Scribd
Upload
4 views1 page

Hybrid Model for Cancer Detection

This study presents a hybrid deep learning model for cancer detection that classifies cervical cell images into five categories using the SIPaKMeD dataset. The model combines DenseNet201, Vision Transformer, and ConvNeXt-Tiny, achieving a test accuracy of 97.37%, which outperforms individual models. The findings highlight the effectiveness of hybrid models in medical image classification and suggest future research directions for enhancing model interpretability and generalizability.

Uploaded by

sarveshkun05
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
4 views1 page

Hybrid Model for Cancer Detection

This study presents a hybrid deep learning model for cancer detection that classifies cervical cell images into five categories using the SIPaKMeD dataset. The model combines DenseNet201, Vision Transformer, and ConvNeXt-Tiny, achieving a test accuracy of 97.37%, which outperforms individual models. The findings highlight the effectiveness of hybrid models in medical image classification and suggest future research directions for enhancing model interpretability and generalizability.

Uploaded by

sarveshkun05
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Conclusion:

Here in this study, we create a hybrid deep learning model for detecting cancer by classifying
cervical cell images into a total of five categories using SIPaKMeD dataset. The hybrid model
combines DenseNet201, Vision Transformer (ViT), and ConvNeXt-Tiny through a weighted soft
voting mechanism.

So rather than concentrating on optimizing the individual models, the primary objective was to
harness the pros of different architectures - both convolutional and transformer based to enhance the
classification accuracy. Our hybrid approach archieved a notable test accuracy of 97.37%,
surpassing the individual model's precision, recall and F1-score. The results show that combining
multiple architectures that complement each other can result more accurate predictions, particularly
in complex tasks such as medical image classification.

The findings tells us the potential of hybrid deep learning models in aiding cancer detection by
automating and enhancing the diagnostic process. Future research could focus on refining the
ensemble strategy, improving model interpretability for clinical application, and validating the
approach on other medical imaging datasets to evaluate generalizability and reliability.

You might also like