Automated measurement of bone scan index from a whole-body bone scintigram

• DOI: 10.1007/s11548-019-02105-x
• Publication Date: 2019-12-13T18:03:34Z
• Authors: Akinobu Shimizu; Hayato Wakabayashi; Takumi Kanamori; Atsushi Saito; Kazuhiro Nishikawa; Hiromitsu Daisaki; Shigeaki Higashiyama; Joji Kawabe
• Keywords: Skeleton (computer programming); Bone scintigraphy

Abstract Purpose We propose a deep learning-based image interpretation system for skeleton segmentation and extraction of hot spots bone metastatic lesion from whole-body scintigram followed by automated measurement scan index (BSI), which will be clinically useful. Methods The proposed employs butterfly-type networks (BtrflyNets) lesions, in pair anterior posterior images are processed simultaneously. BSI is then measured using the segmented bones extracted spots. To further improve networks, supervision (DSV) residual learning technologies were introduced. Results evaluated performance 246 scintigrams prostate cancer terms accuracy segmentation, spot extraction, measurement, as well computational cost. In threefold cross-validation experiment, best was achieved BtrflyNet with DSV blocks. cross-correlation between true 0.9337, time case 112.0 s. Conclusion proved its effectiveness study scintigrams. automatically deemed acceptable reliable.