2023 · 38 citations · 13 references
ComputingComputer ScienceOncologyComputer VisionFeature EngineeringMachine LearningCancer Early DetectionComputer-aided DiagnosisEarly Cancer DetectionFeature DetectionImage Feature EngineeringMedical Image AnalysisBenchmark DatasetsCancer Risk Prediction ModelsEarly Severity DetectionLong-term Prognosis EstimationScale-invariant Feature DetectionScale-invariant Feature TransformScale-invariant Feature Transform (Sift)Severity Analysis
Blood cancer is an ever-growing global health concern. Early detection and accurate diagnosis are key to successful treatment and long-term prognosis of the disease. Despite various advances in medical technologies, current methods for blood cancer detection are far from perfect and often suffer from a lack of accuracy. This paper proposes the use of Scale-Invariant Feature Transform (SIFT) based approaches for predicting the spreading of blood cancer and assessing its severity. SIFT is a computer vision technique that extracts meaningful features from an image. Through SIFT, meaningful patterns in medical imaging datasets, such as those used for blood cancer detection, can be identified. The output from SIFT can then be used to build predictive models which can accurately predict the spread of the disease and its severity. Additionally, this paper explains how machine-learning algorithms and artificial neural networks can be used to further enhance the accuracy of these predictions. Finally, the proposed approach is validated using benchmark datasets and its accuracy is compared with existing methods for blood cancer prediction. The proposed model reached 89.71% accuracy, 89.25% precision, 90.71% recall and 89.40% f1-score results. The results show that prediction accuracy increases significantly when SIFT is used. It is concluded that SIFT can be a powerful tool for predicting blood cancer and estimating its severity.
13
Pattern Recognition of Acute Lymphoblastic Leukemia (ALL) Using Computational Deep Learning
Nasmin Jiwani, Ketan Gupta, Giovanni Pau et al. · IEEE Access · 2023 · 67 citations · Full text