An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks
Skin cancer is a prevalent type of malignancy on a global scale, and the early and accurate diagnosis of this condition is of utmost importance for the survival of patients. The clinical assessment of cutaneous lesions is a crucial aspect of medical practice, although it encounters several obstacles...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-03-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/14/6/636 |
| _version_ | 1797241468626141184 |
|---|---|
| author | Kavita Behara Ernest Bhero John Terhile Agee |
| author_facet | Kavita Behara Ernest Bhero John Terhile Agee |
| author_sort | Kavita Behara |
| collection | DOAJ |
| description | Skin cancer is a prevalent type of malignancy on a global scale, and the early and accurate diagnosis of this condition is of utmost importance for the survival of patients. The clinical assessment of cutaneous lesions is a crucial aspect of medical practice, although it encounters several obstacles, such as prolonged waiting time and misinterpretation. The intricate nature of skin lesions, coupled with variations in appearance and texture, presents substantial barriers to accurate classification. As such, skilled clinicians often struggle to differentiate benign moles from early malignant tumors in skin images. Although deep learning-based approaches such as convolution neural networks have made significant improvements, their stability and generalization continue to experience difficulties, and their performance in accurately delineating lesion borders, capturing refined spatial connections among features, and using contextual information for classification is suboptimal. To address these limitations, we propose a novel approach for skin lesion classification that combines snake models of active contour (AC) segmentation, ResNet50 for feature extraction, and a capsule network with a fusion of lightweight attention mechanisms to attain the different feature channels and spatial regions within feature maps, enhance the feature discrimination, and improve accuracy. We employed the stochastic gradient descent (SGD) optimization algorithm to optimize the model’s parameters. The proposed model is implemented on publicly available datasets, namely, HAM10000 and ISIC 2020. The experimental results showed that the proposed model achieved an accuracy of 98% and AUC-ROC of 97.3%, showcasing substantial potential in terms of effective model generalization compared to existing state-of-the-art (SOTA) approaches. These results highlight the potential for our approach to reshape automated dermatological diagnosis and provide a helpful tool for medical practitioners. |
| first_indexed | 2024-04-24T18:23:48Z |
| format | Article |
| id | doaj.art-b15844f811e04cf1a3b8beb90232bfe8 |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-04-24T18:23:48Z |
| publishDate | 2024-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-b15844f811e04cf1a3b8beb90232bfe82024-03-27T13:33:23ZengMDPI AGDiagnostics2075-44182024-03-0114663610.3390/diagnostics14060636An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule NetworksKavita Behara0Ernest Bhero1John Terhile Agee2Department of Electrical Engineering, Mangosuthu University of Technology, Durban 4031, South AfricaDiscipline of Electrical, Electronic and Computer Engineering, University of KwaZulu-Natal, Durban 4041, South AfricaDiscipline of Electrical, Electronic and Computer Engineering, University of KwaZulu-Natal, Durban 4041, South AfricaSkin cancer is a prevalent type of malignancy on a global scale, and the early and accurate diagnosis of this condition is of utmost importance for the survival of patients. The clinical assessment of cutaneous lesions is a crucial aspect of medical practice, although it encounters several obstacles, such as prolonged waiting time and misinterpretation. The intricate nature of skin lesions, coupled with variations in appearance and texture, presents substantial barriers to accurate classification. As such, skilled clinicians often struggle to differentiate benign moles from early malignant tumors in skin images. Although deep learning-based approaches such as convolution neural networks have made significant improvements, their stability and generalization continue to experience difficulties, and their performance in accurately delineating lesion borders, capturing refined spatial connections among features, and using contextual information for classification is suboptimal. To address these limitations, we propose a novel approach for skin lesion classification that combines snake models of active contour (AC) segmentation, ResNet50 for feature extraction, and a capsule network with a fusion of lightweight attention mechanisms to attain the different feature channels and spatial regions within feature maps, enhance the feature discrimination, and improve accuracy. We employed the stochastic gradient descent (SGD) optimization algorithm to optimize the model’s parameters. The proposed model is implemented on publicly available datasets, namely, HAM10000 and ISIC 2020. The experimental results showed that the proposed model achieved an accuracy of 98% and AUC-ROC of 97.3%, showcasing substantial potential in terms of effective model generalization compared to existing state-of-the-art (SOTA) approaches. These results highlight the potential for our approach to reshape automated dermatological diagnosis and provide a helpful tool for medical practitioners.https://www.mdpi.com/2075-4418/14/6/636attention mechanismcapsule networkclassificationdynamic routingsegmentationskin cancer |
| spellingShingle | Kavita Behara Ernest Bhero John Terhile Agee An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks Diagnostics attention mechanism capsule network classification dynamic routing segmentation skin cancer |
| title | An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks |
| title_full | An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks |
| title_fullStr | An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks |
| title_full_unstemmed | An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks |
| title_short | An Improved Skin Lesion Classification Using a Hybrid Approach with Active Contour Snake Model and Lightweight Attention-Guided Capsule Networks |
| title_sort | improved skin lesion classification using a hybrid approach with active contour snake model and lightweight attention guided capsule networks |
| topic | attention mechanism capsule network classification dynamic routing segmentation skin cancer |
| url | https://www.mdpi.com/2075-4418/14/6/636 |
| work_keys_str_mv | AT kavitabehara animprovedskinlesionclassificationusingahybridapproachwithactivecontoursnakemodelandlightweightattentionguidedcapsulenetworks AT ernestbhero animprovedskinlesionclassificationusingahybridapproachwithactivecontoursnakemodelandlightweightattentionguidedcapsulenetworks AT johnterhileagee animprovedskinlesionclassificationusingahybridapproachwithactivecontoursnakemodelandlightweightattentionguidedcapsulenetworks AT kavitabehara improvedskinlesionclassificationusingahybridapproachwithactivecontoursnakemodelandlightweightattentionguidedcapsulenetworks AT ernestbhero improvedskinlesionclassificationusingahybridapproachwithactivecontoursnakemodelandlightweightattentionguidedcapsulenetworks AT johnterhileagee improvedskinlesionclassificationusingahybridapproachwithactivecontoursnakemodelandlightweightattentionguidedcapsulenetworks |