Melanoma Detection Assignment

Assignment to build a CNN based model which can accurately detect melanoma.

Table of Contents :

• Problem Statement
• Objectives
• Approach
• Technologies/Libraries Used
• Conclusions
• Acknowledgements
• Glossary
• Author

Problem Statement

To build a CNN based model which can accurately detect melanoma. Melanoma is a type of cancer that can be deadly if not detected early. It accounts for 75% of skin cancer deaths. A solution that can evaluate images and alert dermatologists about the presence of melanoma has the potential to reduce a lot of manual effort needed in diagnosis.

You can download the dataset here

The dataset consists of 2357 images of malignant and benign oncological diseases, which were formed from the International Skin Imaging Collaboration (ISIC). All images were sorted according to the classification taken with ISIC, and all subsets were divided into the same number of images, with the exception of melanomas and moles, whose images are slightly dominant.

The data set contains the following diseases:

• Actinic keratosis
• Basal cell carcinoma
• Dermatofibroma
• Melanoma
• Nevus
• Pigmented benign keratosis
• Seborrheic keratosis
• Squamous cell carcinoma
• Vascular lesion

NOTE:

• Should not use any pre-trained model using Transfer learning. All the model building processes should be based on a custom model.
• Some of the elements introduced in the assignment are new, but proper steps have been taken to ensure smooth learning.
• The model training may take time to train as you will be working with large epochs. It is advised to use GPU runtime in Google Colab.

Objectives

The primary objective of this assignment is to develop a custom convolutional neural network (CNN) model capable of accurately classifying images of various skin lesions, specifically focusing on melanoma detection. The model aims to enhance early diagnosis and provide a reliable tool for dermatologists by automating the analysis of skin images, thereby potentially reducing the manual effort involved in identifying melanoma and other oncological diseases. Additionally, the assignment seeks to explore different model architectures and optimization techniques to improve classification accuracy and robustness.

Approach

• Step 1: Import Necessary Libraries
• Step 2: Load the Data and Understanding the Data
• Step 3: Data Preparation Steps - Train-Validation Split and Test Set
• Step 4: Building the Initial Model (Model 1)
• Step 5: Data Augmentation - Build Model after Applying Augmentation
• Step 6: Fix Class Imbalance and Build the Model
• Step 7: Fine-tune and Optimize the Model
• Step 8: Making Predictions Using the Models
• Step 9: Conclusion

Technologies/Libraries Used

• numpy : 1.26.4
• pandas : 2.2.2
• matplotlib : 3.7.1
• tensorflow : 2.17.0
• keras : 3.4.1
• augmentor : 0.2.12

Conclusions

Based on the model comparison and evaluation, the following insights can be drawn: -

• Model 9 (Ten Convolution Layers + Augmented Layers + Balanced Class + Dropouts + Batch Normalization + Controlled LR) achieved the highest overall performance, with a validation accuracy of 0.73 and a test accuracy of 0.50. The combination of more layers, dropout regularization, and controlled learning rates made this model the best at generalizing to unseen data.
• Models 4 and 5 had moderate performance, achieving validation accuracies of 0.70 and 0.65, respectively, with a test accuracy of 0.45. These models underline the importance of deeper architectures and more sophisticated regularization techniques for achieving better results
• Base Models (1, 2, 3) suffered from overfitting, especially Model 1, which had a high training accuracy (0.91) but very low test accuracy (0.33). This confirms that deeper architectures with data augmentation and dropout regularization are essential for better generalization.
• In summary, deeper convolutional models with controlled learning rates, batch normalization, and dropout regularization, like Model 9, performed the best. While more epochs can help in training accuracy (Model 10), they do not always lead to better test performance.
• Overall, applying data augmentation, class balancing, learning rate control, dropout optimization, and deeper architectures significantly enhanced the performance of the CNN models for melanoma detection, with Model 8 emerging as the most balanced in terms of validation and test performance. Further optimization could continue to improve this performance.

**Q :- Which class has the least number of samples? Answer - seborrheic keratosis has the least number of samples - 77 **Q :- Which classes dominate the data in terms proportionate number of samples? Answer - pigmented benign keratosis dominates the data in terms proportionate number of samples - 462

Acknowledgements

• The project reference course materieals from upGrads curriculm .
• The project references from presentation in upgrad live class given by Shivam Garg
• The project references insights and inferences from presentation in upgrad live class given by Sivaram A.

Glossary

• Data Augmentation
• Class Imbalance
• Train-Validation Split
• Test Set
• Convolutional Neural Network (CNN)
• Dropout
• Learning Rate (LR)
• Overfitting
• Early Stopping
• Cross-Entropy Loss
• Accuracy
• Batch Normalization
• Max Pooling
• Softmax
• Learning Rate Scheduler (ReduceLROnPlateau)

Author

• Arnab Bera