A new intelligent hybrid feature extraction model for automating cancer diagnosis: a focus on breast cancer

Cancer, or malignant tumor, is a group of diseases that arises from the abnormal proliferation of body cells, which have the ability to invade or spread to other parts of the body. Many researchers have proposed various methods to detect breast cancer; however, the accuracy of these methods has ofte...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:The Journal of supercomputing Ročník 81; číslo 5; s. 651
Hlavní autoři: Rahmani, Roozbeh, Akbarpour, Shahin, Farzan, Ali, Anari, Babak, Afshord, Saeid Taghavi
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York Springer US 24.03.2025
Springer Nature B.V
Témata:
Accuracy
Algorithms
Automation
Back propagation
Breast cancer
Case studies
Compilers
Computer Science
Diagnosis
Feature extraction
Field programmable gate arrays
Fourier transforms
Image analysis
Image retrieval
Interpreters
Machine learning
Mammography
Medical diagnosis
Neural networks
Online data bases
Processor Architectures
Programming Languages
Sensitivity
Support vector machines
Tumors
Wavelet transforms
Simulated annealing random algorithm
Feature extraction
Feature space
Automatic cancer diagnosis
ISSN:1573-0484, 0920-8542, 1573-0484
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Cancer, or malignant tumor, is a group of diseases that arises from the abnormal proliferation of body cells, which have the ability to invade or spread to other parts of the body. Many researchers have proposed various methods to detect breast cancer; however, the accuracy of these methods has often been insufficient due to ineffective features selection and a lack of appropriate analytical techniques. To address this issue, we need an accurate feature extraction model. In this paper, we propose an intelligent hybrid feature extraction model for automating cancer diagnosis (IHFEACD) with high accuracy. This mathematical model generates more efficient features based on the structure of previous feature formulas. Furthermore, the proposed model combines new features with existing ones to create a new feature space for early cancer detection. Although this model can be applied to detect different types of cancer, we focus on breast cancer in women as our case study. To validate our approach, we investigated the mammographic image analysis society (MIAS) database and curated the breast imaging subset of digital database for screening mammography (CBIS-DDSM). The results indicate that the proposed method effectively classifies normal/abnormal and benign/malignant cases. By optimizing the feature structure in this new space, we have achieved improved accuracy in breast cancer diagnosis. The simulation results demonstrate high performance, showing an accuracy of 99.8%, sensitivity of 98%, and specificity of 99.4% using the naive bayes (NB) classifier on the MIAS database. Additionally, the proposed IHFEACD approach outperforms other methods in terms of accuracy metrics, achieving a 0.8 training test rate on the MIAS database, along with improvements of 0.3%, 1%, 6.8%, and 0.5% compared to IAIS-ABC-CDS, CADx, OKMT-SGO, and ANN-t-SNE approaches, respectively. For the CBIS-DDSM database, the performance results for breast cancer detection are also remarkable, with an accuracy of 99.5%, sensitivity of 98.8%, and specificity of 99.3% using both simple and naive bayes classifiers. This research provides a clearer picture of the robustness of the model across different databases. The proposed approach demonstrates significant improvements compared to previous methods from various comparative perspectives. Finally, this model has the potential to assist medical professionals in making informed decisions regarding breast cancer diagnosis.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1573-0484
0920-8542
1573-0484
DOI:10.1007/s11227-025-07077-1