Osteosarcoma Patients Classification Using Plain X-Rays and Metabolomic Data.

Area of Science:

• Oncology
• Medical Imaging
• Bioinformatics

Background:

• Osteosarcoma is the most prevalent bone cancer, primarily diagnosed via X-rays.
• Image analysis can quantify diagnostic features for computational study.
• Integrating diverse data types may enhance diagnostic accuracy.

Purpose of the Study:

• To classify patients with benign bone tumors versus osteosarcoma.
• To evaluate the efficacy of combining imaging features and metabolomic data for diagnosis.
• To compare the performance of random forest and support vector machine (SVM) classifiers.

Main Methods:

• Utilized image features from X-rays and metabolomic data.
• Applied feature selection algorithms: recursive feature elimination and information gain.
• Assessed classification performance using random forest and SVM models.
• Evaluated models via receiver operating characteristic (ROC) curves.

Main Results:

• The random forest classifier demonstrated superior performance over SVM.
• Achieved a sensitivity of 0.92 and a specificity of 0.78 with the random forest model.
• Feature selection effectively identified relevant diagnostic indicators.

Conclusions:

• Combined imaging and metabolomic data improve osteosarcoma classification accuracy.
• Random forest is a highly effective model for diagnosing osteosarcoma.
• This approach offers potential for more accurate and objective bone cancer diagnosis.