Scribd
Upload
4 views2 pages

Enhanced Alexnet Malignancy Classification in Lung and Colon Histopathology Via Csip-Ehe Hyperparameter Optimization

This paper presents a novel hyperparameter optimization method, CSIP-EHE, to enhance the performance of AlexNet in classifying lung and colon malignancies from histopathological images. The optimized model demonstrates improved accuracy, F1-score, and AUC compared to baseline methods, while also providing faster convergence and reduced overfitting. Additionally, it generates malignancy probability heatmaps to assist pathologists in identifying high-risk areas, highlighting the importance of intelligent hyperparameter optimization in cancer diagnosis.

Uploaded by

kokilarena
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views2 pages

Enhanced Alexnet Malignancy Classification in Lung and Colon Histopathology Via Csip-Ehe Hyperparameter Optimization

This paper presents a novel hyperparameter optimization method, CSIP-EHE, to enhance the performance of AlexNet in classifying lung and colon malignancies from histopathological images. The optimized model demonstrates improved accuracy, F1-score, and AUC compared to baseline methods, while also providing faster convergence and reduced overfitting. Additionally, it generates malignancy probability heatmaps to assist pathologists in identifying high-risk areas, highlighting the importance of intelligent hyperparameter optimization in cancer diagnosis.

Uploaded by

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

Enhanced AlexNet Malignancy Classification in Lung and Colon Histopathology via

CSIP-EHE Hyperparameter Optimization

Abstract

Early and precise identification of malignancies in histopathological images is crucial for


enhancing cancer diagnosis and patient outcomes. Conventional convolutional neural
networks (CNNs), illustrated by AlexNet, have demonstrated encouraging efficacy in the
classification of tumor tissues; nevertheless, their performance is significantly influenced by
the precise selection of hyperparameters. This paper presents an innovative hyperparameter
optimization methodology employing a Cooperative Swarm-Intensified Particle with Elite
Hybrid Exploration (CSIP-EHE) algorithm to augment AlexNet for the automated
identification of lung and colon cancer from histopathological images. The CSIP-EHE system
combines a global search that looks at the whole population with a local search that looks at
the best individuals to quickly explore hyperparameter spaces that include learning rate, batch
size, momentum, dropout, and data augmentation parameters. We test the improved model on
publically accessible lung and colon histopathology datasets that have a balanced class
distribution and stain normalization to deal with differences between domains. Experimental
findings show that the classification accuracy, F1-score, and AUC have all improved a lot
compared to the baseline AlexNet and standard tuning methods like grid and random search.
The optimized model also converges faster and is more resistant to overfitting by using
promising hyperparameter solutions in an adaptive way. Moreover, patch-based slide
inference facilitates the creation of malignancy probability heatmaps to assist pathologists in
pinpointing high-risk tumor areas. The results confirm that CSIP-EHE is a useful meta-
optimization method for analyzing medical images. It improves network generalization and
cuts down on the need for human tuning. This study emphasizes the significance of
intelligent hyperparameter optimization in deep learning-driven cancer diagnoses and
develops an improved AlexNet pipeline suited for implementation as a dependable clinical
decision-support instrument.
Keywords: CSIP-EHE, Hyperparameter Optimization, AlexNet, Deep Learning, Lung
Cancer, Colon Cancer, Histopathology Image Classification, Medical Image Analysis,
Cancer Detection, Swarm Intelligence
Objectives:
 To improve AlexNet's ability to classify malignancies in lung and colon
histopathology images by employing the best training architecture.
 To use a CSIP-EHE-based hyperparameter optimization method to rapidly and
efficiently adjust important learning parameters.
 To enhance diagnostic precision and resilience using equitable data preprocessing,
stain normalization, and enhancement techniques.
 To evaluate the optimized model on benchmark medical datasets using a wide range
of metrics, such as accuracy, F1-score, and AUC.
 To provide comprehensible malignancy localization through patch-based heatmaps to
facilitate clinical decision-making.

You might also like