Loading [MathJax]/jax/element/mml/optable/BasicLatin.js
Research article Special Issues

Performance comparison of deep learning models for MRI-based brain tumor detection

  • Received: 02 November 2024 Revised: 27 December 2024 Accepted: 02 January 2025 Published: 15 January 2025
  • Brain tumors pose a significant threat to human health, as they can severely affect both physical well-being and quality of life. These tumors often lead to increased intracranial pressure and neurological complications. Traditionally, brain tumors are diagnosed through manual interpretation via medical imaging techniques such as Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). While these methods are effective, they are time-consuming and subject to errors in consistency and accuracy. This study explored the use of several deep-learning models, including YOLOv8, YOLOv9, Faster R-CNN, and ResNet18, for the detection of brain tumors from MR images. The results demonstrate that YOLOv9 outperforms the other models in terms of accuracy, precision, and recall, highlighting its potential as the most effective deep-learning approach for brain tumor detection.

    Citation: Abdulmajeed Alsufyani. Performance comparison of deep learning models for MRI-based brain tumor detection[J]. AIMS Bioengineering, 2025, 12(1): 1-21. doi: 10.3934/bioeng.2025001

    Related Papers:

    [1] Jalil Manafian, Onur Alp Ilhan, Sizar Abid Mohammed . Forming localized waves of the nonlinearity of the DNA dynamics arising in oscillator-chain of Peyrard-Bishop model. AIMS Mathematics, 2020, 5(3): 2461-2483. doi: 10.3934/math.2020163
    [2] Mahmoud A. E. Abdelrahman, Sherif I. Ammar, Kholod M. Abualnaja, Mustafa Inc . New solutions for the unstable nonlinear Schrödinger equation arising in natural science. AIMS Mathematics, 2020, 5(3): 1893-1912. doi: 10.3934/math.2020126
    [3] M. Hafiz Uddin, M. Ali Akbar, Md. Ashrafuzzaman Khan, Md. Abdul Haque . New exact solitary wave solutions to the space-time fractional differential equations with conformable derivative. AIMS Mathematics, 2019, 4(2): 199-214. doi: 10.3934/math.2019.2.199
    [4] Haikun Liu, Yongqiang Fu . On the variable exponential fractional Sobolev space Ws(·),p(·). AIMS Mathematics, 2020, 5(6): 6261-6276. doi: 10.3934/math.2020403
    [5] Samir Kumar Bhandari, Dhananjay Gopal, Pulak Konar . Probabilistic α-min Ciric type contraction results using a control function. AIMS Mathematics, 2020, 5(2): 1186-1198. doi: 10.3934/math.2020082
    [6] Manikandan G, Perumal R . Mate and mutual mate functions in a seminearring. AIMS Mathematics, 2020, 5(5): 4974-4982. doi: 10.3934/math.2020318
    [7] Huaji Cheng, Yanxia Hu . Exact solutions of the generalized (2+1)-dimensional BKP equation by the G'/G-expansion method and the first integral method. AIMS Mathematics, 2017, 2(3): 562-579. doi: 10.3934/Math.2017.2.562
    [8] Johnny Henderson, Abdelghani Ouahab, Samia Youcefi . Existence results for ϕ-Laplacian impulsive differential equations with periodic conditions. AIMS Mathematics, 2019, 4(6): 1610-1633. doi: 10.3934/math.2019.6.1610
    [9] Abdelkrim Salim, Sabri T. M. Thabet, Imed Kedim, Miguel Vivas-Cortez . On the nonlocal hybrid (k,φ)-Hilfer inverse problem with delay and anticipation. AIMS Mathematics, 2024, 9(8): 22859-22882. doi: 10.3934/math.20241112
    [10] Murali Ramdoss, Divyakumari Pachaiyappan, Inho Hwang, Choonkil Park . Stability of an n-variable mixed type functional equation in probabilistic modular spaces. AIMS Mathematics, 2020, 5(6): 5903-5915. doi: 10.3934/math.2020378
  • Brain tumors pose a significant threat to human health, as they can severely affect both physical well-being and quality of life. These tumors often lead to increased intracranial pressure and neurological complications. Traditionally, brain tumors are diagnosed through manual interpretation via medical imaging techniques such as Computed Tomography (CT) and Magnetic Resonance Imaging (MRI). While these methods are effective, they are time-consuming and subject to errors in consistency and accuracy. This study explored the use of several deep-learning models, including YOLOv8, YOLOv9, Faster R-CNN, and ResNet18, for the detection of brain tumors from MR images. The results demonstrate that YOLOv9 outperforms the other models in terms of accuracy, precision, and recall, highlighting its potential as the most effective deep-learning approach for brain tumor detection.



    Fractional partial differential equations (FPDEs) are the generalizations of classical partial differential equations with integer orders, which are used to describe several phenomena in many fields of sciences, such as mechanics, signal processing, plasma physics, systems identification, electricity, chemistry, biology, control theory and other areas.

    The exact solutions of FPDEs play a crucial role in the study of nonlinear sciences. It is used to describe observed various qualitative and quantitative features of nonlinear phenomenons in many fields of mathematical physics, it can let our better understand some complex physics phenomena. Therefore, it is an important task to seek more exact solutions of different forms for the FPDEs.

    In recent decades, with the development of science and technology, especially symbolic computation package such as Maple and Mathematica, many researchers have presented many direct and powerful approaches to establish exact solutions of fractional partial differential equations. For example, the fractional sub-equation method [1,2], the first integral method [3], the extended fractional Riccati expansion method [4], the fractional complex transform [5], the Jacobi elliptic equation method [6], the fractional mapping method [7], the (G/G)-expansion method [8], the improved fractional (DαG/G) method [9], the extended fractional (DαξG/G)-expansion method [10], the separation variables approach [11], the modified extended tanh method [12], the exp(Φ(ξ)) method [13,14], the invariant subspace method [15], and other methods [16,17,18,19]. Due to these methods, various exact solutions or numerical solutions of FPDEs have been established successfully.

    The two variable (ϕ/ϕ,1/ϕ)-expansion method is the generalization of (G/G)-expansion method, the main idea of this method is that the solutions to FPDEs are represented as a polynomial in two variables (ϕ/ϕ) and (1/ϕ), wherein ϕ=ϕ(ξ) satisfies the second order ODE ϕ+δϕ=μ, where δ and μ are constants. The objective of this article is to establish further general and some fresh close form solitary wave solution to the time-fractional Kuramoto-Sivashinsky (K-S) equation, (3+1)-dimensional time-fractional KdV-Zakharov-Kuznetsov (KdV-ZK) equation and time-fractional Sharma-Tasso-Olver (FSTO) equation by means of the two variable (ϕ/ϕ,1/ϕ)-expansion method, the results suggest that the method is significative and further general.

    The organization of the paper is as follows. In Section 2, the description of conformable fractional derivative and its properties are given. In Section 3, we describe the algorithm for solving time-fractional partial differential equations by using the two variable (ϕ/ϕ,1/ϕ)-expansion method with the help of fractional complex transform. In Section 4, we apply the two variable (ϕ/ϕ,1/ϕ)-expansion to the time-fractional K-S equation, (3+1)-dimensional KdV-ZK equation and FSTO equation. In Section 5, some typical wave figures of the exact solutions are given. Results and discussion part are added in Section 6. The conclusion part is in Section 7.

    In fractional calculus, the most famous fractional derivatives are the Riemann-Liouville and the Caputo fractional derivatives, the Riemann-Liouville fractional derivative is defined as follows [20]:

    If n is a positive integer and αϵ[n1,n], the α derivative of a function f is given by

     Dαt(f)(t)=1Γ(nα)dndtntaf(x)(tx)αn+1dx. (2.1)

    Also, the Caputo fractional derivative is defined as follows [20]

     Dαt(f)(t)=1Γ(nα)dndtntaf(n)(x)(tx)αn+1dx. (2.2)

    Some flaws arise in these definitions of these fractional derivatives, for example, all these derivatives do not satisfy: the known formula of the derivative of the product of two functions, the known formula of the derivative of the quotient of two functions and the chain rule of two functions.

    In 2014, Khalil et al. [21] introduced a novel definition of fractional derivative named the conformable fractional derivative to overcome the flaws found in Riemann-Liouville and the Caupto fractional derivatives.

    Definition 1. Suppose f: [0,)R is a function. Then, the conformable fractional derivative of f of order α is defined as

     Dαt(f)(t)=limϵ0f(t+ϵt1α)f(t)ϵ, (2.3)

    for all t>0 and α(0,1]. If f is α-differentiable in some (0,a), a>0, and limt0+f(α)(t) exists, then f(α)(0)=limt0+f(α)(t).

    Some properties of the conformable fractional derivative are given below as in [21]

    Thereom 1. Suppose α(0,1], and f=f(t) and g=g(t) are αdifferentiable at t>0. Then

     Dαt(af+bg)=aDαt(f)+bDαt(g),     a,bR. (2.4)
     Dαt(tμ)=μtμα,     μR. (2.5)
     Dαt(fg)=fDαt(g)+gDαt(f). (2.6)
     Dαt(fg)=gDαt(f)fDαt(g)g2. (2.7)

    If, in addition to f differentiable, then

     Dαt(f)(t)=t1αdfdt. (2.8)

    Thereom 2. Suppose functions f,g: [0,)R be αdifferentiable, where (0<α1). Then the following rule is obtained

     Dαt(fg)(t)=t1αg(t)f(g(t)). (2.9)

    The above equations play an important role in fractional calculus in the following sections.

    In this section we give the description of the two variable (ϕ/ϕ,1/ϕ)-expansion method to find exact traveling wave solutions of time-fractional partial differential equation.

    Suppose that a time-fractional partial differential equation in the variables x,y,z,t is given by

     P(u,ux,uy,uz,uxx,uxy,uxz,uxxx,,Dαtu,D2αttu,)=0,    0<α1, (3.1)

    where Dαtu,D2αttu are fraction-order derivatives of u with respect to t, P is a polynomial of u=u(x,y,z,t) and its various partial conformable derivatives including the highest order derivatives and nonlinear terms.

    We use the conformable wave transformation:

     u(x,y,z,t)=u(ξ),ξ=c(x+y+zυtαα), (3.2)

    where c and υ are constant to be determine later, the FPDE (3.1) is reduced to the following nonlinear ordinary differential equation (ODE) for u(x,y,z,t)=u(ξ):

     P(u,cu,c2u,c3u,,cυu,)=0, (3.3)

    where u=uξ,u=uξξ,.

    We suppose the solution u of (3.3) can be expressed in the following form:

    u=ni=0ai(ϕϕ)i+n1j=0bi(ϕϕ)j1ϕ, (3.4)

    where ai,bj(i=0,1,2,,n;j=0,1,2,,n1) are constants and anbn10. The positive number n can be determined by considering the homogeneous balance between the highest order derivatives and nonlinear terms appearing in (3.3). The function ϕ=ϕ(ξ) satisfies the second order linear ODE in the form

    ϕ+δϕ=μ, (3.5)

    where δ and μ are constants. Equation (3.5) has three types of general solution with double arbitrary parameters as follows [22]:

     ϕ(ξ)={A1cosh(δξ)+A2sinh(δξ)+μδ, when   δ<0,A1cos(δξ)+A2sin(δξ)+μδ,         when   δ>0,A1ξ+A2+μ2ξ2,                             when   δ=0.  (3.6)

    and

     (ϕϕ)2={(δA21δA22μ2δ)(1ϕ)2δ+2μϕ,   when δ<0,                (δA21+δA22μ2δ)(1ϕ)2δ+2μϕ,   when δ>0,                 (A212μA2)(1ϕ)2+2μϕ,              when δ=0.                  (3.7)

    where A1,A2 are arbitrary constants.

    By substituting (3.4) into (3.3) and using the second order linear ODE (3.5) and (3.7), collecting all terms with the same order of 1ϕi and 1ϕiϕϕ together, the left-hand side of (3.3) is converted into another polynomial in 1ϕi and 1ϕiϕϕ. Equating each coefficient of this different power terms to zero yields a set of algebraic equations for ai,bj(i=0,1,2,,n;j=0,1,2,,n1),δ,μ,c and υ.

    Assuming constants ai,bj(i=0,1,2,,n;j=0,1,2,,n1),δ,μ,c and υ can be determined by solving the nonlinear algebraic equations. Then substituting these terms and the general solutions (3.6) of (3.5) into (3.4), we can obtain more exact traveling wave solutions of (3.1).

    In this subsection, we investigate more general and new exact traveling wave solutions of time-fractional differential equations by means of the two variable (ϕ/ϕ,1/ϕ)-expansion method.

    we consider the time-fractional Kuramoto-Sivashinsky (K-S) equation [23]:

     Dαtu+auux+buxx+kuxxxx=0, (4.1)

    where 0<α1 and a,b,k are arbitrary constants.

    The K-S equation (4.1) was examined as a prototypical example of spatiotemporal chaos in one space dimension. This equation represents the motion of a fluid going down a vertical wall, the variations of the position of a flame front, or a spatially uniform oscillating chemical reaction in a homogeneous medium.

    To solve (4.1), we use the wave transformations:

     u=u(ξ),   ξ=c(xυtαα), (4.2)

    then (4.1) is reduced into a nonlinear ODE in the form

     cυu(ξ)+acu(ξ)u(ξ)+bc2u(ξ)+kc4u(4)(ξ)=0. (4.3)

    By reducing (4.3), we get

     υu(ξ)+au(ξ)u(ξ)+bcu(ξ)+kc3u(4)(ξ)=0. (4.4)

    By balancing the highest order derivative term u(4) and nonlinear term uu in (4.4), the value of n can be determined, which is n=3 in this problem. Therefore, by (3.4), we have the following ansatz:

    u(ξ)=a0+a1(ϕϕ)+a2(ϕϕ)2+a3(ϕϕ)3+b01ϕ+b1(ϕϕ)1ϕ+b2(ϕϕ)21ϕ, (4.5)

    where a0,a1,a2,a3,b0,b1 and b2 are constants to be determined later, and function ϕ(ξ) satisfies (3.5).

    By substituting (4.5) into (4.4) and using the second order linear ODE (3.5) and (3.7), collecting all terms with the same order of 1ϕi and 1ϕiϕϕ together, the left-hand side of (4.4) is converted into another polynomial in 1ϕi and 1ϕiϕϕ. Equating each coefficient of this different power terms to zero yields a set of algebraic equations for a0,a1,a2,a3,b0,b1,b2,δ,μ,k,a,b,c and υ. Solving this system of algebraic equations, with the aid of Maple, we obtain

    1.a1=±3b2δ2Ξi, a2=b2μ2Ξi, a3=±b2Ξi,b0=b2(μ2+Ξiδ)Ξi, b1=3b2μ2Ξi, b=±19b2aδ60cΞi,  
     k=±ab260c3Ξi,  υ=a(b2μδ+2a0Ξi)2Ξi,                                                               (4.6)
    2.a1=±21b2δ22Ξi, a2=b2μ2Ξi, a3=±b2Ξi,b0=b2(11μ2+5Ξiδ)11Ξi,  b1=3b2μ2Ξi,                 
     b=19b2aδ660cΞi,  k=±ab260c3Ξi,  υ=a(b2μδ+2a0Ξi)2Ξi,                                          (4.7)

    where Ξi(i=1,2,3) are given later.

    Substituting (4.6) and (4.7) and the general solutions (3.6) of Eq. (3.5) into (4.5), we obtain more exact traveling wave solutions of (4.1):

    u1i(ξ)=a0±3b2δ2ΞiΦib2μ2ΞiΦ2i±b2ΞiΦ3i+Ψi(b2(μ2+Ξiδ)Ξi3b2μ2ΞiΦi+b2Φ2i), (4.8)

    where ξ=c(x(b2δμ+2a0Ξi)a2Ξiαtα),  b=±19b2aδ60cΞi,  k=±ab260c3Ξi.

    u2i(ξ)=a0±21b2δ22ΞiΦib2μ2ΞiΦ2i±b2ΞiΦ3i+Ψi(b2(11μ2+5Ξiδ)11Ξi3b2μ2ΞiΦi+b2Φ2i), (4.9)

    where ξ=c(x(b2δμ+2a0Ξi)a2Ξiαtα),  b=19b2aδ660cΞi,  k=±ab260c3Ξi.

    When i=1 for δ<0, i=2 for δ>0, i=3 for δ=0. And

    Φ1=A1sinh(δξ)+A2cosh(δξ)A1cosh(δξ)+A2sinh(δξ)+μ/δδ,                          (4.10)
    Ψ1=1A1cosh(δξ)+A2sinh(δξ)+μ/δ,  Ξ1=A21δA22δμ2/δ. (4.11)
    Φ2=A1sin(δξ)+A2cos(δξ)A1cos(δξ)+A2sin(δξ)+μ/δδ,                                     (4.12)
    Ψ2=1A1cos(δξ)+A2sin(δξ)+μ/δ,  Ξ2=A21δ+A22δμ2/δ.         (4.13)
    Φ3=A1+μξA1ξ+A2+μξ2/2, Ψ3=1A1ξ+A2+μξ2/2,  Ξ3=A212A2μ.    (4.14)

    Since A1 and A2 are arbitrary constants, one may choose arbitrarily their values. For example, if we choose A1=0, A20 and μ=0 in (4.8), we obtain some traveling wave solutions.

    Case 1.1 when δ<0, we have

    u1(ξ)=a0+b2δA2(32coth(δξ)coth3(δξ)csch3(δξ)),(A2>0),    (4.15)

    where ξ=c(xa0aαtα),b=19b2aδ60A2cδ,  k=ab260c3A2δ.

    u2(ξ)=a0b2δA2(32coth(δξ)coth3(δξ)+csch3(δξ)),(A2<0),    (4.16)

    where ξ=c(xa0aαtα),b=19b2aδ60A2cδ,  k=ab260c3A2δ.

    Case 1.2 when δ>0, we have

    u3(ξ)=a0+b2δA2(32cot(δξ)+cot3(δξ)+csc3(δξ)),(A2>0),    (4.17)

    where ξ=c(xa0aαtα),b=19b2aδ60A2cδ,  k=ab260c3A2δ.

    u4(ξ)=a0b2δA2(32cot(δξ)+cot3(δξ)csc3(δξ)),(A2<0),    (4.18)

    where ξ=c(xa0aαtα),b=19b2aδ60A2cδ,  k=ab260c3A2δ.

    Again if A2=0, A10 and μ=0 in (4.8), we obtain some traveling wave solutions.

    Case 2.1 when δ<0, we have

    u5(ξ)=a0+b2δδA21δ(32tanh(δξ)tanh3(δξ))+b2δA1sech3(δξ), (4.19)

    where ξ=c(xa0aαtα),b=19b2aδ60cA21δ,  k=ab260c3A21δ.

    Case 2.2 when δ>0, we have

    u6(ξ)=a0b2δA1(32tan(δξ)+tan3(δξ))sec3(δξ)),(A1>0), (4.20)

    where ξ=c(xa0aαtα),b=19b2aδ60cA1δ,  k=ab260c3A1δ.

    u7(ξ)=a0+b2δA1(32tan(δξ)+tan3(δξ))+sec3(δξ)),(A1<0), (4.21)

    where ξ=c(xa0aαtα),b=19b2aδ60cA1δ,  k=ab260c3A1δ.

    If we choose A1=0, A20 and μ=0 in (4.9), we obtain some traveling wave solutions, for example

    Case 3.1 when δ<0, we have

    u8(ξ)=a0+b2δA2(2122coth(δξ)coth3(δξ)+(511coth2(δξ))csch(δξ)),(A2>0),    (4.22)

    where ξ=c(xa0aαtα),b=19b2aδ660A2cδ,  k=ab260c3A2δ.

    u9(ξ)=a0b2δA2(2122coth(δξ)coth3(δξ)(511coth2(δξ))csch(δξ)),(A2<0),    (4.23)

    where ξ=c(xa0aαtα),b=19b2aδ660A2cδ,  k=ab260c3A2δ.

    Case 3.2 when δ>0, we have

    u10(ξ)=a0+b2δA2(2122cot(δξ)+cot3(δξ)+(511+cot2(δξ))csc(δξ)),(A2>0),    (4.24)

    where ξ=c(xa0aαtα),b=19b2aδ660A2cδ,  k=ab260c3A2δ.

    u11(ξ)=a0b2δA2(2122cot(δξ)+cot3(δξ)(511+cot2(δξ))csc(δξ)),(A2<0),    (4.25)

    where ξ=c(xa0aαtα),b=19b2aδ660A2cδ,  k=ab260c3A2δ.

    Again if A2=0, A10 and μ=0 in (4.9), we obtain some traveling wave solutions.

    Case 4.1 when δ<0, we have

    u12(ξ)=a0+b2δδA21δ(2122tanh(δξ)tanh3(δξ))+b2δA1(511tanh2(δξ))sech(δξ), (4.26)

    where ξ=c(xa0aαtα),b=19b2aδ660cA21δ,  k=ab260c3A21δ.

    Case 4.2 when δ>0, we have

    u13(ξ)=a0b2δA1(2122tan(δξ)+tan3(δξ)(511+tan2(δξ))sec(δξ)),(A1>0),    (4.27)

    where ξ=c(xa0aαtα),b=19b2aδ660A1cδ,  k=ab260c3A1δ.

    u14(ξ)=a0+b2δA1(2122tan(δξ)+tan3(δξ)+(511+tan2(δξ))sec(δξ)),(A1<0),    (4.28)

    where ξ=c(xa0aαtα),b=19b2aδ660A1cδ,  k=ab260c3A1δ.

    Consider the (3+1)-dimensional time-fractional KdV-Zakharov-Kuznetsov (KdV-ZK) equation [24,25]:

     Dαtu+auux+uxxx+b(uxyy+uxzz)=0, (4.29)

    where 0<α1 and a,b are arbitrary constants.

    It is well known that the Korteweg-de Vries (KdV) equation arises as an model for one-dimensional long wavelength surface waves propagating in weakly nonlinear dispersive media, as well as the evolution of weakly nonlinear ion acoustic waves in plasmas. The ZK equation is one of two well-studied canonical two-dimensional extensions of the Korteweg-de Vries equation. In recent, S. Shoo et al. [25] found some new exact traveling wave solutions of Eq. (4.29) by the simplest equation method.

    To solve (4.29), we use the wave transformations:

     u=u(ξ),   ξ=c(x+y+zυtαα), (4.30)

    then (4.29) is reduced into a nonlinear ODE in the form

     cυu(ξ)+acu(ξ)u(ξ)+c3(1+2b)u(ξ)=0. (4.31)

    By reducing (4.31), we get

     υu(ξ)+au(ξ)u(ξ)+c2(1+2b)u(ξ)=0. (4.32)

    Further by integrating (4.32) with respect to ξ, we get

     υu(ξ)+a2u2(ξ)+c2(1+2b)u(ξ)=0. (4.33)

    By balancing the highest order derivative term u and nonlinear term u2 in (4.33), the value of n can be determined, which is n=2 in this problem. Therefore, by equation (3.4), we have the following ansatz:

    u(ξ)=a0+a1(ϕϕ)+a2(ϕϕ)2+b01ϕ+b1(ϕϕ)1ϕ, (4.34)

    where a0,a1,a2,b0 and b1 are constants to be determined later, and function ϕ(ξ) satisfies (3.5).

    By substituting (4.34) into (4.33) and using the second order linear ODE (3.5) and (3.7), collecting all terms with the same order of 1ϕi and 1ϕiϕϕ together, the left-hand side of (4.33) is converted into another polynomial in 1ϕi and 1ϕiϕϕ. Equating each coefficient of this different power terms to zero yields a set of algebraic equations for a0,a1,a2,b0,b1,δ,μ,a,b,c and υ. Solving this system of algebraic equations, with the aid of Maple, we obtain

     1.a0=±b1δΞi,  a1=0,  a2=±b1Ξi,  b0=b1μΞi,  b=6c2Ξi±ab112c2Ξi,  υ=±δab16Ξi.   (4.35)
     2.a0=±2b1δ3Ξi,  a1=0,  a2=±b1Ξi,  b0=b1μΞi,  b=6c2Ξi±ab112c2Ξi,  υ=δab16Ξi.   (4.36)

    Substituting (4.35) and (4.36) and the general solutions (3.6) of Eq. (3.5) into (4.34), we obtain more exact solutions of (4.29):

    u1i(ξ)=±b1δΞi±b1ΞiΦ2i+b1Ψi(μΞi+Φi), (4.37)

    where ξ=c(x+y+zδab16Ξiαtα),  b=6c2Ξi±ab112c2Ξi.

    u2i(ξ)=±2b1δ3Ξi±b1ΞiΦ2i+b1Ψi(μΞi+Φi), (4.38)

    where ξ=c(x+y+z±δab16Ξiαtα),  b=6c2Ξi±ab112c2Ξi, in which i=1 for δ<0, i=2 for δ>0, i=3 for δ=0, and Φi,Ψi,Ξi, (i=1,2,3) see (4.10)-(4.14).

    Here A1 and A2 are arbitrary constants. Therefore, one can freely select their values. If we choose A1=0, A20, μ=0 and δ>0 in (4.37), we obtain some traveling wave solutions.

    u1(ξ)=b1δA2(csc2(δξ)+cot(δξ)csc(δξ)),(A2>0),  (4.39)

    where ξ=c(x+y+zδab16A2αδtα),b=6c2A2δ+ab112c2A2δ.

    u2(ξ)=b1δA2(csc2(δξ)cot(δξ)csc(δξ)),(A2<0),  (4.40)

    where ξ=c(x+y+z+δab16A2αδtα),b=6c2A2δ+ab112c2A2δ.

    If we choose A1=0, A20, μ=0 and δ<0 in (4.38), we obtain some other traveling wave solutions.

    u3(ξ)=b1δA2(23coth2(δξ)coth(δξ)csch(δξ)),(A2>0),  (4.41)

    where ξ=c(x+y+z+δab16A2αδtα),b=6c2A2δ+ab112c2A2δ.

    u4(ξ)=b1δA2(23coth2(δξ)+coth(δξ)csch(δξ)),(A2<0),  (4.42)

    where ξ=c(x+y+zδab16A2αδtα),b=6c2A2δ+ab112c2A2δ.

    If we choose A1=0, A20, μ=0 and δ>0 in (4.38), we obtain some traveling wave solutions.

    u5(ξ)=b1δA2(23+cot2(δξ)+cot(δξ)csc(δξ)),(A2>0),  (4.43)

    where ξ=c(x+y+z+δab16A2αδtα),b=6c2A2δ+ab112c2A2δ.

    u6(ξ)=b1δA2(23+cot2(δξ)cot(δξ)csc(δξ)),(A2>0),  (4.44)

    where ξ=c(x+y+zδab16A2αδtα),b=6c2A2δ+ab112c2A2δ.

    Similarly, we can write down the other families of exact solutions of Eq. (4.29) which are omitted for convenience.

    Consider the time-fractional Sharma-Tasso-Olver (FSTO) equation [26,27]:

     Dαtu+3au2x+3au2ux+3auuxx+auxxx=0, (4.45)

    where a is an arbitrary constant and 0<α1. The function u(x,t) is assumed to be a causal function of time. i.e. vanishing for t<0. The general response expression contains a parameter describing the order of the fractional derivative that can be varied to obtain various responses. In the case of α=1, Eq. (4.45) reduces to the classical nonlinear STO equation. L. Song [26] found a rational approximation solution of Eq. (4.45) by the variational iteration method, the Adomian decomposition method and the homotopy perturbation method.

    To solve (4.45), we use the wave transformations:

     u=u(ξ),   ξ=c(xυtαα), (4.46)

    then (4.45) is reduced into a nonlinear ODE in the form

     cυu(ξ)+3ac2u2(ξ)+3acu2(ξ)u(ξ)+3ac2u(ξ)u(ξ)+ac3u(ξ)=0. (4.47)

    By reducing (4.47), we get

     υu(ξ)+3acu2(ξ)+3au2(ξ)u(ξ)+3acu(ξ)u(ξ)+ac2u(ξ)=0. (4.48)

    Further by integrating (4.48) with respect to ξ, we get

     υu(ξ)+3acu(ξ)u(ξ)+au3(ξ)+ac2u(ξ)=0. (4.49)

    By balancing the highest order derivative term u and nonlinear term u3 in (4.49), the value of n can be determined, which is n=1 in this problem. Therefore, by Eq. (3.4), we have the following ansatz:

    u(ξ)=a0+a1(ϕϕ)+b01ϕ, (4.50)

    where a0,a1 and b0 are constants to be determined later, and function ϕ(ξ) satisfies (3.5).

    By substituting (4.50) into (4.49) and using the second order linear ODE (3.5) and (3.7), collecting all terms with the same order of 1ϕi and 1ϕiϕϕ together, the left-hand side of (4.49) is converted into another polynomial in 1ϕi and 1ϕiϕϕ. Equating each coefficient of this different power terms to zero yields a set of algebraic equations for a0,a1,b0,δ,μ,a,c and υ. Solving this system of algebraic equations, with the aid of Maple, we obtain

     1.a0=0,  a1=c,  b0=±Ξic,  υ=ac2δ.                  (4.51)
     2.a0=0,  a1=±b0Ξi,  c=±2b0Ξi,  υ=b20aδΞi.             (4.52)
     3.a0=±δΞib0,  a1=b0Ξi,  c=2b0Ξi,  υ=4b20aδΞi.    (4.53)
     4.a0=±δΞib0,  a1=b0Ξi,  c=2b0Ξi,  υ=4b20aδΞi. (4.54)

    Substituting (4.51)-(4.54) and the general solutions (3.6) of Eq. (3.5) into (4.50), we obtain more exact solutions of (4.45):

    u1i(ξ)=cΦi±ΞicΨi, (4.55)

    where ξ=c(x+ac2δtαα).

    u2i(ξ)=±b0ΞiΦi+b0Ψi, (4.56)

    where ξ=±2b0Ξi(x+b20aδtαΞiα).

    u3i(ξ)=±δΞib0+b0ΞiΦi+b0Ψi, (4.57)

    where ξ=2b0Ξi(x+4b20aδtαΞiα).

    u4i(ξ)=±δΞib0b0ΞiΦi+b0Ψi, (4.58)

    where ξ=2b0Ξi(x+4b20aδtαΞiα), in which i=1 for δ<0, i=2 for δ>0, i=3 for δ=0, and Φi,Ψi,Ξi, (i=1,2,3) see (4.10)-(4.14).

    Here A1 and A2 are arbitrary constants. Therefore, one can freely select their values. If we choose A1=0, A20, μ=0 and δ<0 in (4.55), we the traveling wave solution:

    u1(ξ)=cδ(coth(δξ)±csch(δξ)).  (4.59)

    If we choose A1=0, A20, μ=0 and δ>0 in (4.55), we obtain the traveling wave solution:

    u2(ξ)=cδ(cot(δξ)±csc(δξ)).  (4.60)

    If we choose A2=0, A10, μ=0 and δ>0 in (4.55), we obtain the traveling wave solution:

    u3(ξ)=cδ(tan(δξ)±sec(δξ)).  (4.61)

    where ξ=c(x+ac2δαtα).

    Similarly, we can write down the other families of exact solutions of Eq. (4.45) which are omitted for convenience.

    In this section, some typical wave figures are given as follows (Figure 1-3):

    Figure 1.  (a) 2D figure of solution u11(ξ) in (4.8) with A1=1,μ=1,δ=1,A2=1,c=0.5,a0=2,a=1,b2=1, (b) 2D figure of solution u12(ξ) in (4.8) with A1=1,μ=1,δ=1,A2=1,c=0.5,a0=2,a=1,b2=1.
    Figure 2.  (a) 2D figure of solution u11(ξ) in (4.37) with A1=1,μ=2,δ=1,A2=1,c=1,a=1,b1=1,y=1,z=1, (b) 2D figure of solution u12(ξ) in (4.37) with A1=1,μ=1,δ=1,A2=1,c=1,a=1,b1=1,y=1,z=1.
    Figure 3.  (a) 2D figure of solution u11(ξ) in (4.55) with A1=1,μ=2,δ=1,A2=1,c=1,a=1, (b) 2D figure of solution u12(ξ) in (4.55) with A1=1,μ=1,δ=1,A2=1,c=1,a=1.

    The basic idea of the two variable (ϕ/ϕ,1/ϕ)-expansion method is to research the new exact traveling wave solutions of the mentioned Eqs. (4.1), (4.29) and Eq. (4.45). The Eqs. (4.1), (4.29) and (4.29) have been studied using various techniques, among them, Authors obtained some new solutions, but the researches considered the Jumaries modified Riemann-Liouville derivative sense for their solution techniques. Nonetheless, the existing analytical solutions reported in [23,24,27] are not correct because the utilized definitions of fractional derivative have some shortcomings that could not be overlooked [21]. Chen et al. [25] found some new solutions of Eq. (4.29) expressed by tanh, coth, tan and cot form. In our case, first time we considered the conformable fractional derivative sense and two variable (ϕ/ϕ,1/ϕ)-expansion method for the Eqs. (4.1), (4.29) and (4.45), we found some new solutions expressed by tanh, coth, sech, csch, tan, cot, sec and csc form. To our knowledge, the solutions obtained have not been reported in former literature. So, all the solutions are new in this article.

    In this study, the two variable (ϕ/ϕ,1/ϕ)-expansion method with the help of conformable wave transformation has been applied to find out exact traveling wave solutions of time-fractional differential equations. We have obtained some new and further general solitary wave solutions to three nonlinear time fractional differential equation, namely, time-fractional K-S equation, (3+1)-dimensional KdV-ZK equation and FSTO equation in terms of hyperbolic, trigonometric and rational function solution involving parameters. These solutions have important physical implications, for example, these solutions forces are convenient to characterize the hydromagnetic waves in cold plasma, acoustic waves in inharmonic crystals and acoustic-gravity waves incompressible fluids. The obtained results show that the two variable (ϕ/ϕ,1/ϕ)-expansion method is direct, consistent, reliable, very much attractive and an effective powerful mathematical tool for obtaining the exact solutions of other time fractional differential equations, and it can be generalized to nonlinear space-time fractional differential equations and space fractional differential equations. Finally, our results in this article have been checked using the Maple by putting them back into the original equation.

    The authors would like to express their deepest appreciation to the reviewers for their valuable suggestions and comments to improve the article.

    This work was supported by the Middle-Aged Academic Backbone of Honghe University (No. 2014GG0105; No.2015GG0207), Scientific Research Foundation of Yunnan Education Department (No. 2018JS479) Yunnan Applied Basic Research Project(No.2018FH001013; 2018FH001014) and the Natural Science Foundation of Education Committee of Yunnan Province (No. 2019J0558).

    The authors declare no conflict of interest.



    1 https://www.kaggle.com/datasets/pkdarabi/medical-image-dataset-brain-tumor-detection/data

    Conflict of interest



    The author declares no conflict of interest.

    [1] Aldape K, Brindle KM, Chesler L, et al. (2019) Challenges to curing primary brain tumours. Nat Rev Clin Oncol 16: 509-520. https://doi.org/10.1038/s41571-019-0177-5
    [2] Franceschi E, Frappaz D, Rudà R, et al. (2020) Rare primary central nervous system tumors in adults: an overview. Front Oncol 10: 996. https://doi.org/10.3389/fonc.2020.00996
    [3] McRobbie DW, Moore EA, Graves MJ, et al. (2017) MRI: From Picture to Proton.Cambridge University Press. https://doi.org/10.3174/ajnr.A0980
    [4] Batİ CT, Ser G (2023) Effects of data augmentation methods on YOLO v5s: application of deep learning with pytorch for individual cattle identification. Yüz Yıl Üniv Tarım Bilim Derg 33: 363-376. https://doi.org/10.29133/yyutbd.1246901
    [5] Krichen M (2023) Convolutional neural networks: a survey. Computers 12: 151. https://doi.org/10.3390/computers12080151
    [6] Muksimova S, Umirzakova S, Mardieva S, et al. (2023) Enhancing medical image denoising with innovative teacher-student model-based approaches for precision diagnostics. Sensors 23: 9502. https://doi.org/10.3390/s23239502
    [7] Rayed ME, Islam SMS, Niha SI, et al. (2024) Deep learning for medical image segmentation: state-of-the-art advancements and challenges. Inform Med Unlocked 47: 101504. https://doi.org/10.1016/j.imu.2024.101504
    [8] Huang SC, Pareek A, Seyyedi S, et al. (2020) Fusion of medical imaging and electronic health records using deep learning: a systematic review and implementation guidelines. NPJ Digit Med 3: 136. https://doi.org/10.1038/s41746-020-00341-z
    [9] Abdusalomov AB, Mukhiddinov M, Whangbo TK (2023) Brain tumor detection based on deep learning approaches and magnetic resonance imaging. Cancers 15: 4172. https://doi.org/10.3390/cancers15164172
    [10] ZainEldin H, Gamel SA, El-Kenawy E-SM, et al. (2022) Brain tumor detection and classification using deep learning and sine-cosine fitness grey wolf optimization. Bioengineering 10: 18. https://doi.org/10.3390/bioengineering10010018
    [11] Passa RS, Nurmaini S, Rini DP (2023) YOLOv8 based on data augmentation for MRI brain tumor detection. Sci J Inform 10: 8. https://doi.org/10.15294/sji.v10i3.45361
    [12] Hussain M (2023) YOLO-v1 to YOLO-v8, the rise of YOLO and its complementary nature toward digital manufacturing and industrial defect detection. Machines 11: 677. https://doi.org/10.3390/machines11070677
    [13] Patil S, Waghule S, Waje S, et al. (2024) Efficient object detection with YOLO: a comprehensive guide. Int J Adv Res Sci Commun Technol 4: 519-531. https://doi.org/10.48175/ijarsct-18483
    [14] Alshahrani A, Mustafa J, Almatrafi M, et al. (2024) A comparative study of deep learning techniques for Alzheimer's disease detection in medical radiography. IJCSNS 24: 53-63. https://doi.org/10.22937/IJCSNS.2024.24.5.6
    [15] Kesav N, Jibukumar MG (2022) Efficient and low complex architecture for detection and classification of brain tumor using RCNN with two channel CNN. J King Saud Univ Comput Inf Sci 34: 6229-6242. https://doi.org/10.1016/j.jksuci.2021.05.008
    [16] Avşar E, Salçin K (2019) Detection and classification of brain tumours from MRI images using faster R-CNN. Teh Glas 13: 337-342. https://doi.org/10.31803/tg-20190712095507
    [17] Senan EM, Jadhav ME, Rassem TH, et al. (2022) Early diagnosis of brain tumour MRI images using hybrid techniques between deep and machine learning. Comput Math Methods Med 2022: 8330833. https://doi.org/10.1155/2022/8330833
    [18] Mathivanan SK, Sonaimuthu S, Murugesan S, et al. (2024) Employing deep learning and transfer learning for accurate brain tumor detection. Sci Rep 14: 7232. https://doi.org/10.1038/s41598-024-57970-7
    [19] El-Kenawy ESM, Khodadadi N, Mirjalili S, et al. (2024) Greylag goose optimization: nature-inspired optimization algorithm. Expert Syst Appl 238: 122147. https://doi.org/10.1016/j.eswa.2023.122147
    [20] El-Kenawy E, El-Sayed M, Rizk FH, et al. (2024) iHow optimization algorithm: a human-inspired metaheuristic approach for complex problem solving and feature selection. J Artif Intell Eng Pract 1: 36-53. https://doi.org/10.21608/jaiep.2024.386694
    [21] El-Kenawy ESM, Rizk FH, Zaki AM, et al. (2024) Football Optimization Algorithm (FbOA): a novel metaheuristic inspired by team strategy dynamics. J Artif Intell Eng Product 8: 21-38. https://doi.org/10.54216/JAIM.080103
    [22] Simo AMD, Kouanou AT, Monthe V, et al. (2024) Introducing a deep learning method for brain tumor classification using MRI data towards better performance. Inform Med Unlocked 44: 101423. https://doi.org/10.1016/j.imu.2023.101423
    [23] Ren S, He K, Girshick R, et al. (2017) Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Trans Pattern Anal Mach Intell 39: 1137-1149. https://doi.org/10.1109/tpami.2016.2577031
    [24] Holzinger A, Langs G, Denk H, et al. (2019) Causability and explainability of artificial intelligence in medicine. Wiley Interdiscip Rev Data Min Knowl Discov 9: e1312. https://doi.org/10.1002/widm.1312
    [25] Edalatifar M, Bote-Curiel L, Sánchez-Cano A, et al. (2021) A hybrid neuro-fuzzy algorithm for prediction of reference impact in scientific publications. IEEE Access 9: 35784-33579. https://doi.org/10.1007/978-3-319-99834-3_31
    [26] Redmon J, Farhadi A (2018) YOLOv3: an incremental improvement. arXiv: 180402767 . https://doi.org/10.48550/arXiv.1804.02767
    [27] Bochkovskiy A, Wang CY, Liao HM (2020) YOLOv4: optimal speed and accuracy of object detection. arXiv: 200410934 . https://doi.org/10.48550/arXiv.2004.10934
    [28] Chicco D, Jurman G (2020) The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genom 21: 6. https://doi.org/10.1186/s12864-019-6413-7
    [29] Powers DM (2011) Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. J Mach Learn Technol 2: 37-63. https://doi.org/10.48550/arXiv.2010.16061
    [30] Shorten C, Khoshgoftaar TM (2019) A survey on image data augmentation for deep learning. J Big Data 6: 1-48. https://doi.org/10.1186/s40537-019-0197-0
    [31] Yosinski J, Clune J, Bengio Y, et al. (2014) How transferable are features in deep neural networks?. Adv Neural Inf Process Syst 27: 3320-3328. https://doi.org/10.48550/arXiv.1411.1792
    [32] Tan C, Sun F, Kong T, et al. (2018) A survey on deep transfer learning. ICANN 2018: Artificial Neural Networks and Machine Learning. Cham, Switzerland: Springer International Publishing 270-279.
    [33] Selvaraju RR, Cogswell M, Das A, et al. (2017) Grad-CAM: visual explanations from deep networks via gradient-based localization. 2017 IEEE International Conference on Computer Vision (ICCV). Venice, Italy: IEEE 618-626.
    [34] Lundberg SM, Lee SI (2017) A unified approach to interpreting model predictions. arXiv: 170507874 . https://doi.org/10.48550/arXiv.1705.07874
    [35] Samek W, Wiegand T, Müller KR (2017) Explainable artificial intelligence: understanding, visualizing and interpreting deep learning models. ITU J: ICT Discov 1: 39-48. https://doi.org/10.48550/arXiv.1708.08296
    [36] Arrieta AB, Díaz-Rodríguez N, Del Ser J, et al. (2020) Explainable artificial intelligence (XAI): concepts, taxonomies, opportunities and challenges toward responsible AI. Inf Fusion 58: 82-115. https://doi.org/10.1016/j.inffus.2019.12.012
    [37] Tjoa E, Guan C (2020) A survey on explainable artificial intelligence (XAI): towards medical XAI. Artif Intell Rev 53: 591-641. https://doi.org/10.1007/s10462-020-09825-y
  • This article has been cited by:

    1. Aniqa Aniqa, Jamshad Ahmad, Soliton solution of fractional Sharma-Tasso-Olever equation via an efficient (G′/G)-expansion method, 2022, 13, 20904479, 101528, 10.1016/j.asej.2021.06.014
    2. Hongkua Lin, Explicit Exact Solutions of Space-time Fractional Drinfel’d-Sokolov-Wilson Equations, 2021, 2068, 1742-6588, 012005, 10.1088/1742-6596/2068/1/012005
    3. Sevil Çulha Ünal, Ayşegül Daşcıoğlu, Exact solutions of time fractional Korteweg–de Vries–Zakharov–Kuznetsov equation, 2021, 44, 0170-4214, 9557, 10.1002/mma.7379
    4. Shumaila Naz, Qazi Mahmood Ul-Hassan, Jamshad ahmad, Aniqa Zulfiqar, Dynamics of new optical solutions for nonlinear equations via a novel analytical technique, 2022, 54, 0306-8919, 10.1007/s11082-022-03793-7
    5. Mst Munny Khatun, M Ali Akbar, Zarina K Fayazova, Onur Alp İlhan, Various wave solutions to the nonlinear fractional Korteweg-de Vries Zakharov-Kuznetsov equation by a new approach to the two-variable expansion scheme, 2024, 99, 0031-8949, 035218, 10.1088/1402-4896/ad23ba
    6. Ramima Akter, Shuvo Sarker, Anish Adhikary, M. Ali Akbar, Pinakee Dey, M.S. Osman, Dynamics of geometric shape solutions for space-time fractional modified equal width equation with beta derivative, 2024, 11, 26668181, 100841, 10.1016/j.padiff.2024.100841
    7. Yating Hao, Ben Gao, Exact solutions to the (3+1)-dimensional time-fractional KdV–Zakharov–Kuznetsov equation and modified KdV equation with variable coefficients, 2024, 98, 0973-7111, 10.1007/s12043-023-02693-z
  • Reader Comments
  • © 2025 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(532) PDF downloads(34) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog