Automated Whole-Body Bone Lesion Detection for Multiple Myeloma on ⁶⁸Ga-Pentixafor PET/CT Imaging Using Deep Learning Methods

Table 1

Experimental results of phantom study using synthetic PET/CT using V-Net, random forest (RF), -Nearest Neighbor (k-NN), and support vector machine (SVM).

Performance (%)	Sensitivity	Specificity	Precision	Dice
V-Net	89.71	99.68	88.82	89.26
RF with	99.16	89.49	12.18	21.69
kNN with = 15	98.52	90.38	12.41	23.09
SVM with	98.76	92.15	15.60	26.94