Erratum: EFNet: evidence fusion network for tumor segmentation from PET-CT volumes (Phys. Med. Biol. (2021) 66 (205005) DOI: 10.1088/1361-6560/ac299a)

Zhaoshuo Diao, Huiyan Jiang, Xian Hua Han, Yu Dong Yao, Tianyu Shi

Research output: Contribution to journalComment/debate

2 Scopus citations

Abstract

Due to an error that occurred during the production process, two paragraphs that were the intended content of sections 4.3 and 4.5.1 were deleted from this paper. Please find the missing paragraphs below: 4.3. Evaluation measures The evaluation measures are Dice, Sensitivity, Precision and IOU. Define TP as true positive, FP as false positive,TNas true negative and FN as false negative. Then Dice=2TP/(2TP+FP+FN), Sensitivity=TP/(TP+FN), Precision=TP/(TP+FP), IOU=TP/(TP+FP+FN). 4.5.1. Single modal versus multimodal The highest effect of single modal is PET E-Net. The Dice value is 0.74, which is lower than the Dice value of 0.78 in the method proposed in this paper. In addition, the Dice value of CT E-Net based on CT modal is only 0.40, and the Dice value of CT E-Net+Fusion based on fusion with PET feature map after upsampling is 0.72. Therefore, if the multi-modal fusion method is feasible, the multi-modal approach is better than the singlemodal approach because it provides more information.

Original languageEnglish
Article number249501
JournalPhysics in Medicine and Biology
Volume66
Issue number24
DOIs
StatePublished - 21 Dec 2021

Fingerprint

Dive into the research topics of 'Erratum: EFNet: evidence fusion network for tumor segmentation from PET-CT volumes (Phys. Med. Biol. (2021) 66 (205005) DOI: 10.1088/1361-6560/ac299a)'. Together they form a unique fingerprint.

Cite this