN model. The CNN automatically
extracts important features like color variation, texture, shape, and irregular borders of skin
lesions. The model learns these patterns during training and later uses this knowledge to
classify new images.
Once the model is trained and tested, it is integrated into a user interface. The user can upload
a dermoscopic image through the interface, and the system processes the image in the backend.
The trained model then predicts the result and displays whether the lesion is likely benign or
malignant along with a confidence score.
The system is designed to be simple, fast, and easy to use. It does not replace medical
professionals but acts as a supportive screening tool that can help in early detection. By
combining medical imaging with AI technology, the project aims to improve diagnostic
efficiency, reduce manual effort, and make skin cancer screening more accessible, especially
in areas where specialist doctors are not easily available.

8
 USEFULNESS

The proposed system, Skin Cancer Detection using Dermoscopic Images, is highly useful in
improving the process of early skin cancer diagnosis. Skin cancer can become dangerous if not
detected at an early stage, but timely identification greatly increases the chances of successful
treatment. This project provides an AI-based solution that helps in analyzing skin lesion images
quickly and accurately, making early screening easier.
One of the major usefulness of this application is that it assists dermatologists in decision-
making. The system can analyze dermoscopic images and highlight the possibility of cancer,
which helps doctors perform faster and more informed examinations. It reduces the workload
of manual image analysis and saves valuable time in clinical settings.
The system is also beneficial in rural and remote areas where specialist doctors may not be
easily available. Patients can upload images and get a preliminary assessment, which helps
them decide whether they need immediate medical consultation. This improves accessibility to
healthcare and promotes early awareness about skin diseases.
Another important usefulness is large-scale screening. Hospitals and healthcare organizations
can use this system to examine a large number of patients efficiently. Since the model provides
consistent results, it reduces the chances of human error and improves diagnostic reliability.
The application should be used as a supportive tool rather than a replacement for medical
professionals. It is most useful during preliminary screening, follow-up monitoring, and
educational purposes. Additionally, the system can be extended in the future to detect multiple
skin diseases, making it more valuable in the healthcare domain.
Overall, this project helps in faster detection, better accessibility, reduced workload for doctors,
and improved patient awareness, contributing positively to modern healthcare using Artificial
Intelligence.

9
 PROCESS DEVELOPMENT MODEL PROPOSED

Model Used: Agile Software Development Model

Key Points
1. Iterative Development
o The project is developed in small phases (sprints).
o Each phase focuses on a specific task such as data collection, model training,
or interface development.
2. Flexibility
o Requirements in AI projects may change based on dataset quality and model
performance.
o Agile allows easy modifications without affecting the whole project.
3. Continuous Testing
o Testing is performed after every development stage.
o Helps identify errors early and improves system reliability.
4. Regular Feedback
o Faculty feedback after presentations can be incorporated quickly.
o Enhances accuracy, usability, and overall quality.
5. Incremental Improvement
o The CNN model can be retrained and optimized multiple times.
o Each iteration improves performance and reduces errors.
6. Better Task Management
o Work can be divided between team members efficiently.
o Progress can be tracked easily.
7. Risk Reduction
o Problems are identified early instead of at the final stage.
o Reduces chances of project failure.
8. Suitable for AI Projects
o Supports experimentation, tuning, and repeated training cycles.
o Ideal for machine learning and research-based development.
The Agile model is chosen because it supports flexibility, continuous improvement, regular
feedback, and efficient development, making it highly suitable for the skin cancer detection
project.

10
Fig: Process Development Model Proposed for Skin Cancer Detection
using Dermoscopic Images

11
 ADVANTAGES / BENEFITS OF THE PROPOSED SYSTEM

1. Early Detection
 Helps identify suspicious skin lesions at an early stage.
 Increases chances of successful treatment and recovery.
 Useful for detecting dangerous skin cancer types quickly.
2. Time Efficiency
 Analyzes dermoscopic images within seconds.
 Reduces time required for manual examination.
 Helps doctors handle more patients efficiently.
3. Improved Accuracy and Consistency
 Uses deep learning to detect patterns that may be missed by humans.
 Provides consistent results for similar cases.
 Reduces chances of diagnostic errors.
4. Accessibility
 Can be used in rural and remote areas with limited medical facilities.
 Allows preliminary screening before visiting a specialist.
 Improves healthcare reach.
5. Cost-Effective Solution
 Reduces repeated manual screening costs.
 Supports faster diagnosis, saving time and resources.
6. Scalability
 Can be extended to detect multiple skin diseases.
 Supports future integration with mobile and cloud platforms.
7. User-Friendly Interface
 Easy image upload and result display.
 Can be used by both doctors and general users.
8. Real-Time Prediction
 Provides quick results with confidence scores.
 Useful for screening large populations.
9. Reduced Workload for Doctors
 Assists dermatologists in decision-making.
 Helps focus on complex cases rather than routine screening.
Overall, the proposed system enhances early detection, improves diagnostic efficiency,
increases accessibility, and supports modern AI-based healthcare solutions.

12
 TECHNOLOGY USED

i) Front End
The front end of the system is responsible for user interaction and display of results. It
provides a simple and user-friendly interface where users can upload dermoscopic images
and view prediction outcomes.
Technologies used:
 HTML: To create the structure of web pages.
 CSS: To design and style the interface for better user experience.
 JavaScript: To handle user actions such as image upload and dynamic content
display.
 [Link] (optional): To build a responsive and interactive user interface.
Front End Features:
 Image upload option
 Display of prediction results
 Clean and easy navigation
 Responsive design for different devices

ii) Back End

The back end handles image processing, model execution, data handling, and communication
between the user interface and the AI model. It is responsible for analyzing uploaded images
and generating predictions.
Technologies used:
 Python: Main programming language for AI model development.
 Flask / Django: Framework used to connect the frontend with the machine learning
model.
 TensorFlow / PyTorch: Deep learning libraries used to build and train the CNN
model.
 OpenCV: Used for image preprocessing and manipulation.
 NumPy & Pandas: Used for data handling and numerical operations.
 Scikit-learn: Used for model evaluation and performance metrics.
Back End Features:
 Image preprocessing
 CNN model training and prediction
 Result generation with confidence score
 System integration and data handling
These technologies together enable the development of an efficient AI-based skin cancer
detection system.

13
 REFERENCES

[1] K. Nawaz et al., “Skin cancer detection using dermoscopic images with convolutional
neural network,” Scientific Reports, vol. 15, 2025. (Nature)

[2] M. A. A. Mahmud et al., “Explainable deep learning approaches for high precision early
melanoma detection using dermoscopic images,” Scientific Reports, vol. 15, 2025. (Nature)

[3] A. S. Al-Waisy et al., “A deep learning framework for automated early diagnosis and
classification of skin cancer lesions in dermoscopy images,” Scientific Reports, 2025. (Nature)

[4] M. Mateen et al., “Hybrid deep learning framework for melanoma diagnosis using
dermoscopic medical images,” Diagnostics, vol. 14, no. 19, 2024. (PubMed)

[5] S. Zhi et al., “A multiclassification model for skin diseases using dermatoscopy images
with Inception-v2,” Applied Sciences, vol. 14, no. 22, 2024. (MDPI)

[6] “Recent advances in diagnosis of skin lesions using dermoscopic images based on deep
learning,” IEEE Xplore, 2024. (IEEE Xplore)

14
Signature:
Student(s):
Chanchal Ratnawat
Riya Yogi
Date: __________________
Place: __________________

15