تحولی در تشخیص لوسمی با هوش مصنوعی
کد: G-1595
نویسندگان: Shirin Fakhar * ℗, Asiyeh Jebelli
زمان بندی: زمان بندی نشده!
برچسب: تشخیص و درمان سرطان
دانلود: دانلود پوستر
خلاصه مقاله:
خلاصه مقاله
Background and aims: Leukemia is a hematological disease affecting bone marrow and lymphatic system. Because the mortality rate is too high, early diagnosis is pivotal. This study focuses on improving Leukemia detection using Artificial Intelligence (AI). Method & Results: Machine learning (ML) is a significant tool for building detection models. A hybrid SVM-PSO model improves the accuracy of Leukemia detection by representing a two-dimensional image and completing the classification process. This model has high accuracy, superior performance, outperforming stand-alone algorithms, an enhanced confusion matrix, and a higher detection rate. FOADCNN-LDC technique using ShuffleNetv2 and CDAE models eradicates the image noise for Leukemia diagnosis and classification. The FOADCNN-LDC approach has a superior accuracy value of 99.62% over existing techniques. CapsNet is a neural network method that analyzes complex features and spatial relationships within images. However, a hybrid model that combines a genetic algorithm with ResNet-50V2, and SVM/JAYA demonstrated the superiority of this method in different terms. DDRNet is another model used to classify blood cell images. The strategic combination of three blocks, DRDB, GLFEB, and CSAB, in DDRNet addresses specific challenges in the classification process, resulting in improved discrimination of features important for accurate multi-class blood cell image identification. Their effective integration within the model can lead to the superior performance of DDRNet. The DDRNet model has a high testing accuracy of 91.98%, minimal computational complexity, and enhanced feature discrimination ability. Moreover, several novel deep ML-based approaches like DeepSHAP Autoencoder Filter for Gene Selection were utilized to select the most informative hub genes for time-to-therapy-need. The results indicate the effectiveness of techniques in detecting genes with independent predictive power, suggesting biomarkers to improve the diagnosis for further investigation. A range of ML-boosting algorithms can identify the individual risk of death, treatment, infection, and their combination. As genomic heterogeneity inherent in Leukemia is too vast for traditional prediction methods, ML-based methods facilitate the identification of novel Leukemia subclasses. Conclusion: The objective of using AI is to decrease the death rate through early diagnosis of Leukemia, thus offering individuals a better chance of survival from this disease.
کلمات کلیدی
Leukemia Diagnosis, Artificial Intelligence, Machine Learning
