Abstract
Early and accurate diagnosis of melanoma, a potentially life-threatening skin cancer, is crucial for improving patient outcomes. In this study, we propose a novel approach for melanoma detection, termed Panoptic Region Slicing Segmentation (PRS2) using an optimized convolution neural network (PRS2-OCNN) based on AlexNet. The proposed system integrates several advanced methods to enhance the accuracy and efficiency of melanoma identification. The initial step involves preprocessing the dermoscopic image using a 2D Fusion Filter, which enhances the image quality and prepares it for subsequent analysis. Next, the Panoptic Region Slicing Segmentation (PRS2) method is applied to emphasize the boundary regions, allowing for precise localization of melanoma-affected areas. To assess color variations within the segmented regions, we employ Threshold Histogram Evaluation (THE), which effectively characterizes melanoma-specific color patterns. The features extracted using Spread Spectral Menzies’s Feature Selection (SSMFS), reducing the dimensionality and improving the efficiency of the subsequent analysis. The core of our proposed approach lies in the optimized convolution neural network, derived from the influential AlexNet architecture. By fine-tuning the AlexNet-OCNN on the reduced feature set, we maximize its ability to accurately classify melanoma lesions based on their risk level. The PRS2-OCNN identifies melanoma classes according to their risk severity, aiding dermatologists in making informed decisions for timely and appropriate treatment.Experimental evaluations were conducted on a diverse and extensive dataset of dermoscopic images. The proposed system demonstrated superior performance compared to existing methods, exhibiting heightened detection accuracy by deeply analyzing the melanoma-affected regions.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Data Availability
The dataset produced and scrutinized in this study are accessible from the corresponding author upon reasonable request.
References
Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, Thrun S. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):115–8.
Haenssle HA, Fink C, Schneiderbauer R, Toberer F, Buhl T, Blum A, Hofmann-Wellenhof R. Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Ann Oncol. 2018;29(8):1836–42.
Tschandl P, Rosendahl C, Kittler H. The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Sci data. 2018;5(1):1–9.
Codella NC, Gutman D, Celebi ME, Helba B, Marchetti MA, Dusza SW, Halpern A. Skin lesion analysis toward melanoma detection: A challenge at the 2017 International Symposium on Biomedical Imaging (ISBI), hosted by the International Skin Imaging Collaboration (ISIC). arXiv preprint arXiv. 2018:1710.05006.
Rajpurkar P, Irvin J, Ball RL, Zhu K, Yang B, Mehta H, Duggal P. Deep learning for chest radiograph diagnosis: a retrospective comparison of the CheXNeXt algorithm to practicing radiologists. PLoS Med. 2017;15(11), e1002686.
Han SS, Park GH, Lim W, Kim MS, Na JI, Park I. Deep neural networks show an equivalent and often superior performance to dermatologists in onychomycosis diagnosis: automatic construction of onychomycosis datasets by region-based convolutional deep neural network. PLoS ONE. 2018;13(3), e0191493.
Fujisawa Y, Otomo Y, Ogata Y, Nakamura Y. Image recognition in dermatology: representation learning, classification, and big data. Front Med. 2018;5:226.
Gessert N, Kaps R, Werner J. Accurate skin lesion segmentation with convolutional neural networks and attention mechanisms. Int J Comput Assist Radiol Surg. 2021;16(4):695–702.
Brinker TJ, Hekler A, Enk AH, Berking C. Deep learning outperformed 11 pathologists in the classification of histopathological melanoma images. Eur J Cancer. 2019;119:57–65.
Yu L, Chen H, Dou Q, Qin J, Heng PA. Integrating online and offline knowledge for 3D anatomical Landmark Detection via Multi-task Learning. IEEE Trans Med Imaging. 2020;39(7):2201–12.
Acknowledgements
The authors acknowledged the Alvas Institute of Engineering and Technology, Mangaluru, Karnataka, India ; A J Institute of Engineering and Technology, Mangaluru Karnataka, India for their invaluable support in facilitating the research through provision of necessary facilities.
Funding
No funding received for this research.
Author information
Authors and Affiliations
Contributions
This research was a collective effort, made possible through the collaboration and contributions of all authors involved.
Corresponding author
Ethics declarations
Conflict of Interest
No conflict of interest.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Ganesh, V.N., Gulappagol, L., Pushparani, M.K. et al. Panoptic Region Slicing Segmentation and Optimized Alexnet-Based CNN for Early Melanoma Diagnosis. SN COMPUT. SCI. 5, 756 (2024). https://doi.org/10.1007/s42979-024-03131-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s42979-024-03131-6