• Optics and Precision Engineering
  • Vol. 32, Issue 4, 609 (2024)
Daxiang LI, Fujie YANG*, Ying LIU, and Yao TANG
Author Affiliations
  • School of Communications and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an710121, China
  • show less
    DOI: 10.37188/OPE.20243204.0609 Cite this Article
    Daxiang LI, Fujie YANG, Ying LIU, Yao TANG. Skin lesion segmentation network with cross-attention coding[J]. Optics and Precision Engineering, 2024, 32(4): 609 Copy Citation Text show less

    Abstract

    Owing to the limitations of convolutional operations, existing skin lesion image segmentation networks are unable to model the global contextual information in images, resulting in their inability to effectively capture the target structural information of images. In this paper, a U-shaped hybrid network with cross-self-attention coding was designed for skin lesion image segmentation. Firstly, the designed multi-head gated position cross self-attention encoder was introduced in the last two layers of the U-shaped network to enable it to learn the long-term dependencies of semantic information in images and to compensate for the lack of global modelling capability of the convolutional operation; Secondly, a novel position channel attention mechanism was implemented in the skip connection part to encode the channel information of the fused features and retain the positional information to improve the network's ability to capture the target structure; finally, a regularised dice loss function was designed to enable the network to trade off between false positives and false negatives to improve the network's segmentation results. Experimental results on ISBI2017 and ISIC2018 datasets show that the network presented in this paper achieves Dice score of 91.48% and 91.30%, and IoU of 84.42% and 84.12%, respectively. The network outperforms other networks in terms of segmentation accuracy with fewer parameters and lower computational complexity. Therefore, it can efficiently segment the target region of skin lesion images and aid in the adjunctive diagnosis of skin diseases.
    Q=f01×1FK=f11×1FV=f21×1F(1)

    View in Article

    zi=u=1H+W-1SoftMaxqiTKi,uVi,u(2)

    View in Article

    zi=u=1H+W-1SoftMaxqiTKi,u+GiqqiTriq+Gi,ukKi,uTri,ukVi,u(3)

    View in Article

    Z=GPCSAF(4)

    View in Article

    Z'=MhGPCSA(BN(F))Z=F+MhGPCSAZ'F¯=Z+ConvReLUBNZ(5)

    View in Article

    Z'=MhGPCSAF=ConcatGPCSA1F,,GPCSA8F,(6)

    View in Article

    FGAP=1H×Wi=1Hj=1WF(c,i,j)c=1C(7)

    View in Article

    FXavg =1Wj=1WF(c,h,j)c=1,,C,h=1,,HFYavg =1Hi=1HF(c,i,w)c=1,,C,w=1,,W(8)

    View in Article

    FXmax=maxj1...W(F(c,h,j)c=1,,C,h=1,,H)FYmax=maxi1...H(F(c,i,w)c=1,,C,w=1,,W)(9)

    View in Article

    MX=σC1Dk×1×1α1FXavg+β1FXmax(10)

    View in Article

    MY=σC1Dk×1×1α2FYavg+β2FYmax(11)

    View in Article

    k=log2Cγ+bγodd(12)

    View in Article

    F'=FMYMX(13)

    View in Article

    LDice pi,p^i=1-2i=1Npip^ii=1Npi+i=1Np^i(14)

    View in Article

    L=LDice pi,p^i-βNi=1N1-piIn1-p^i(15)

    View in Article

    IoU(X,Y)=|XY||XY|(16)

    View in Article

    Dice=2TP2TP+FP+FN(17)

    View in Article

    Precesion=TPTP+FP(18)

    View in Article

    Recall=TPTP+FN(19)

    View in Article

    HDY˜,Y=maxhY˜,Y,hY,Y˜(20)

    View in Article

    hY˜,Y=maxy˜Y˜{minyYy˜-y}hY,Y˜=maxy˜Y{minyY˜y-y˜}(21)

    View in Article

    Daxiang LI, Fujie YANG, Ying LIU, Yao TANG. Skin lesion segmentation network with cross-attention coding[J]. Optics and Precision Engineering, 2024, 32(4): 609
    Download Citation