Machine-learning-assisted interrogation of sulfide mineral LA-ICP-MS data: Klaza Epithermal Deposit, Yukon, Canada

In order to investigate the geochemical evolution of the Klaza hydrothermal system and metal distribution among sulfide minerals, pyrite, sphalerite, and arsenopyrite were mapped using LA-ICP-MS. The raster data from each map were processed using unsupervised machine learning (K-means clustering and PCA) to classify and filter pixels unrelated to the mineral of interest and to subsequently attribute primary and late features in the mineral grain.

为探究克拉扎热液系统的地球化学演化过程以及硫化物矿物中的金属分布特征，研究人员采用激光剥蚀电感耦合等离子体质谱法（LA-ICP-MS）对黄铁矿（pyrite）、闪锌矿（sphalerite）与毒砂（arsenopyrite）开展成像测绘。随后，通过无监督机器学习方法（K均值聚类与主成分分析（PCA））处理每张成像图的栅格数据，以分类并筛除与目标矿物无关的像素点，进而对矿物颗粒中的原生及后期特征进行归属标注。