P-060: An effective approach using ensemble learning based on a random subspace method in laser-induced breakdown spectroscopy to diagnose blood cancer

Introduction: Blood cancer poses serious threats to human health, and the diagnosis of blood cancer is still facing certain challenges. The chemometrics method combined with laser-induced breakdown spectroscopy (LIBS) can be used for cancer detection. However, it was easily influenced by the spectral feature redundancy and noise, resulting in a low accuracy rate. Thus, it is essential to develop more effective methods for blood cancer diagnosis.

Methods: We proposed an approach using LIBS combined with the ensemble learning based on the random subspace method (RSM). The blood cancer serum samples including lymphoma, multiple myeloma (MM), chronic myelogenous leukemia (CML) and acute myeloid leukemia (AML) were dripped onto a boric acid substrate for LIBS spectrum collection.

Results: The results showed that these four cancer types could be distinguished by the RSM-LDA model. Compare with linear discriminant analysis (LDA) and k nearest neighbors (kNN), the RSM-LDA model has the highest average accuracy rate and Area Under Curve (AUC), suggesting the RSM-LDA model has the best classification performance. With the RSM-LDA model, the average accuracy rates for lymphoma vs healthy control (HC), MM vs HC, CML vs HC and AML vs HC were from 94.38%, 94.61%, 94.49% and 94.33% to 96.62%, 98.78%, 96.54% and 98.77%, respectively. For cancerous samples classification, the detectable rate was evaluated with the average detectable rates of lymphoma vs HC, MM vs HC, CML vs HC and AML vs HC and were 89.86%, 92.94%, 89.17% and 93.33%, respectively. Furthermore, the average accuracy rate was improved to 91.00%, and 8% of MM spectra and 6% of lymphoma spectra were misidentified to CML calculated. For blood cancer types identification, the detectable rates of lymphoma, MM, CML and AML were 93.75%, 90.00%, 86.67% and 93.33%, respectively, which means the RSM-LDA model can improve the diagnostic performance.

Conclusions: In conclusion, the RSM-LDA model provides an effective pattern recognition method for LIBS analysis in blood cancer discrimination.