W-net for whole-body bone lesion detection on 68 ga-pentixafor PET/CT imaging of multiple myeloma patients

Abstract

The assessment of bone lesion is crucial for the diagnostic and therapeutic planning of multiple myeloma (MM). 68 Ga-Pentixafor PET/CT can capture the abnormal molecular expression of CXCR-4 in addition to anatomical changes. However, the whole-body detection of dozens of lesions on hybrid imaging is tedious and error-prone. In this paper, we adopt a cascaded convolutional neural networks (CNN) to form a W-shaped architecture (W-Net). This deep learning method leverages multimodal information for lesion detection. The first part of W-Net extracts skeleton from CT scan and the second part detect and segment lesions. The network was tested on 12 68 Ga-Pentixafor PET/CT scans of MM patients using 3-folder cross validation. The preliminary results showed that W-Net can automatically learn features from multimodal imaging for MM bone lesion detection. The proof-of-concept study encouraged further development of deep learning approach for MM lesion detection with increased number of subjects.