AI Article Synopsis
- Deep learning struggles with detecting and diagnosing medical image anomalies due to data imbalance, variability, and complexity, particularly in skin diseases that show significant differences in appearance and texture.
- A new hybrid architecture combining wavelet decomposition with EfficientNet models has been developed to address these challenges, utilizing advanced techniques for data augmentation, loss functions, and optimization.
- The proposed model demonstrated impressive accuracy rates of 94.7% and 92.2% when tested on the HAM10000 and ISIC2017 datasets, respectively.
Article Abstract
Faced with anomalies in medical images, Deep learning is facing major challenges in detecting, diagnosing, and classifying the various pathologies that can be treated via medical imaging. The main challenges encountered are mainly due to the imbalance and variability of the data, as well as its complexity. The detection and classification of skin diseases is one such challenge that researchers are trying to overcome, as these anomalies present great variability in terms of appearance, texture, color, and localization, which sometimes makes them difficult to identify accurately and quickly, particularly by doctors, or by the various Deep Learning techniques on offer. In this study, an innovative and robust hybrid architecture is unveiled, underscoring the symbiotic potential of wavelet decomposition in conjunction with EfficientNet models. This approach integrates wavelet transformations with an EfficientNet backbone and incorporates advanced data augmentation, loss function, and optimization strategies. The model tested on the publicly accessible HAM10000 and ISIC2017 datasets has achieved an accuracy rate of 94.7%, and 92.2% respectively.
Download full-text PDF |
Source |
|---|---|
| http://www.ncbi.nlm.nih.gov/pmc/articles/PMC11685683 | PMC |
| http://dx.doi.org/10.7150/jca.101574 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
The development of China's National Carbon Market has strengthened the inherent link between the carbon market and the broader energy market, providing a potential for cross-market risk transmission resonance. Studying the risk spillover effects between China's National Carbon Market and the crude oil futures market is of significant practical importance, both in terms of carbon market development and carbon risk management. Based on the Maximal Overlap Discrete Wavelet Transform (MODWT), the price series are decomposed across multiple scales, and the risk spillover effects between the carbon market and the crude oil futures market are examined from both the time domain and the frequency domain.
View Article and Find Full Text PDFHybrid Model with Wavelet Decomposition and EfficientNet for Accurate Skin Cancer Classification.
J Cancer
January 2025
Department of Information and Communication Engineering, Yeungnam University, Gyeongsan, 38541, Republic of Korea.
Article Synopsis
- Deep learning struggles with detecting and diagnosing medical image anomalies due to data imbalance, variability, and complexity, particularly in skin diseases that show significant differences in appearance and texture.
- A new hybrid architecture combining wavelet decomposition with EfficientNet models has been developed to address these challenges, utilizing advanced techniques for data augmentation, loss functions, and optimization.
- The proposed model demonstrated impressive accuracy rates of 94.7% and 92.2% when tested on the HAM10000 and ISIC2017 datasets, respectively.
Fourier analysis of signal dependent noise images.
Sci Rep
December 2024
Cancer Epidemiology Department, H. Lee Moffitt Cancer Center and Research Institute, 12902 Bruce B. Downs Blvd, Tampa, FL, 33612, USA.
An archetype signal dependent noise (SDN) model is a component used in analyzing images or signals acquired from different technologies. This model-component may share properties with stationary normal white noise (WN). Measurements from WN images were used as standards for making comparisons with SDN in both the image domain (ID) and Fourier domain (FD).
View Article and Find Full Text PDFWeak fault feature extraction method of rolling bearing based on wavelet packet decomposition and optimized MCKD.
Rev Sci Instrum
December 2024
School of Mechatronic Engineering, Southwest Petroleum University, Chendu 610500, China.
The early fault characteristics of rolling bearings are weak, especially in a strong noise environment, which are more difficult to extract; therefore, a method based on wavelet packet decomposition, multi-verse optimizer, and maximum correlated kurtosis deconvolution for weak fault feature extraction of rolling bearings is proposed. First, the original vibration signal is decomposed using wavelet packet decomposition, followed by proposing a signal reconstruction method combining the Pearson correlation coefficient and energy ratio to effectively remove noise from the original signal. Second, the parameters L and M of Maximum Correlated Kurtosis Deconvolution (MCKD) are optimized using the multi-verse optimizer algorithm to obtain optimal filter settings.
View Article and Find Full Text PDFA comparative study of wavelet families for schizophrenia detection.
Front Hum Neurosci
December 2024
Department of Electrical Engineering, Mathematics and Science, University of Gävle, Gävle, Sweden.
Schizophrenia (SZ) is a chronic mental disorder, affecting approximately 1% of the global population, it is believed to result from various environmental factors, with psychological factors potentially influencing its onset and progression. Discrete wavelet transform (DWT)-based approaches are effective in SZ detection. In this report, we aim to investigate the effect of wavelet and decomposition levels in SZ detection.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!